diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,29266 @@
+{
+  "best_metric": 2.3052062590891975e-13,
+  "best_model_checkpoint": "./checkpoint-4000",
+  "epoch": 2.999922845459455,
+  "global_step": 4860,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.5135,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.8103,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 9.8101,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 10.1898,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 9.8821,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 9.6497,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 9.9717,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 9.3271,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 9.3293,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 9.3993,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 9.319,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 9.9098,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-07,
+      "loss": 9.3977,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 9.5338,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 8.8868,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 9.3684,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 9.0794,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 9.0645,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 9.0123,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 9.4119,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4e-07,
+      "loss": 8.8021,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-07,
+      "loss": 8.5539,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 8.7908,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 8.8862,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.08e-06,
+      "loss": 8.5395,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.14e-06,
+      "loss": 8.2233,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 8.126,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.26e-06,
+      "loss": 8.2975,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 7.7976,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.38e-06,
+      "loss": 7.9925,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 8.0983,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 7.3953,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 7.1967,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.62e-06,
+      "loss": 7.5035,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.68e-06,
+      "loss": 7.3995,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 7.2042,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8e-06,
+      "loss": 7.0243,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.86e-06,
+      "loss": 7.0224,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 6.5678,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 6.75,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 6.2525,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 6.2871,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.16e-06,
+      "loss": 6.2398,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.22e-06,
+      "loss": 6.0467,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.28e-06,
+      "loss": 5.6491,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.34e-06,
+      "loss": 5.6776,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 5.254,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.46e-06,
+      "loss": 5.0795,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 5.0805,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.58e-06,
+      "loss": 4.627,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 7.2491,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7e-06,
+      "loss": 7.3335,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.76e-06,
+      "loss": 6.9999,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 7.0554,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 6.6526,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 6.3875,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 6.597,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.06e-06,
+      "loss": 6.2403,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 6.2675,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.18e-06,
+      "loss": 6.1487,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.24e-06,
+      "loss": 5.9913,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3e-06,
+      "loss": 5.8557,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.36e-06,
+      "loss": 5.7814,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 5.784,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-06,
+      "loss": 5.744,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-06,
+      "loss": 5.5437,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.54e-06,
+      "loss": 5.8307,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6e-06,
+      "loss": 5.6559,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 5.5548,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.72e-06,
+      "loss": 5.5231,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 5.4228,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 5.5118,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.9e-06,
+      "loss": 5.4285,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.96e-06,
+      "loss": 5.1592,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 5.2248,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 5.1573,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.14e-06,
+      "loss": 5.1023,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 5.153,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.26e-06,
+      "loss": 5.0788,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.32e-06,
+      "loss": 4.9816,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 4.8121,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.44e-06,
+      "loss": 4.9162,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5e-06,
+      "loss": 4.7705,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-06,
+      "loss": 4.7357,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.62e-06,
+      "loss": 4.646,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.68e-06,
+      "loss": 4.8589,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 4.6516,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 4.672,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 4.5723,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.92e-06,
+      "loss": 4.6102,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 4.5179,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.04e-06,
+      "loss": 4.3337,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.1e-06,
+      "loss": 4.2993,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.16e-06,
+      "loss": 4.1927,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.22e-06,
+      "loss": 4.1839,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 4.0563,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.34e-06,
+      "loss": 3.7496,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.4e-06,
+      "loss": 3.9954,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.46e-06,
+      "loss": 3.5369,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.52e-06,
+      "loss": 3.457,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.58e-06,
+      "loss": 6.2043,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.64e-06,
+      "loss": 6.0894,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 5.8912,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 5.688,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 5.4945,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 5.2314,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 5.3442,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6e-06,
+      "loss": 5.2601,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 5.3084,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.12e-06,
+      "loss": 5.0756,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.18e-06,
+      "loss": 5.1925,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 5.108,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.3e-06,
+      "loss": 5.0921,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.36e-06,
+      "loss": 5.0332,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.42e-06,
+      "loss": 4.8082,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.48e-06,
+      "loss": 5.0208,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.54e-06,
+      "loss": 4.9231,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.6e-06,
+      "loss": 4.765,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 4.8059,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.72e-06,
+      "loss": 4.8748,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.78e-06,
+      "loss": 4.88,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 4.8964,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 4.9392,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.96e-06,
+      "loss": 4.6879,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 4.7324,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.08e-06,
+      "loss": 4.6488,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.14e-06,
+      "loss": 4.6249,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2e-06,
+      "loss": 4.5444,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.26e-06,
+      "loss": 4.6106,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.32e-06,
+      "loss": 4.4823,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 4.5877,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.44e-06,
+      "loss": 4.5227,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5e-06,
+      "loss": 4.5071,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 4.498,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.62e-06,
+      "loss": 4.4264,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 4.4352,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-06,
+      "loss": 4.2828,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.8e-06,
+      "loss": 4.1946,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 4.2553,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.92e-06,
+      "loss": 4.2524,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 7.98e-06,
+      "loss": 4.0794,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 3.9417,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.1e-06,
+      "loss": 3.8751,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 3.6861,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 3.7071,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-06,
+      "loss": 3.6065,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 3.5654,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.5693,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 3.2224,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 3.1884,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.58e-06,
+      "loss": 5.9042,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 5.7351,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.7e-06,
+      "loss": 5.6756,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 5.3656,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.82e-06,
+      "loss": 5.2135,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.88e-06,
+      "loss": 5.0475,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 5.1011,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9e-06,
+      "loss": 4.9306,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.06e-06,
+      "loss": 4.8511,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.12e-06,
+      "loss": 4.7725,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.18e-06,
+      "loss": 4.8765,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.24e-06,
+      "loss": 4.6816,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-06,
+      "loss": 4.8654,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.36e-06,
+      "loss": 5.0027,
+      "step": 164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.42e-06,
+      "loss": 4.7675,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.48e-06,
+      "loss": 4.7165,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54e-06,
+      "loss": 4.7454,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 4.6712,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.66e-06,
+      "loss": 4.6788,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.72e-06,
+      "loss": 4.6381,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 4.6172,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84e-06,
+      "loss": 4.582,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.9e-06,
+      "loss": 4.5599,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 4.5146,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.002e-05,
+      "loss": 4.6056,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.008e-05,
+      "loss": 4.4705,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 4.5853,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.02e-05,
+      "loss": 4.3406,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 4.4274,
+      "step": 179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.032e-05,
+      "loss": 4.3961,
+      "step": 180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 4.4047,
+      "step": 181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.044e-05,
+      "loss": 4.2428,
+      "step": 182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.05e-05,
+      "loss": 4.2947,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 4.2004,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.062e-05,
+      "loss": 4.2383,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.068e-05,
+      "loss": 4.1848,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.074e-05,
+      "loss": 4.2353,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.08e-05,
+      "loss": 4.1847,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.086e-05,
+      "loss": 4.0992,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.092e-05,
+      "loss": 3.9924,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.098e-05,
+      "loss": 3.915,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.104e-05,
+      "loss": 3.9317,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.11e-05,
+      "loss": 3.7332,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.116e-05,
+      "loss": 3.6268,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 3.5605,
+      "step": 195
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.128e-05,
+      "loss": 3.5955,
+      "step": 196
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.134e-05,
+      "loss": 3.4729,
+      "step": 197
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 3.4687,
+      "step": 198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.146e-05,
+      "loss": 3.0271,
+      "step": 199
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.152e-05,
+      "loss": 3.0714,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 5.8277,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.164e-05,
+      "loss": 5.5827,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 5.347,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 5.2894,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.182e-05,
+      "loss": 5.3213,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 4.9755,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 4.9224,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2e-05,
+      "loss": 4.7249,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 4.6133,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 4.7302,
+      "step": 210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 4.6769,
+      "step": 211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.224e-05,
+      "loss": 4.5529,
+      "step": 212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 4.6437,
+      "step": 213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.236e-05,
+      "loss": 4.7464,
+      "step": 214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.242e-05,
+      "loss": 4.61,
+      "step": 215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 4.5005,
+      "step": 216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.254e-05,
+      "loss": 4.6151,
+      "step": 217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.26e-05,
+      "loss": 4.411,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 4.4964,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.272e-05,
+      "loss": 4.4687,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.278e-05,
+      "loss": 4.501,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.284e-05,
+      "loss": 4.3885,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.29e-05,
+      "loss": 4.3404,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.296e-05,
+      "loss": 4.336,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.302e-05,
+      "loss": 4.3989,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.308e-05,
+      "loss": 4.2526,
+      "step": 226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.314e-05,
+      "loss": 4.3715,
+      "step": 227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.32e-05,
+      "loss": 4.2924,
+      "step": 228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.326e-05,
+      "loss": 4.2099,
+      "step": 229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 4.2481,
+      "step": 230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.338e-05,
+      "loss": 4.2145,
+      "step": 231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.344e-05,
+      "loss": 4.1204,
+      "step": 232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 4.1879,
+      "step": 233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.356e-05,
+      "loss": 4.16,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.362e-05,
+      "loss": 4.2281,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 4.1161,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.374e-05,
+      "loss": 3.9482,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 4.0938,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 3.916,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.392e-05,
+      "loss": 4.044,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 3.8559,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 3.8818,
+      "step": 242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 3.8,
+      "step": 243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.416e-05,
+      "loss": 3.693,
+      "step": 244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.422e-05,
+      "loss": 3.6825,
+      "step": 245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.428e-05,
+      "loss": 3.5745,
+      "step": 246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.434e-05,
+      "loss": 3.6192,
+      "step": 247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.44e-05,
+      "loss": 3.1873,
+      "step": 248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.446e-05,
+      "loss": 3.0572,
+      "step": 249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.452e-05,
+      "loss": 2.9737,
+      "step": 250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.458e-05,
+      "loss": 5.5844,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.464e-05,
+      "loss": 5.5559,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.47e-05,
+      "loss": 5.2466,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 5.2629,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.482e-05,
+      "loss": 4.7461,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.488e-05,
+      "loss": 4.7748,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 4.8032,
+      "step": 257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5e-05,
+      "loss": 4.7036,
+      "step": 258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.506e-05,
+      "loss": 4.6931,
+      "step": 259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 4.5777,
+      "step": 260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.518e-05,
+      "loss": 4.6151,
+      "step": 261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.524e-05,
+      "loss": 4.6325,
+      "step": 262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.53e-05,
+      "loss": 4.5461,
+      "step": 263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 4.5339,
+      "step": 264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.542e-05,
+      "loss": 4.3584,
+      "step": 265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.548e-05,
+      "loss": 4.3681,
+      "step": 266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.554e-05,
+      "loss": 4.426,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.56e-05,
+      "loss": 4.2358,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 4.4445,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 4.4885,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.578e-05,
+      "loss": 4.3558,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.584e-05,
+      "loss": 4.1791,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.59e-05,
+      "loss": 4.3216,
+      "step": 273
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.596e-05,
+      "loss": 4.4837,
+      "step": 274
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 4.3334,
+      "step": 275
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 4.1932,
+      "step": 276
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.614e-05,
+      "loss": 4.1232,
+      "step": 277
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.62e-05,
+      "loss": 4.1301,
+      "step": 278
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.626e-05,
+      "loss": 4.2543,
+      "step": 279
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 4.3296,
+      "step": 280
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 4.1564,
+      "step": 281
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 4.0627,
+      "step": 282
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.65e-05,
+      "loss": 3.9426,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.656e-05,
+      "loss": 4.2111,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 4.1505,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 3.9769,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 4.1299,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 4.0305,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.686e-05,
+      "loss": 3.9243,
+      "step": 289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 3.8297,
+      "step": 290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.698e-05,
+      "loss": 3.8364,
+      "step": 291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.704e-05,
+      "loss": 3.592,
+      "step": 292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.71e-05,
+      "loss": 3.6456,
+      "step": 293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.716e-05,
+      "loss": 3.4557,
+      "step": 294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 3.5229,
+      "step": 295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.728e-05,
+      "loss": 3.3416,
+      "step": 296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.734e-05,
+      "loss": 3.1807,
+      "step": 297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.74e-05,
+      "loss": 3.1484,
+      "step": 298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.746e-05,
+      "loss": 3.1251,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 2.9249,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.758e-05,
+      "loss": 5.457,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.764e-05,
+      "loss": 5.5533,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.77e-05,
+      "loss": 5.2891,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.776e-05,
+      "loss": 5.0216,
+      "step": 304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.782e-05,
+      "loss": 4.8655,
+      "step": 305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 4.7591,
+      "step": 306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.794e-05,
+      "loss": 4.4591,
+      "step": 307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8e-05,
+      "loss": 4.5946,
+      "step": 308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.806e-05,
+      "loss": 4.4732,
+      "step": 309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.812e-05,
+      "loss": 4.534,
+      "step": 310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.818e-05,
+      "loss": 4.4876,
+      "step": 311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.824e-05,
+      "loss": 4.5201,
+      "step": 312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.83e-05,
+      "loss": 4.4685,
+      "step": 313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.836e-05,
+      "loss": 4.3556,
+      "step": 314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.842e-05,
+      "loss": 4.3997,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848e-05,
+      "loss": 4.2941,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854e-05,
+      "loss": 4.461,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86e-05,
+      "loss": 4.2949,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.866e-05,
+      "loss": 4.1936,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.872e-05,
+      "loss": 4.3339,
+      "step": 320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.878e-05,
+      "loss": 4.2792,
+      "step": 321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.884e-05,
+      "loss": 4.2333,
+      "step": 322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 4.1548,
+      "step": 323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.896e-05,
+      "loss": 4.318,
+      "step": 324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.902e-05,
+      "loss": 4.1554,
+      "step": 325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.908e-05,
+      "loss": 4.2348,
+      "step": 326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.914e-05,
+      "loss": 4.1746,
+      "step": 327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 4.136,
+      "step": 328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 4.1199,
+      "step": 329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.932e-05,
+      "loss": 4.3064,
+      "step": 330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.938e-05,
+      "loss": 4.0753,
+      "step": 331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.944e-05,
+      "loss": 3.9913,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.95e-05,
+      "loss": 3.9594,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 3.9248,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 3.7247,
+      "step": 335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.968e-05,
+      "loss": 4.0716,
+      "step": 336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.974e-05,
+      "loss": 3.8347,
+      "step": 337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.98e-05,
+      "loss": 3.7996,
+      "step": 338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 3.7686,
+      "step": 339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 3.7768,
+      "step": 340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 3.7082,
+      "step": 341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.004e-05,
+      "loss": 3.6877,
+      "step": 342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.01e-05,
+      "loss": 3.6209,
+      "step": 343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.016e-05,
+      "loss": 3.3822,
+      "step": 344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 3.3779,
+      "step": 345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 3.331,
+      "step": 346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 3.3149,
+      "step": 347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.04e-05,
+      "loss": 3.1964,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.046e-05,
+      "loss": 2.981,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 2.9907,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 5.3625,
+      "step": 351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.064e-05,
+      "loss": 5.1021,
+      "step": 352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.07e-05,
+      "loss": 5.0612,
+      "step": 353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 4.8129,
+      "step": 354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.082e-05,
+      "loss": 4.5588,
+      "step": 355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.088e-05,
+      "loss": 4.5592,
+      "step": 356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.094e-05,
+      "loss": 4.4664,
+      "step": 357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1e-05,
+      "loss": 4.2769,
+      "step": 358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 4.4438,
+      "step": 359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 4.4762,
+      "step": 360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.118e-05,
+      "loss": 4.3446,
+      "step": 361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.124e-05,
+      "loss": 4.3148,
+      "step": 362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.13e-05,
+      "loss": 4.3088,
+      "step": 363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.136e-05,
+      "loss": 4.3826,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 4.3101,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.148e-05,
+      "loss": 4.2935,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.154e-05,
+      "loss": 4.3897,
+      "step": 367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.16e-05,
+      "loss": 4.3086,
+      "step": 368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.166e-05,
+      "loss": 4.1998,
+      "step": 369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.172e-05,
+      "loss": 4.3241,
+      "step": 370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.178e-05,
+      "loss": 4.2554,
+      "step": 371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.184e-05,
+      "loss": 4.151,
+      "step": 372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.19e-05,
+      "loss": 4.1235,
+      "step": 373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.196e-05,
+      "loss": 4.0923,
+      "step": 374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.202e-05,
+      "loss": 4.082,
+      "step": 375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.208e-05,
+      "loss": 4.1963,
+      "step": 376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.214e-05,
+      "loss": 4.2074,
+      "step": 377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.22e-05,
+      "loss": 3.8363,
+      "step": 378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.226e-05,
+      "loss": 3.9323,
+      "step": 379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.232e-05,
+      "loss": 4.0137,
+      "step": 380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.238e-05,
+      "loss": 4.0713,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 3.9883,
+      "step": 382
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.25e-05,
+      "loss": 3.9573,
+      "step": 383
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.256e-05,
+      "loss": 4.0523,
+      "step": 384
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.262e-05,
+      "loss": 3.9059,
+      "step": 385
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.268e-05,
+      "loss": 3.9414,
+      "step": 386
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.274e-05,
+      "loss": 3.7984,
+      "step": 387
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 3.987,
+      "step": 388
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.286e-05,
+      "loss": 3.7249,
+      "step": 389
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.292e-05,
+      "loss": 3.7599,
+      "step": 390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.298e-05,
+      "loss": 3.7645,
+      "step": 391
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.304e-05,
+      "loss": 3.5248,
+      "step": 392
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 3.5559,
+      "step": 393
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 3.4376,
+      "step": 394
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.322e-05,
+      "loss": 3.4318,
+      "step": 395
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.328e-05,
+      "loss": 3.4078,
+      "step": 396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.334e-05,
+      "loss": 3.2615,
+      "step": 397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 3.1577,
+      "step": 398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 2.9043,
+      "step": 399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 2.868,
+      "step": 400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.358e-05,
+      "loss": 5.401,
+      "step": 401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.364e-05,
+      "loss": 5.0022,
+      "step": 402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.37e-05,
+      "loss": 4.9048,
+      "step": 403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 4.6424,
+      "step": 404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 4.5142,
+      "step": 405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 4.4719,
+      "step": 406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.394e-05,
+      "loss": 4.4708,
+      "step": 407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4e-05,
+      "loss": 4.4867,
+      "step": 408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 4.4333,
+      "step": 409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 4.4252,
+      "step": 410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 4.2036,
+      "step": 411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 4.3523,
+      "step": 412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.43e-05,
+      "loss": 4.4641,
+      "step": 413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 4.4383,
+      "step": 414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.442e-05,
+      "loss": 4.37,
+      "step": 415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.448e-05,
+      "loss": 4.2788,
+      "step": 416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.454e-05,
+      "loss": 4.398,
+      "step": 417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 4.2757,
+      "step": 418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 4.2294,
+      "step": 419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.472e-05,
+      "loss": 4.263,
+      "step": 420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.478e-05,
+      "loss": 4.2489,
+      "step": 421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.484e-05,
+      "loss": 4.1857,
+      "step": 422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.49e-05,
+      "loss": 4.3053,
+      "step": 423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 4.2507,
+      "step": 424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.502e-05,
+      "loss": 4.0672,
+      "step": 425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.508e-05,
+      "loss": 4.0247,
+      "step": 426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.514e-05,
+      "loss": 4.0634,
+      "step": 427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.52e-05,
+      "loss": 3.9653,
+      "step": 428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.526e-05,
+      "loss": 4.0949,
+      "step": 429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 3.8963,
+      "step": 430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.538e-05,
+      "loss": 4.0393,
+      "step": 431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.544e-05,
+      "loss": 3.8684,
+      "step": 432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.55e-05,
+      "loss": 3.9259,
+      "step": 433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.556e-05,
+      "loss": 3.9106,
+      "step": 434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.562e-05,
+      "loss": 3.8304,
+      "step": 435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.568e-05,
+      "loss": 3.8423,
+      "step": 436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.574e-05,
+      "loss": 3.7872,
+      "step": 437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.58e-05,
+      "loss": 3.8766,
+      "step": 438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.586e-05,
+      "loss": 3.7726,
+      "step": 439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.592e-05,
+      "loss": 3.6208,
+      "step": 440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 3.581,
+      "step": 441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.604e-05,
+      "loss": 3.4726,
+      "step": 442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.61e-05,
+      "loss": 3.4554,
+      "step": 443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.616e-05,
+      "loss": 3.3654,
+      "step": 444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.622e-05,
+      "loss": 3.4321,
+      "step": 445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.628e-05,
+      "loss": 3.2067,
+      "step": 446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 2.9969,
+      "step": 447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.64e-05,
+      "loss": 2.921,
+      "step": 448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.646e-05,
+      "loss": 3.1139,
+      "step": 449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.652e-05,
+      "loss": 2.8419,
+      "step": 450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.658e-05,
+      "loss": 5.0731,
+      "step": 451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 4.7888,
+      "step": 452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 4.6427,
+      "step": 453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.676e-05,
+      "loss": 4.6325,
+      "step": 454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.682e-05,
+      "loss": 4.3141,
+      "step": 455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.688e-05,
+      "loss": 4.4109,
+      "step": 456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 4.476,
+      "step": 457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 4.4199,
+      "step": 458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 4.4244,
+      "step": 459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.712e-05,
+      "loss": 4.1334,
+      "step": 460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.718e-05,
+      "loss": 4.2254,
+      "step": 461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.724e-05,
+      "loss": 4.2982,
+      "step": 462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 4.1048,
+      "step": 463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 4.2011,
+      "step": 464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 4.2732,
+      "step": 465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.748e-05,
+      "loss": 4.426,
+      "step": 466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.754e-05,
+      "loss": 4.2054,
+      "step": 467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 4.202,
+      "step": 468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 4.0487,
+      "step": 469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 4.1709,
+      "step": 470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.778e-05,
+      "loss": 4.0376,
+      "step": 471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.784e-05,
+      "loss": 4.1988,
+      "step": 472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.79e-05,
+      "loss": 4.0949,
+      "step": 473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 4.148,
+      "step": 474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 3.9856,
+      "step": 475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 4.0207,
+      "step": 476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 4.0782,
+      "step": 477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 4.2249,
+      "step": 478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.826e-05,
+      "loss": 4.0086,
+      "step": 479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.832e-05,
+      "loss": 3.9341,
+      "step": 480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.838e-05,
+      "loss": 3.9545,
+      "step": 481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.844e-05,
+      "loss": 4.1198,
+      "step": 482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 3.9543,
+      "step": 483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.856e-05,
+      "loss": 4.011,
+      "step": 484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.862e-05,
+      "loss": 3.9453,
+      "step": 485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.868e-05,
+      "loss": 3.8082,
+      "step": 486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.874e-05,
+      "loss": 3.8562,
+      "step": 487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.88e-05,
+      "loss": 3.5906,
+      "step": 488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 3.8926,
+      "step": 489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.892e-05,
+      "loss": 3.6742,
+      "step": 490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.898e-05,
+      "loss": 3.6165,
+      "step": 491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.904e-05,
+      "loss": 3.6059,
+      "step": 492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.91e-05,
+      "loss": 3.4886,
+      "step": 493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.916e-05,
+      "loss": 3.4836,
+      "step": 494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.922e-05,
+      "loss": 3.3144,
+      "step": 495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.928e-05,
+      "loss": 3.1079,
+      "step": 496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.934e-05,
+      "loss": 3.0898,
+      "step": 497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.94e-05,
+      "loss": 3.102,
+      "step": 498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.946e-05,
+      "loss": 2.8913,
+      "step": 499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 2.6458,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "eval_bleu": 0.0,
+      "eval_loss": 4.375434398651123,
+      "eval_runtime": 2564.062,
+      "eval_samples_per_second": 5.756,
+      "eval_steps_per_second": 0.72,
+      "step": 500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.958e-05,
+      "loss": 5.265,
+      "step": 501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.964e-05,
+      "loss": 4.8463,
+      "step": 502
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.97e-05,
+      "loss": 4.5924,
+      "step": 503
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.976e-05,
+      "loss": 4.6515,
+      "step": 504
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.982e-05,
+      "loss": 4.4683,
+      "step": 505
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 4.3256,
+      "step": 506
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.994e-05,
+      "loss": 4.3975,
+      "step": 507
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3e-05,
+      "loss": 4.2111,
+      "step": 508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9993119266055046e-05,
+      "loss": 4.3889,
+      "step": 509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9986238532110094e-05,
+      "loss": 4.2647,
+      "step": 510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.997935779816514e-05,
+      "loss": 4.3963,
+      "step": 511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9972477064220187e-05,
+      "loss": 4.0669,
+      "step": 512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.996559633027523e-05,
+      "loss": 4.0018,
+      "step": 513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9958715596330276e-05,
+      "loss": 4.2353,
+      "step": 514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9951834862385324e-05,
+      "loss": 4.1876,
+      "step": 515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.994495412844037e-05,
+      "loss": 4.2228,
+      "step": 516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.993807339449541e-05,
+      "loss": 4.1881,
+      "step": 517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.993119266055046e-05,
+      "loss": 4.1387,
+      "step": 518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9924311926605503e-05,
+      "loss": 4.1752,
+      "step": 519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.991743119266055e-05,
+      "loss": 4.1388,
+      "step": 520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9910550458715596e-05,
+      "loss": 4.1184,
+      "step": 521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.990366972477064e-05,
+      "loss": 4.0703,
+      "step": 522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.989678899082569e-05,
+      "loss": 4.0857,
+      "step": 523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9889908256880734e-05,
+      "loss": 4.0663,
+      "step": 524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9883027522935782e-05,
+      "loss": 3.9165,
+      "step": 525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9876146788990827e-05,
+      "loss": 3.8922,
+      "step": 526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9869266055045872e-05,
+      "loss": 3.9902,
+      "step": 527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.986238532110092e-05,
+      "loss": 4.0894,
+      "step": 528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9855504587155965e-05,
+      "loss": 3.9642,
+      "step": 529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9848623853211013e-05,
+      "loss": 4.0372,
+      "step": 530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9841743119266058e-05,
+      "loss": 3.8658,
+      "step": 531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9834862385321102e-05,
+      "loss": 3.9573,
+      "step": 532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9827981651376147e-05,
+      "loss": 3.8124,
+      "step": 533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9821100917431192e-05,
+      "loss": 3.7518,
+      "step": 534
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9814220183486237e-05,
+      "loss": 3.6785,
+      "step": 535
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9807339449541285e-05,
+      "loss": 3.8491,
+      "step": 536
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.980045871559633e-05,
+      "loss": 3.761,
+      "step": 537
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793577981651378e-05,
+      "loss": 3.6594,
+      "step": 538
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9786697247706423e-05,
+      "loss": 3.5953,
+      "step": 539
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9779816513761467e-05,
+      "loss": 3.7618,
+      "step": 540
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9772935779816515e-05,
+      "loss": 3.6021,
+      "step": 541
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.976605504587156e-05,
+      "loss": 3.4332,
+      "step": 542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.975917431192661e-05,
+      "loss": 3.4905,
+      "step": 543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9752293577981653e-05,
+      "loss": 3.3176,
+      "step": 544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9745412844036698e-05,
+      "loss": 3.3787,
+      "step": 545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9738532110091746e-05,
+      "loss": 3.0901,
+      "step": 546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.973165137614679e-05,
+      "loss": 3.1733,
+      "step": 547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9724770642201836e-05,
+      "loss": 2.9312,
+      "step": 548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.971788990825688e-05,
+      "loss": 2.98,
+      "step": 549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9711009174311925e-05,
+      "loss": 2.858,
+      "step": 550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9704128440366973e-05,
+      "loss": 5.0107,
+      "step": 551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9697247706422018e-05,
+      "loss": 4.8224,
+      "step": 552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9690366972477063e-05,
+      "loss": 4.7002,
+      "step": 553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.968348623853211e-05,
+      "loss": 4.4887,
+      "step": 554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9676605504587156e-05,
+      "loss": 4.2167,
+      "step": 555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9669724770642204e-05,
+      "loss": 4.3358,
+      "step": 556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.966284403669725e-05,
+      "loss": 4.3923,
+      "step": 557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9655963302752294e-05,
+      "loss": 4.3584,
+      "step": 558
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9649082568807342e-05,
+      "loss": 4.311,
+      "step": 559
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9642201834862386e-05,
+      "loss": 4.2315,
+      "step": 560
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9635321100917435e-05,
+      "loss": 4.1853,
+      "step": 561
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.962844036697248e-05,
+      "loss": 4.1671,
+      "step": 562
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9621559633027524e-05,
+      "loss": 4.136,
+      "step": 563
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9614678899082572e-05,
+      "loss": 4.2823,
+      "step": 564
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9607798165137614e-05,
+      "loss": 4.2179,
+      "step": 565
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.960091743119266e-05,
+      "loss": 4.1742,
+      "step": 566
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9594036697247707e-05,
+      "loss": 4.2129,
+      "step": 567
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.958715596330275e-05,
+      "loss": 4.0231,
+      "step": 568
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.95802752293578e-05,
+      "loss": 4.0831,
+      "step": 569
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9573394495412844e-05,
+      "loss": 4.0044,
+      "step": 570
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.956651376146789e-05,
+      "loss": 4.0104,
+      "step": 571
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9559633027522937e-05,
+      "loss": 4.0856,
+      "step": 572
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9552752293577982e-05,
+      "loss": 4.1314,
+      "step": 573
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.954587155963303e-05,
+      "loss": 4.0691,
+      "step": 574
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9538990825688075e-05,
+      "loss": 4.0251,
+      "step": 575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.953211009174312e-05,
+      "loss": 3.8316,
+      "step": 576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9525229357798168e-05,
+      "loss": 4.0223,
+      "step": 577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9518348623853213e-05,
+      "loss": 3.8936,
+      "step": 578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9511467889908257e-05,
+      "loss": 3.9537,
+      "step": 579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9504587155963306e-05,
+      "loss": 3.9282,
+      "step": 580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9497706422018347e-05,
+      "loss": 4.0381,
+      "step": 581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9490825688073395e-05,
+      "loss": 3.8899,
+      "step": 582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.948394495412844e-05,
+      "loss": 3.8215,
+      "step": 583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9477064220183485e-05,
+      "loss": 3.8345,
+      "step": 584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9470183486238533e-05,
+      "loss": 3.9464,
+      "step": 585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9463302752293578e-05,
+      "loss": 3.875,
+      "step": 586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9456422018348626e-05,
+      "loss": 3.564,
+      "step": 587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.944954128440367e-05,
+      "loss": 3.6126,
+      "step": 588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9442660550458715e-05,
+      "loss": 3.6567,
+      "step": 589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9435779816513764e-05,
+      "loss": 3.5191,
+      "step": 590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9428899082568808e-05,
+      "loss": 3.5913,
+      "step": 591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9422018348623853e-05,
+      "loss": 3.4763,
+      "step": 592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.94151376146789e-05,
+      "loss": 3.3917,
+      "step": 593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9408256880733946e-05,
+      "loss": 3.4918,
+      "step": 594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9401376146788994e-05,
+      "loss": 3.4219,
+      "step": 595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.939449541284404e-05,
+      "loss": 3.1842,
+      "step": 596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.938761467889908e-05,
+      "loss": 3.0021,
+      "step": 597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.938073394495413e-05,
+      "loss": 3.0101,
+      "step": 598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9373853211009173e-05,
+      "loss": 2.8107,
+      "step": 599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.936697247706422e-05,
+      "loss": 2.5839,
+      "step": 600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9360091743119266e-05,
+      "loss": 4.9169,
+      "step": 601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.935321100917431e-05,
+      "loss": 4.6634,
+      "step": 602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.934633027522936e-05,
+      "loss": 4.5291,
+      "step": 603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9339449541284404e-05,
+      "loss": 4.5098,
+      "step": 604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9332568807339452e-05,
+      "loss": 4.4073,
+      "step": 605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9325688073394497e-05,
+      "loss": 4.2773,
+      "step": 606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.931880733944954e-05,
+      "loss": 4.2872,
+      "step": 607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.931192660550459e-05,
+      "loss": 4.269,
+      "step": 608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9305045871559635e-05,
+      "loss": 4.2167,
+      "step": 609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.929816513761468e-05,
+      "loss": 4.2467,
+      "step": 610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9291284403669727e-05,
+      "loss": 4.201,
+      "step": 611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9284403669724772e-05,
+      "loss": 4.1585,
+      "step": 612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9277522935779817e-05,
+      "loss": 4.1873,
+      "step": 613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9270642201834862e-05,
+      "loss": 3.9574,
+      "step": 614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9263761467889907e-05,
+      "loss": 4.091,
+      "step": 615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9256880733944955e-05,
+      "loss": 4.0931,
+      "step": 616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.925e-05,
+      "loss": 4.1573,
+      "step": 617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9243119266055048e-05,
+      "loss": 4.1755,
+      "step": 618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9236238532110092e-05,
+      "loss": 3.9992,
+      "step": 619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9229357798165137e-05,
+      "loss": 4.079,
+      "step": 620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9222477064220185e-05,
+      "loss": 3.9554,
+      "step": 621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.921559633027523e-05,
+      "loss": 4.0354,
+      "step": 622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9208715596330275e-05,
+      "loss": 4.066,
+      "step": 623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9201834862385323e-05,
+      "loss": 3.9509,
+      "step": 624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9194954128440368e-05,
+      "loss": 3.9634,
+      "step": 625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9188073394495416e-05,
+      "loss": 4.0105,
+      "step": 626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.918119266055046e-05,
+      "loss": 4.0852,
+      "step": 627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9174311926605506e-05,
+      "loss": 3.9165,
+      "step": 628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.916743119266055e-05,
+      "loss": 3.7882,
+      "step": 629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9160550458715595e-05,
+      "loss": 3.8876,
+      "step": 630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9153669724770643e-05,
+      "loss": 3.8349,
+      "step": 631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9146788990825688e-05,
+      "loss": 3.9002,
+      "step": 632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9139908256880733e-05,
+      "loss": 3.7714,
+      "step": 633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.913302752293578e-05,
+      "loss": 3.7369,
+      "step": 634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9126146788990826e-05,
+      "loss": 3.6978,
+      "step": 635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9119266055045874e-05,
+      "loss": 3.6612,
+      "step": 636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.911238532110092e-05,
+      "loss": 3.6976,
+      "step": 637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9105504587155963e-05,
+      "loss": 3.6376,
+      "step": 638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.909862385321101e-05,
+      "loss": 3.6539,
+      "step": 639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9091743119266056e-05,
+      "loss": 3.3996,
+      "step": 640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.90848623853211e-05,
+      "loss": 3.5956,
+      "step": 641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.907798165137615e-05,
+      "loss": 3.4429,
+      "step": 642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9071100917431194e-05,
+      "loss": 3.4197,
+      "step": 643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9064220183486242e-05,
+      "loss": 3.3457,
+      "step": 644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9057339449541284e-05,
+      "loss": 3.1827,
+      "step": 645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.905045871559633e-05,
+      "loss": 2.9943,
+      "step": 646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9043577981651377e-05,
+      "loss": 3.0895,
+      "step": 647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.903669724770642e-05,
+      "loss": 2.8351,
+      "step": 648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.902981651376147e-05,
+      "loss": 2.7697,
+      "step": 649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9022935779816514e-05,
+      "loss": 2.5398,
+      "step": 650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.901605504587156e-05,
+      "loss": 4.9471,
+      "step": 651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9009174311926607e-05,
+      "loss": 4.7836,
+      "step": 652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9002293577981652e-05,
+      "loss": 4.3295,
+      "step": 653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8995412844036697e-05,
+      "loss": 4.3086,
+      "step": 654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8988532110091745e-05,
+      "loss": 4.3622,
+      "step": 655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.898165137614679e-05,
+      "loss": 4.152,
+      "step": 656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8974770642201838e-05,
+      "loss": 4.2805,
+      "step": 657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8967889908256883e-05,
+      "loss": 4.2046,
+      "step": 658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8961009174311927e-05,
+      "loss": 4.2099,
+      "step": 659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8954128440366975e-05,
+      "loss": 4.2261,
+      "step": 660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8947247706422017e-05,
+      "loss": 4.2323,
+      "step": 661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8940366972477065e-05,
+      "loss": 4.2631,
+      "step": 662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.893348623853211e-05,
+      "loss": 4.0397,
+      "step": 663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8926605504587155e-05,
+      "loss": 4.1473,
+      "step": 664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8919724770642203e-05,
+      "loss": 4.0582,
+      "step": 665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8912844036697247e-05,
+      "loss": 4.0104,
+      "step": 666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8905963302752296e-05,
+      "loss": 4.0893,
+      "step": 667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.889908256880734e-05,
+      "loss": 3.9466,
+      "step": 668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8892201834862385e-05,
+      "loss": 4.0809,
+      "step": 669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8885321100917433e-05,
+      "loss": 4.1124,
+      "step": 670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8878440366972478e-05,
+      "loss": 4.0932,
+      "step": 671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.8871559633027523e-05,
+      "loss": 3.9526,
+      "step": 672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.886467889908257e-05,
+      "loss": 3.9067,
+      "step": 673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8857798165137616e-05,
+      "loss": 4.052,
+      "step": 674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8850917431192664e-05,
+      "loss": 3.8522,
+      "step": 675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.884403669724771e-05,
+      "loss": 3.9389,
+      "step": 676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.883715596330275e-05,
+      "loss": 3.9657,
+      "step": 677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.88302752293578e-05,
+      "loss": 3.8425,
+      "step": 678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8823394495412843e-05,
+      "loss": 3.8184,
+      "step": 679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.881651376146789e-05,
+      "loss": 3.8927,
+      "step": 680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8809633027522936e-05,
+      "loss": 3.8285,
+      "step": 681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.880275229357798e-05,
+      "loss": 3.7147,
+      "step": 682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.879587155963303e-05,
+      "loss": 3.8218,
+      "step": 683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8788990825688074e-05,
+      "loss": 3.8177,
+      "step": 684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.878211009174312e-05,
+      "loss": 3.6834,
+      "step": 685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.8775229357798167e-05,
+      "loss": 3.6111,
+      "step": 686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.876834862385321e-05,
+      "loss": 3.7068,
+      "step": 687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.876146788990826e-05,
+      "loss": 3.5426,
+      "step": 688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8754587155963304e-05,
+      "loss": 3.7393,
+      "step": 689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.874770642201835e-05,
+      "loss": 3.5434,
+      "step": 690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8740825688073397e-05,
+      "loss": 3.3851,
+      "step": 691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8733944954128442e-05,
+      "loss": 3.5201,
+      "step": 692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8727064220183487e-05,
+      "loss": 3.5369,
+      "step": 693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.872018348623853e-05,
+      "loss": 3.3044,
+      "step": 694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8713302752293576e-05,
+      "loss": 3.2446,
+      "step": 695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8706422018348625e-05,
+      "loss": 3.0815,
+      "step": 696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.869954128440367e-05,
+      "loss": 2.9655,
+      "step": 697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8692660550458717e-05,
+      "loss": 2.8263,
+      "step": 698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8685779816513762e-05,
+      "loss": 2.8308,
+      "step": 699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8678899082568807e-05,
+      "loss": 2.6302,
+      "step": 700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8672018348623855e-05,
+      "loss": 4.9892,
+      "step": 701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.86651376146789e-05,
+      "loss": 4.6829,
+      "step": 702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8658256880733945e-05,
+      "loss": 4.461,
+      "step": 703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.8651376146788993e-05,
+      "loss": 4.4058,
+      "step": 704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8644495412844038e-05,
+      "loss": 4.2581,
+      "step": 705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8637614678899086e-05,
+      "loss": 4.2228,
+      "step": 706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.863073394495413e-05,
+      "loss": 4.3244,
+      "step": 707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8623853211009175e-05,
+      "loss": 4.206,
+      "step": 708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.861697247706422e-05,
+      "loss": 4.0038,
+      "step": 709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8610091743119265e-05,
+      "loss": 4.136,
+      "step": 710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8603211009174313e-05,
+      "loss": 4.1242,
+      "step": 711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8596330275229358e-05,
+      "loss": 4.1839,
+      "step": 712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8589449541284403e-05,
+      "loss": 3.9573,
+      "step": 713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.858256880733945e-05,
+      "loss": 4.0964,
+      "step": 714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8575688073394496e-05,
+      "loss": 4.1754,
+      "step": 715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.856880733944954e-05,
+      "loss": 3.9622,
+      "step": 716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.856192660550459e-05,
+      "loss": 4.0737,
+      "step": 717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8555045871559633e-05,
+      "loss": 4.0793,
+      "step": 718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.854816513761468e-05,
+      "loss": 4.1239,
+      "step": 719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.8541284403669726e-05,
+      "loss": 4.1435,
+      "step": 720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.853440366972477e-05,
+      "loss": 4.0308,
+      "step": 721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.852752293577982e-05,
+      "loss": 4.0292,
+      "step": 722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8520642201834864e-05,
+      "loss": 3.9296,
+      "step": 723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8513761467889912e-05,
+      "loss": 4.0227,
+      "step": 724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8506880733944953e-05,
+      "loss": 3.9908,
+      "step": 725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 4.0004,
+      "step": 726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8493119266055046e-05,
+      "loss": 3.8598,
+      "step": 727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.848623853211009e-05,
+      "loss": 4.0313,
+      "step": 728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.847935779816514e-05,
+      "loss": 3.7461,
+      "step": 729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8472477064220184e-05,
+      "loss": 3.7496,
+      "step": 730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.846559633027523e-05,
+      "loss": 3.8251,
+      "step": 731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8458715596330277e-05,
+      "loss": 3.9388,
+      "step": 732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8451834862385322e-05,
+      "loss": 3.8479,
+      "step": 733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8444954128440367e-05,
+      "loss": 3.8724,
+      "step": 734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8438073394495415e-05,
+      "loss": 3.8073,
+      "step": 735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.843119266055046e-05,
+      "loss": 3.6338,
+      "step": 736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8424311926605508e-05,
+      "loss": 3.6718,
+      "step": 737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8417431192660552e-05,
+      "loss": 3.4171,
+      "step": 738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8410550458715597e-05,
+      "loss": 3.5832,
+      "step": 739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8403669724770645e-05,
+      "loss": 3.544,
+      "step": 740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8396788990825687e-05,
+      "loss": 3.4338,
+      "step": 741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8389908256880735e-05,
+      "loss": 3.382,
+      "step": 742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.838302752293578e-05,
+      "loss": 3.3707,
+      "step": 743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8376146788990824e-05,
+      "loss": 3.1947,
+      "step": 744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8369266055045873e-05,
+      "loss": 3.211,
+      "step": 745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8362385321100917e-05,
+      "loss": 2.9358,
+      "step": 746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8355504587155962e-05,
+      "loss": 2.9177,
+      "step": 747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.834862385321101e-05,
+      "loss": 2.8263,
+      "step": 748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8341743119266055e-05,
+      "loss": 2.7954,
+      "step": 749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8334862385321103e-05,
+      "loss": 2.597,
+      "step": 750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8327981651376148e-05,
+      "loss": 4.8941,
+      "step": 751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8321100917431193e-05,
+      "loss": 4.7563,
+      "step": 752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.831422018348624e-05,
+      "loss": 4.4566,
+      "step": 753
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8307339449541286e-05,
+      "loss": 4.6128,
+      "step": 754
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8300458715596334e-05,
+      "loss": 4.2577,
+      "step": 755
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.829357798165138e-05,
+      "loss": 4.2274,
+      "step": 756
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.828669724770642e-05,
+      "loss": 4.1637,
+      "step": 757
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8279816513761468e-05,
+      "loss": 4.327,
+      "step": 758
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8272935779816513e-05,
+      "loss": 4.0671,
+      "step": 759
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.826605504587156e-05,
+      "loss": 4.0039,
+      "step": 760
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8259174311926606e-05,
+      "loss": 3.9931,
+      "step": 761
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.825229357798165e-05,
+      "loss": 4.23,
+      "step": 762
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.82454128440367e-05,
+      "loss": 3.9882,
+      "step": 763
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8238532110091744e-05,
+      "loss": 4.0574,
+      "step": 764
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.823165137614679e-05,
+      "loss": 4.1767,
+      "step": 765
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8224770642201836e-05,
+      "loss": 4.1173,
+      "step": 766
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.821788990825688e-05,
+      "loss": 4.0091,
+      "step": 767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.821100917431193e-05,
+      "loss": 3.9234,
+      "step": 768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.8204128440366974e-05,
+      "loss": 4.0761,
+      "step": 769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.819724770642202e-05,
+      "loss": 4.0232,
+      "step": 770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8190366972477067e-05,
+      "loss": 3.8495,
+      "step": 771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8183486238532112e-05,
+      "loss": 3.9073,
+      "step": 772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8176605504587157e-05,
+      "loss": 3.9088,
+      "step": 773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.81697247706422e-05,
+      "loss": 3.846,
+      "step": 774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8162844036697246e-05,
+      "loss": 3.8645,
+      "step": 775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8155963302752294e-05,
+      "loss": 3.9733,
+      "step": 776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.814908256880734e-05,
+      "loss": 3.7986,
+      "step": 777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8142201834862384e-05,
+      "loss": 3.8844,
+      "step": 778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8135321100917432e-05,
+      "loss": 3.8554,
+      "step": 779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8128440366972477e-05,
+      "loss": 3.8316,
+      "step": 780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8121559633027525e-05,
+      "loss": 3.8446,
+      "step": 781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.811467889908257e-05,
+      "loss": 3.7988,
+      "step": 782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8107798165137615e-05,
+      "loss": 3.6744,
+      "step": 783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8100917431192663e-05,
+      "loss": 3.7535,
+      "step": 784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.8094036697247707e-05,
+      "loss": 3.5543,
+      "step": 785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8087155963302756e-05,
+      "loss": 3.5793,
+      "step": 786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.80802752293578e-05,
+      "loss": 3.4917,
+      "step": 787
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8073394495412845e-05,
+      "loss": 3.6259,
+      "step": 788
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.806651376146789e-05,
+      "loss": 3.5827,
+      "step": 789
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8059633027522935e-05,
+      "loss": 3.4085,
+      "step": 790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8052752293577983e-05,
+      "loss": 3.4976,
+      "step": 791
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8045871559633028e-05,
+      "loss": 3.4456,
+      "step": 792
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8038990825688072e-05,
+      "loss": 3.2931,
+      "step": 793
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.803211009174312e-05,
+      "loss": 3.1863,
+      "step": 794
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8025229357798165e-05,
+      "loss": 3.1603,
+      "step": 795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.801834862385321e-05,
+      "loss": 3.0813,
+      "step": 796
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8011467889908258e-05,
+      "loss": 2.856,
+      "step": 797
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.8004587155963303e-05,
+      "loss": 2.9176,
+      "step": 798
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.799770642201835e-05,
+      "loss": 2.6086,
+      "step": 799
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.7990825688073396e-05,
+      "loss": 2.7443,
+      "step": 800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.798394495412844e-05,
+      "loss": 4.9625,
+      "step": 801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.797706422018349e-05,
+      "loss": 4.6492,
+      "step": 802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7970183486238534e-05,
+      "loss": 4.4824,
+      "step": 803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7963302752293582e-05,
+      "loss": 4.3578,
+      "step": 804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7956422018348623e-05,
+      "loss": 4.3899,
+      "step": 805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7949541284403668e-05,
+      "loss": 4.2422,
+      "step": 806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7942660550458716e-05,
+      "loss": 4.1877,
+      "step": 807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.793577981651376e-05,
+      "loss": 4.0871,
+      "step": 808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7928899082568806e-05,
+      "loss": 4.1366,
+      "step": 809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7922018348623854e-05,
+      "loss": 4.1255,
+      "step": 810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.79151376146789e-05,
+      "loss": 4.0374,
+      "step": 811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7908256880733947e-05,
+      "loss": 4.0985,
+      "step": 812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.790137614678899e-05,
+      "loss": 4.0384,
+      "step": 813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7894495412844036e-05,
+      "loss": 4.1181,
+      "step": 814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7887614678899085e-05,
+      "loss": 3.9733,
+      "step": 815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.788073394495413e-05,
+      "loss": 3.9691,
+      "step": 816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7873853211009177e-05,
+      "loss": 4.0078,
+      "step": 817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.7866972477064222e-05,
+      "loss": 3.9389,
+      "step": 818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7860091743119267e-05,
+      "loss": 4.0451,
+      "step": 819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7853211009174315e-05,
+      "loss": 4.0331,
+      "step": 820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7846330275229357e-05,
+      "loss": 4.0056,
+      "step": 821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7839449541284405e-05,
+      "loss": 3.8147,
+      "step": 822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.783256880733945e-05,
+      "loss": 3.9346,
+      "step": 823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7825688073394494e-05,
+      "loss": 4.012,
+      "step": 824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7818807339449542e-05,
+      "loss": 3.8617,
+      "step": 825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7811926605504587e-05,
+      "loss": 3.8247,
+      "step": 826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7805045871559632e-05,
+      "loss": 3.8461,
+      "step": 827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.779816513761468e-05,
+      "loss": 4.0756,
+      "step": 828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7791284403669725e-05,
+      "loss": 3.8744,
+      "step": 829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7784403669724773e-05,
+      "loss": 3.9181,
+      "step": 830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7777522935779818e-05,
+      "loss": 3.9788,
+      "step": 831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7770642201834863e-05,
+      "loss": 3.8619,
+      "step": 832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.776376146788991e-05,
+      "loss": 3.6693,
+      "step": 833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.7756880733944956e-05,
+      "loss": 3.6002,
+      "step": 834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 3.7428,
+      "step": 835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.774311926605505e-05,
+      "loss": 3.6893,
+      "step": 836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.773623853211009e-05,
+      "loss": 3.5815,
+      "step": 837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7729357798165138e-05,
+      "loss": 3.5232,
+      "step": 838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7722477064220183e-05,
+      "loss": 3.5726,
+      "step": 839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7715596330275228e-05,
+      "loss": 3.5175,
+      "step": 840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7708715596330276e-05,
+      "loss": 3.4131,
+      "step": 841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.770183486238532e-05,
+      "loss": 3.3865,
+      "step": 842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.769495412844037e-05,
+      "loss": 3.3571,
+      "step": 843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7688073394495413e-05,
+      "loss": 3.3497,
+      "step": 844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7681192660550458e-05,
+      "loss": 3.176,
+      "step": 845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7674311926605506e-05,
+      "loss": 3.0809,
+      "step": 846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.766743119266055e-05,
+      "loss": 2.8854,
+      "step": 847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.76605504587156e-05,
+      "loss": 2.646,
+      "step": 848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.7653669724770644e-05,
+      "loss": 2.67,
+      "step": 849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.764678899082569e-05,
+      "loss": 2.5463,
+      "step": 850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7639908256880737e-05,
+      "loss": 4.8814,
+      "step": 851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7633027522935782e-05,
+      "loss": 4.6287,
+      "step": 852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7626146788990823e-05,
+      "loss": 4.5501,
+      "step": 853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.761926605504587e-05,
+      "loss": 4.261,
+      "step": 854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7612385321100916e-05,
+      "loss": 4.1093,
+      "step": 855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7605504587155964e-05,
+      "loss": 4.1866,
+      "step": 856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.759862385321101e-05,
+      "loss": 4.1312,
+      "step": 857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7591743119266054e-05,
+      "loss": 4.1041,
+      "step": 858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7584862385321102e-05,
+      "loss": 4.0132,
+      "step": 859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7577981651376147e-05,
+      "loss": 4.0895,
+      "step": 860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7571100917431195e-05,
+      "loss": 4.0549,
+      "step": 861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.756422018348624e-05,
+      "loss": 3.9581,
+      "step": 862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7557339449541284e-05,
+      "loss": 3.9773,
+      "step": 863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7550458715596333e-05,
+      "loss": 3.8936,
+      "step": 864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7543577981651377e-05,
+      "loss": 3.9403,
+      "step": 865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.7536697247706425e-05,
+      "loss": 4.1238,
+      "step": 866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.752981651376147e-05,
+      "loss": 3.9878,
+      "step": 867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 4.0749,
+      "step": 868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.751605504587156e-05,
+      "loss": 4.0034,
+      "step": 869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7509174311926605e-05,
+      "loss": 3.9961,
+      "step": 870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.750229357798165e-05,
+      "loss": 3.9777,
+      "step": 871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7495412844036697e-05,
+      "loss": 3.9771,
+      "step": 872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7488532110091742e-05,
+      "loss": 3.7907,
+      "step": 873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.748165137614679e-05,
+      "loss": 3.8327,
+      "step": 874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7474770642201835e-05,
+      "loss": 3.7032,
+      "step": 875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.746788990825688e-05,
+      "loss": 3.8936,
+      "step": 876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7461009174311928e-05,
+      "loss": 3.8751,
+      "step": 877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7454128440366973e-05,
+      "loss": 3.8665,
+      "step": 878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.744724770642202e-05,
+      "loss": 3.9143,
+      "step": 879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.7440366972477066e-05,
+      "loss": 3.8285,
+      "step": 880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.743348623853211e-05,
+      "loss": 3.7898,
+      "step": 881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.742660550458716e-05,
+      "loss": 3.7204,
+      "step": 882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7419724770642204e-05,
+      "loss": 3.7239,
+      "step": 883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.741284403669725e-05,
+      "loss": 3.6792,
+      "step": 884
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7405963302752293e-05,
+      "loss": 3.6297,
+      "step": 885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7399082568807338e-05,
+      "loss": 3.6415,
+      "step": 886
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7392201834862386e-05,
+      "loss": 3.6796,
+      "step": 887
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.738532110091743e-05,
+      "loss": 3.4787,
+      "step": 888
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7378440366972476e-05,
+      "loss": 3.5835,
+      "step": 889
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7371559633027524e-05,
+      "loss": 3.4229,
+      "step": 890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.736467889908257e-05,
+      "loss": 3.1035,
+      "step": 891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7357798165137617e-05,
+      "loss": 3.4557,
+      "step": 892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.735091743119266e-05,
+      "loss": 3.4771,
+      "step": 893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7344036697247706e-05,
+      "loss": 3.2001,
+      "step": 894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7337155963302754e-05,
+      "loss": 3.0841,
+      "step": 895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.73302752293578e-05,
+      "loss": 3.0297,
+      "step": 896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7323394495412847e-05,
+      "loss": 2.9804,
+      "step": 897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7316513761467892e-05,
+      "loss": 2.777,
+      "step": 898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.7309633027522937e-05,
+      "loss": 2.6816,
+      "step": 899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7302752293577985e-05,
+      "loss": 2.4983,
+      "step": 900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7295871559633026e-05,
+      "loss": 4.9185,
+      "step": 901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.728899082568807e-05,
+      "loss": 4.4774,
+      "step": 902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.728211009174312e-05,
+      "loss": 4.4743,
+      "step": 903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7275229357798164e-05,
+      "loss": 4.2923,
+      "step": 904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7268348623853212e-05,
+      "loss": 4.3921,
+      "step": 905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7261467889908257e-05,
+      "loss": 4.0628,
+      "step": 906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7254587155963302e-05,
+      "loss": 4.1336,
+      "step": 907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.724770642201835e-05,
+      "loss": 4.0265,
+      "step": 908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7240825688073395e-05,
+      "loss": 3.9583,
+      "step": 909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7233944954128443e-05,
+      "loss": 4.0398,
+      "step": 910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7227064220183488e-05,
+      "loss": 4.1711,
+      "step": 911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7220183486238532e-05,
+      "loss": 4.1007,
+      "step": 912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.721330275229358e-05,
+      "loss": 3.9508,
+      "step": 913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.7206422018348625e-05,
+      "loss": 3.9787,
+      "step": 914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.719954128440367e-05,
+      "loss": 3.8947,
+      "step": 915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7192660550458718e-05,
+      "loss": 3.759,
+      "step": 916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.718577981651376e-05,
+      "loss": 3.9466,
+      "step": 917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7178899082568808e-05,
+      "loss": 3.9169,
+      "step": 918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7172018348623853e-05,
+      "loss": 4.0105,
+      "step": 919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7165137614678897e-05,
+      "loss": 3.8629,
+      "step": 920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7158256880733946e-05,
+      "loss": 3.7396,
+      "step": 921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.715137614678899e-05,
+      "loss": 3.8258,
+      "step": 922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.714449541284404e-05,
+      "loss": 3.9341,
+      "step": 923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7137614678899083e-05,
+      "loss": 3.8441,
+      "step": 924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7130733944954128e-05,
+      "loss": 3.6504,
+      "step": 925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7123853211009176e-05,
+      "loss": 3.7832,
+      "step": 926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.711697247706422e-05,
+      "loss": 3.963,
+      "step": 927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.711009174311927e-05,
+      "loss": 3.7235,
+      "step": 928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7103211009174314e-05,
+      "loss": 3.753,
+      "step": 929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.709633027522936e-05,
+      "loss": 3.6928,
+      "step": 930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.7089449541284407e-05,
+      "loss": 3.6579,
+      "step": 931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.708256880733945e-05,
+      "loss": 3.6453,
+      "step": 932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7075688073394493e-05,
+      "loss": 3.5761,
+      "step": 933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.706880733944954e-05,
+      "loss": 3.4798,
+      "step": 934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7061926605504586e-05,
+      "loss": 3.5129,
+      "step": 935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7055045871559634e-05,
+      "loss": 3.5833,
+      "step": 936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.704816513761468e-05,
+      "loss": 3.4915,
+      "step": 937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7041284403669724e-05,
+      "loss": 3.4655,
+      "step": 938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7034403669724772e-05,
+      "loss": 3.4573,
+      "step": 939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7027522935779817e-05,
+      "loss": 3.4471,
+      "step": 940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7020642201834865e-05,
+      "loss": 3.3968,
+      "step": 941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.701376146788991e-05,
+      "loss": 3.308,
+      "step": 942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7006880733944954e-05,
+      "loss": 3.1173,
+      "step": 943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 3.054,
+      "step": 944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.6993119266055047e-05,
+      "loss": 3.1359,
+      "step": 945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.6986238532110092e-05,
+      "loss": 3.031,
+      "step": 946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.697935779816514e-05,
+      "loss": 2.9851,
+      "step": 947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6972477064220185e-05,
+      "loss": 2.7334,
+      "step": 948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.696559633027523e-05,
+      "loss": 2.5527,
+      "step": 949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6958715596330274e-05,
+      "loss": 2.4362,
+      "step": 950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.695183486238532e-05,
+      "loss": 4.8778,
+      "step": 951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6944954128440367e-05,
+      "loss": 4.7592,
+      "step": 952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6938073394495412e-05,
+      "loss": 4.3466,
+      "step": 953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.693119266055046e-05,
+      "loss": 4.3998,
+      "step": 954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6924311926605505e-05,
+      "loss": 4.2682,
+      "step": 955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.691743119266055e-05,
+      "loss": 4.0566,
+      "step": 956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6910550458715598e-05,
+      "loss": 4.1775,
+      "step": 957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6903669724770643e-05,
+      "loss": 4.2825,
+      "step": 958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.689678899082569e-05,
+      "loss": 4.159,
+      "step": 959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6889908256880736e-05,
+      "loss": 4.029,
+      "step": 960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.688302752293578e-05,
+      "loss": 4.1577,
+      "step": 961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.687614678899083e-05,
+      "loss": 4.0538,
+      "step": 962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.6869266055045873e-05,
+      "loss": 3.9848,
+      "step": 963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6862385321100918e-05,
+      "loss": 4.0464,
+      "step": 964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6855504587155963e-05,
+      "loss": 3.8731,
+      "step": 965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6848623853211008e-05,
+      "loss": 3.9625,
+      "step": 966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6841743119266056e-05,
+      "loss": 4.0283,
+      "step": 967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.68348623853211e-05,
+      "loss": 3.9119,
+      "step": 968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6827981651376145e-05,
+      "loss": 3.806,
+      "step": 969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6821100917431194e-05,
+      "loss": 3.8192,
+      "step": 970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.681422018348624e-05,
+      "loss": 3.7399,
+      "step": 971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6807339449541286e-05,
+      "loss": 3.9375,
+      "step": 972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.680045871559633e-05,
+      "loss": 3.8825,
+      "step": 973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6793577981651376e-05,
+      "loss": 3.814,
+      "step": 974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6786697247706424e-05,
+      "loss": 4.0057,
+      "step": 975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.677981651376147e-05,
+      "loss": 3.8324,
+      "step": 976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6772935779816514e-05,
+      "loss": 3.7883,
+      "step": 977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6766055045871562e-05,
+      "loss": 3.9479,
+      "step": 978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6759174311926607e-05,
+      "loss": 3.6448,
+      "step": 979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.6752293577981655e-05,
+      "loss": 3.8703,
+      "step": 980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6745412844036696e-05,
+      "loss": 3.7085,
+      "step": 981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.673853211009174e-05,
+      "loss": 3.8175,
+      "step": 982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.673165137614679e-05,
+      "loss": 3.6114,
+      "step": 983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6724770642201834e-05,
+      "loss": 3.4698,
+      "step": 984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6717889908256882e-05,
+      "loss": 3.5873,
+      "step": 985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6711009174311927e-05,
+      "loss": 3.5026,
+      "step": 986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.670412844036697e-05,
+      "loss": 3.453,
+      "step": 987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.669724770642202e-05,
+      "loss": 3.6156,
+      "step": 988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6690366972477065e-05,
+      "loss": 3.4623,
+      "step": 989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.668348623853211e-05,
+      "loss": 3.3685,
+      "step": 990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6676605504587157e-05,
+      "loss": 3.3438,
+      "step": 991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6669724770642202e-05,
+      "loss": 3.1968,
+      "step": 992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.666284403669725e-05,
+      "loss": 2.9933,
+      "step": 993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6655963302752295e-05,
+      "loss": 3.0781,
+      "step": 994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.664908256880734e-05,
+      "loss": 2.8805,
+      "step": 995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.6642201834862388e-05,
+      "loss": 2.9416,
+      "step": 996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.663532110091743e-05,
+      "loss": 2.9428,
+      "step": 997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6628440366972478e-05,
+      "loss": 2.7499,
+      "step": 998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6621559633027522e-05,
+      "loss": 2.5175,
+      "step": 999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6614678899082567e-05,
+      "loss": 2.3505,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "eval_bleu": 0.0,
+      "eval_loss": 4.307140827178955,
+      "eval_runtime": 2568.1635,
+      "eval_samples_per_second": 5.747,
+      "eval_steps_per_second": 0.718,
+      "step": 1000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6607798165137615e-05,
+      "loss": 4.9677,
+      "step": 1001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.660091743119266e-05,
+      "loss": 4.6597,
+      "step": 1002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6594036697247708e-05,
+      "loss": 4.5241,
+      "step": 1003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6587155963302753e-05,
+      "loss": 4.0314,
+      "step": 1004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6580275229357798e-05,
+      "loss": 4.1454,
+      "step": 1005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6573394495412846e-05,
+      "loss": 3.9822,
+      "step": 1006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.656651376146789e-05,
+      "loss": 3.994,
+      "step": 1007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6559633027522936e-05,
+      "loss": 4.0944,
+      "step": 1008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6552752293577984e-05,
+      "loss": 3.9926,
+      "step": 1009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.654587155963303e-05,
+      "loss": 4.057,
+      "step": 1010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.6538990825688077e-05,
+      "loss": 3.9411,
+      "step": 1011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.653211009174312e-05,
+      "loss": 3.9164,
+      "step": 1012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6525229357798163e-05,
+      "loss": 3.8685,
+      "step": 1013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.651834862385321e-05,
+      "loss": 3.9499,
+      "step": 1014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6511467889908256e-05,
+      "loss": 4.1124,
+      "step": 1015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6504587155963304e-05,
+      "loss": 4.0607,
+      "step": 1016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.649770642201835e-05,
+      "loss": 4.0614,
+      "step": 1017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6490825688073393e-05,
+      "loss": 3.9706,
+      "step": 1018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.648394495412844e-05,
+      "loss": 3.7469,
+      "step": 1019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6477064220183486e-05,
+      "loss": 3.8089,
+      "step": 1020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.647018348623853e-05,
+      "loss": 3.9299,
+      "step": 1021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.646330275229358e-05,
+      "loss": 3.8027,
+      "step": 1022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6456422018348624e-05,
+      "loss": 3.8309,
+      "step": 1023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6449541284403672e-05,
+      "loss": 3.8062,
+      "step": 1024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6442660550458717e-05,
+      "loss": 3.8766,
+      "step": 1025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6435779816513762e-05,
+      "loss": 3.7845,
+      "step": 1026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.642889908256881e-05,
+      "loss": 3.6627,
+      "step": 1027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.6422018348623855e-05,
+      "loss": 3.6386,
+      "step": 1028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.64151376146789e-05,
+      "loss": 3.817,
+      "step": 1029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6408256880733944e-05,
+      "loss": 3.7637,
+      "step": 1030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.640137614678899e-05,
+      "loss": 3.681,
+      "step": 1031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6394495412844037e-05,
+      "loss": 3.6259,
+      "step": 1032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6387614678899082e-05,
+      "loss": 3.5259,
+      "step": 1033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.638073394495413e-05,
+      "loss": 3.5487,
+      "step": 1034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6373853211009175e-05,
+      "loss": 3.5475,
+      "step": 1035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.636697247706422e-05,
+      "loss": 3.5306,
+      "step": 1036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6360091743119268e-05,
+      "loss": 3.354,
+      "step": 1037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6353211009174313e-05,
+      "loss": 3.2691,
+      "step": 1038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6346330275229357e-05,
+      "loss": 3.3927,
+      "step": 1039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6339449541284406e-05,
+      "loss": 3.4417,
+      "step": 1040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.633256880733945e-05,
+      "loss": 3.3018,
+      "step": 1041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.63256880733945e-05,
+      "loss": 3.3356,
+      "step": 1042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6318807339449543e-05,
+      "loss": 3.1892,
+      "step": 1043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.6311926605504588e-05,
+      "loss": 3.0899,
+      "step": 1044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6305045871559633e-05,
+      "loss": 2.9766,
+      "step": 1045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6298165137614678e-05,
+      "loss": 2.6825,
+      "step": 1046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6291284403669726e-05,
+      "loss": 2.7802,
+      "step": 1047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.628440366972477e-05,
+      "loss": 2.7787,
+      "step": 1048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6277522935779815e-05,
+      "loss": 2.5068,
+      "step": 1049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6270642201834863e-05,
+      "loss": 2.2879,
+      "step": 1050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6263761467889908e-05,
+      "loss": 4.5866,
+      "step": 1051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6256880733944953e-05,
+      "loss": 4.4047,
+      "step": 1052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.625e-05,
+      "loss": 4.2422,
+      "step": 1053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6243119266055046e-05,
+      "loss": 4.2296,
+      "step": 1054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6236238532110094e-05,
+      "loss": 4.1701,
+      "step": 1055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.622935779816514e-05,
+      "loss": 4.1685,
+      "step": 1056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6222477064220184e-05,
+      "loss": 4.1848,
+      "step": 1057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6215596330275232e-05,
+      "loss": 4.0955,
+      "step": 1058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6208715596330276e-05,
+      "loss": 4.1015,
+      "step": 1059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6201834862385325e-05,
+      "loss": 3.9915,
+      "step": 1060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.619495412844037e-05,
+      "loss": 3.8984,
+      "step": 1061
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.618807339449541e-05,
+      "loss": 4.0315,
+      "step": 1062
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.618119266055046e-05,
+      "loss": 3.8862,
+      "step": 1063
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6174311926605504e-05,
+      "loss": 3.9672,
+      "step": 1064
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6167431192660552e-05,
+      "loss": 4.0156,
+      "step": 1065
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6160550458715597e-05,
+      "loss": 3.86,
+      "step": 1066
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.615366972477064e-05,
+      "loss": 3.9873,
+      "step": 1067
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.614678899082569e-05,
+      "loss": 3.9597,
+      "step": 1068
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6139908256880734e-05,
+      "loss": 3.7478,
+      "step": 1069
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.613302752293578e-05,
+      "loss": 3.8478,
+      "step": 1070
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6126146788990827e-05,
+      "loss": 3.9226,
+      "step": 1071
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6119266055045872e-05,
+      "loss": 3.8217,
+      "step": 1072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.611238532110092e-05,
+      "loss": 3.7075,
+      "step": 1073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6105504587155965e-05,
+      "loss": 3.7548,
+      "step": 1074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.609862385321101e-05,
+      "loss": 3.6965,
+      "step": 1075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6091743119266058e-05,
+      "loss": 3.7564,
+      "step": 1076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6084862385321103e-05,
+      "loss": 3.6424,
+      "step": 1077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6077981651376147e-05,
+      "loss": 3.6543,
+      "step": 1078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6071100917431192e-05,
+      "loss": 3.6199,
+      "step": 1079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6064220183486237e-05,
+      "loss": 3.6645,
+      "step": 1080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6057339449541285e-05,
+      "loss": 3.5999,
+      "step": 1081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.605045871559633e-05,
+      "loss": 3.5068,
+      "step": 1082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6043577981651375e-05,
+      "loss": 3.5921,
+      "step": 1083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6036697247706423e-05,
+      "loss": 3.5423,
+      "step": 1084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6029816513761468e-05,
+      "loss": 3.5595,
+      "step": 1085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6022935779816516e-05,
+      "loss": 3.6018,
+      "step": 1086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.601605504587156e-05,
+      "loss": 3.4816,
+      "step": 1087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6009174311926605e-05,
+      "loss": 3.5103,
+      "step": 1088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6002293577981654e-05,
+      "loss": 3.3334,
+      "step": 1089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5995412844036698e-05,
+      "loss": 3.3756,
+      "step": 1090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5988532110091746e-05,
+      "loss": 3.2676,
+      "step": 1091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.598165137614679e-05,
+      "loss": 3.0914,
+      "step": 1092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5974770642201836e-05,
+      "loss": 2.9686,
+      "step": 1093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.596788990825688e-05,
+      "loss": 3.1525,
+      "step": 1094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5961009174311926e-05,
+      "loss": 2.894,
+      "step": 1095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5954128440366974e-05,
+      "loss": 2.832,
+      "step": 1096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.594724770642202e-05,
+      "loss": 2.7598,
+      "step": 1097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5940366972477063e-05,
+      "loss": 2.7866,
+      "step": 1098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.593348623853211e-05,
+      "loss": 2.4547,
+      "step": 1099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5926605504587156e-05,
+      "loss": 2.3385,
+      "step": 1100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.59197247706422e-05,
+      "loss": 4.7936,
+      "step": 1101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.591284403669725e-05,
+      "loss": 4.6836,
+      "step": 1102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5905963302752294e-05,
+      "loss": 4.4624,
+      "step": 1103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5899082568807342e-05,
+      "loss": 4.3558,
+      "step": 1104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5892201834862387e-05,
+      "loss": 4.212,
+      "step": 1105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.588532110091743e-05,
+      "loss": 3.8899,
+      "step": 1106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.587844036697248e-05,
+      "loss": 3.9256,
+      "step": 1107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5871559633027525e-05,
+      "loss": 4.1953,
+      "step": 1108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5864678899082573e-05,
+      "loss": 3.9634,
+      "step": 1109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5857798165137614e-05,
+      "loss": 4.0321,
+      "step": 1110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.585091743119266e-05,
+      "loss": 4.0262,
+      "step": 1111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5844036697247707e-05,
+      "loss": 3.963,
+      "step": 1112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5837155963302752e-05,
+      "loss": 3.9172,
+      "step": 1113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5830275229357797e-05,
+      "loss": 3.9844,
+      "step": 1114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5823394495412845e-05,
+      "loss": 3.8552,
+      "step": 1115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.581651376146789e-05,
+      "loss": 3.7526,
+      "step": 1116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5809633027522938e-05,
+      "loss": 3.8854,
+      "step": 1117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5802752293577982e-05,
+      "loss": 3.8484,
+      "step": 1118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5795871559633027e-05,
+      "loss": 3.7844,
+      "step": 1119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5788990825688075e-05,
+      "loss": 3.8565,
+      "step": 1120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.578211009174312e-05,
+      "loss": 3.7393,
+      "step": 1121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5775229357798168e-05,
+      "loss": 3.7788,
+      "step": 1122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5768348623853213e-05,
+      "loss": 3.7832,
+      "step": 1123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5761467889908258e-05,
+      "loss": 3.6034,
+      "step": 1124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5754587155963306e-05,
+      "loss": 3.7786,
+      "step": 1125
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5747706422018347e-05,
+      "loss": 3.7652,
+      "step": 1126
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5740825688073396e-05,
+      "loss": 3.7358,
+      "step": 1127
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.573394495412844e-05,
+      "loss": 3.6439,
+      "step": 1128
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5727064220183485e-05,
+      "loss": 3.7187,
+      "step": 1129
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5720183486238533e-05,
+      "loss": 3.6595,
+      "step": 1130
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5713302752293578e-05,
+      "loss": 3.6654,
+      "step": 1131
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5706422018348623e-05,
+      "loss": 3.6736,
+      "step": 1132
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.569954128440367e-05,
+      "loss": 3.5915,
+      "step": 1133
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5692660550458716e-05,
+      "loss": 3.6259,
+      "step": 1134
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5685779816513764e-05,
+      "loss": 3.6045,
+      "step": 1135
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.567889908256881e-05,
+      "loss": 3.4469,
+      "step": 1136
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5672018348623853e-05,
+      "loss": 3.541,
+      "step": 1137
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.56651376146789e-05,
+      "loss": 3.3137,
+      "step": 1138
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5658256880733946e-05,
+      "loss": 3.2642,
+      "step": 1139
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5651376146788994e-05,
+      "loss": 3.2185,
+      "step": 1140
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.564449541284404e-05,
+      "loss": 3.3112,
+      "step": 1141
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.563761467889908e-05,
+      "loss": 3.2239,
+      "step": 1142
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.563073394495413e-05,
+      "loss": 3.1936,
+      "step": 1143
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5623853211009174e-05,
+      "loss": 3.0771,
+      "step": 1144
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.561697247706422e-05,
+      "loss": 2.8935,
+      "step": 1145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5610091743119267e-05,
+      "loss": 2.9473,
+      "step": 1146
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.560321100917431e-05,
+      "loss": 2.8546,
+      "step": 1147
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.559633027522936e-05,
+      "loss": 2.6932,
+      "step": 1148
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5589449541284404e-05,
+      "loss": 2.4038,
+      "step": 1149
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.558256880733945e-05,
+      "loss": 2.3804,
+      "step": 1150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5575688073394497e-05,
+      "loss": 4.6653,
+      "step": 1151
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5568807339449542e-05,
+      "loss": 4.479,
+      "step": 1152
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.556192660550459e-05,
+      "loss": 4.2815,
+      "step": 1153
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5555045871559635e-05,
+      "loss": 4.1139,
+      "step": 1154
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.554816513761468e-05,
+      "loss": 4.0146,
+      "step": 1155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5541284403669728e-05,
+      "loss": 3.9143,
+      "step": 1156
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5534403669724773e-05,
+      "loss": 4.0616,
+      "step": 1157
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5527522935779817e-05,
+      "loss": 3.9746,
+      "step": 1158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5520642201834862e-05,
+      "loss": 3.9596,
+      "step": 1159
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5513761467889907e-05,
+      "loss": 4.1266,
+      "step": 1160
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5506880733944955e-05,
+      "loss": 3.958,
+      "step": 1161
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.55e-05,
+      "loss": 3.7663,
+      "step": 1162
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5493119266055045e-05,
+      "loss": 3.7501,
+      "step": 1163
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5486238532110093e-05,
+      "loss": 3.8764,
+      "step": 1164
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5479357798165138e-05,
+      "loss": 3.9638,
+      "step": 1165
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5472477064220186e-05,
+      "loss": 3.8085,
+      "step": 1166
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.546559633027523e-05,
+      "loss": 3.8257,
+      "step": 1167
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5458715596330275e-05,
+      "loss": 3.8509,
+      "step": 1168
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5451834862385323e-05,
+      "loss": 3.7888,
+      "step": 1169
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5444954128440368e-05,
+      "loss": 3.759,
+      "step": 1170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5438073394495416e-05,
+      "loss": 3.5881,
+      "step": 1171
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.543119266055046e-05,
+      "loss": 3.9265,
+      "step": 1172
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5424311926605506e-05,
+      "loss": 3.6722,
+      "step": 1173
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.541743119266055e-05,
+      "loss": 3.6571,
+      "step": 1174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5410550458715595e-05,
+      "loss": 3.6893,
+      "step": 1175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.540366972477064e-05,
+      "loss": 3.6684,
+      "step": 1176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.539678899082569e-05,
+      "loss": 3.5403,
+      "step": 1177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5389908256880733e-05,
+      "loss": 3.569,
+      "step": 1178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.538302752293578e-05,
+      "loss": 3.5153,
+      "step": 1179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5376146788990826e-05,
+      "loss": 3.6205,
+      "step": 1180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.536926605504587e-05,
+      "loss": 3.5828,
+      "step": 1181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.536238532110092e-05,
+      "loss": 3.5243,
+      "step": 1182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5355504587155964e-05,
+      "loss": 3.5038,
+      "step": 1183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5348623853211012e-05,
+      "loss": 3.5601,
+      "step": 1184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5341743119266057e-05,
+      "loss": 3.5593,
+      "step": 1185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.53348623853211e-05,
+      "loss": 3.4511,
+      "step": 1186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.532798165137615e-05,
+      "loss": 3.4466,
+      "step": 1187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5321100917431194e-05,
+      "loss": 3.4314,
+      "step": 1188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.531422018348624e-05,
+      "loss": 3.2901,
+      "step": 1189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5307339449541284e-05,
+      "loss": 3.2563,
+      "step": 1190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.530045871559633e-05,
+      "loss": 3.2309,
+      "step": 1191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5293577981651377e-05,
+      "loss": 3.205,
+      "step": 1192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.528669724770642e-05,
+      "loss": 3.2765,
+      "step": 1193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5279816513761466e-05,
+      "loss": 3.0936,
+      "step": 1194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5272935779816515e-05,
+      "loss": 2.8022,
+      "step": 1195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.526605504587156e-05,
+      "loss": 2.8253,
+      "step": 1196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5259174311926607e-05,
+      "loss": 2.8255,
+      "step": 1197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5252293577981652e-05,
+      "loss": 2.6196,
+      "step": 1198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5245412844036697e-05,
+      "loss": 2.5366,
+      "step": 1199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5238532110091745e-05,
+      "loss": 2.4052,
+      "step": 1200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.523165137614679e-05,
+      "loss": 4.6744,
+      "step": 1201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5224770642201838e-05,
+      "loss": 4.4056,
+      "step": 1202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5217889908256883e-05,
+      "loss": 4.2716,
+      "step": 1203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5211009174311928e-05,
+      "loss": 4.0462,
+      "step": 1204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5204128440366976e-05,
+      "loss": 4.1424,
+      "step": 1205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5197247706422017e-05,
+      "loss": 3.9523,
+      "step": 1206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5190366972477062e-05,
+      "loss": 3.9599,
+      "step": 1207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.518348623853211e-05,
+      "loss": 3.8349,
+      "step": 1208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5176605504587155e-05,
+      "loss": 3.8738,
+      "step": 1209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5169724770642203e-05,
+      "loss": 3.8525,
+      "step": 1210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5162844036697248e-05,
+      "loss": 3.9866,
+      "step": 1211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5155963302752293e-05,
+      "loss": 4.0298,
+      "step": 1212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.514908256880734e-05,
+      "loss": 3.9032,
+      "step": 1213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5142201834862386e-05,
+      "loss": 3.8705,
+      "step": 1214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5135321100917434e-05,
+      "loss": 3.8573,
+      "step": 1215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.512844036697248e-05,
+      "loss": 3.7831,
+      "step": 1216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5121559633027523e-05,
+      "loss": 3.9009,
+      "step": 1217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.511467889908257e-05,
+      "loss": 3.8467,
+      "step": 1218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5107798165137616e-05,
+      "loss": 3.6523,
+      "step": 1219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.510091743119266e-05,
+      "loss": 3.6793,
+      "step": 1220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.509403669724771e-05,
+      "loss": 3.8232,
+      "step": 1221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.508715596330275e-05,
+      "loss": 3.6637,
+      "step": 1222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.50802752293578e-05,
+      "loss": 3.8058,
+      "step": 1223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5073394495412843e-05,
+      "loss": 3.7015,
+      "step": 1224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5066513761467888e-05,
+      "loss": 3.7267,
+      "step": 1225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5059633027522936e-05,
+      "loss": 3.5554,
+      "step": 1226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.505275229357798e-05,
+      "loss": 3.7953,
+      "step": 1227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.504587155963303e-05,
+      "loss": 3.643,
+      "step": 1228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5038990825688074e-05,
+      "loss": 3.5547,
+      "step": 1229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.503211009174312e-05,
+      "loss": 3.5521,
+      "step": 1230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5025229357798167e-05,
+      "loss": 3.5506,
+      "step": 1231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5018348623853212e-05,
+      "loss": 3.549,
+      "step": 1232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.501146788990826e-05,
+      "loss": 3.4304,
+      "step": 1233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.5004587155963305e-05,
+      "loss": 3.4582,
+      "step": 1234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.499770642201835e-05,
+      "loss": 3.4577,
+      "step": 1235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4990825688073398e-05,
+      "loss": 3.5314,
+      "step": 1236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4983944954128442e-05,
+      "loss": 3.3779,
+      "step": 1237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4977064220183484e-05,
+      "loss": 3.2672,
+      "step": 1238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4970183486238532e-05,
+      "loss": 3.2761,
+      "step": 1239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4963302752293577e-05,
+      "loss": 3.2113,
+      "step": 1240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4956422018348625e-05,
+      "loss": 3.1446,
+      "step": 1241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.494954128440367e-05,
+      "loss": 3.0828,
+      "step": 1242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4942660550458714e-05,
+      "loss": 3.1032,
+      "step": 1243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4935779816513763e-05,
+      "loss": 2.9934,
+      "step": 1244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4928899082568807e-05,
+      "loss": 3.0943,
+      "step": 1245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4922018348623855e-05,
+      "loss": 2.8689,
+      "step": 1246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.49151376146789e-05,
+      "loss": 2.8334,
+      "step": 1247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4908256880733945e-05,
+      "loss": 2.5169,
+      "step": 1248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4901376146788993e-05,
+      "loss": 2.502,
+      "step": 1249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4894495412844038e-05,
+      "loss": 2.2874,
+      "step": 1250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4887614678899083e-05,
+      "loss": 4.6959,
+      "step": 1251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.488073394495413e-05,
+      "loss": 4.5386,
+      "step": 1252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4873853211009176e-05,
+      "loss": 4.2834,
+      "step": 1253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.486697247706422e-05,
+      "loss": 4.2355,
+      "step": 1254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4860091743119265e-05,
+      "loss": 4.1004,
+      "step": 1255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485321100917431e-05,
+      "loss": 4.0496,
+      "step": 1256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4846330275229358e-05,
+      "loss": 4.0017,
+      "step": 1257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4839449541284403e-05,
+      "loss": 3.9998,
+      "step": 1258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483256880733945e-05,
+      "loss": 4.0986,
+      "step": 1259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4825688073394496e-05,
+      "loss": 4.0278,
+      "step": 1260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481880733944954e-05,
+      "loss": 3.8931,
+      "step": 1261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481192660550459e-05,
+      "loss": 4.102,
+      "step": 1262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4805045871559634e-05,
+      "loss": 3.8769,
+      "step": 1263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4798165137614682e-05,
+      "loss": 3.8575,
+      "step": 1264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4791284403669726e-05,
+      "loss": 3.8678,
+      "step": 1265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478440366972477e-05,
+      "loss": 3.9519,
+      "step": 1266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.477752293577982e-05,
+      "loss": 3.9025,
+      "step": 1267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4770642201834864e-05,
+      "loss": 3.7417,
+      "step": 1268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.476376146788991e-05,
+      "loss": 3.7341,
+      "step": 1269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4756880733944954e-05,
+      "loss": 3.7029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475e-05,
+      "loss": 3.7243,
+      "step": 1271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4743119266055047e-05,
+      "loss": 3.7575,
+      "step": 1272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.473623853211009e-05,
+      "loss": 3.6311,
+      "step": 1273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4729357798165136e-05,
+      "loss": 3.8224,
+      "step": 1274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4722477064220184e-05,
+      "loss": 3.664,
+      "step": 1275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.471559633027523e-05,
+      "loss": 3.5353,
+      "step": 1276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4708715596330277e-05,
+      "loss": 3.6799,
+      "step": 1277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4701834862385322e-05,
+      "loss": 3.6196,
+      "step": 1278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4694954128440367e-05,
+      "loss": 3.4565,
+      "step": 1279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4688073394495415e-05,
+      "loss": 3.5875,
+      "step": 1280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.468119266055046e-05,
+      "loss": 3.6072,
+      "step": 1281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4674311926605505e-05,
+      "loss": 3.6802,
+      "step": 1282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4667431192660553e-05,
+      "loss": 3.5494,
+      "step": 1283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4660550458715597e-05,
+      "loss": 3.3854,
+      "step": 1284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4653669724770646e-05,
+      "loss": 3.5027,
+      "step": 1285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4646788990825687e-05,
+      "loss": 3.493,
+      "step": 1286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4639908256880732e-05,
+      "loss": 3.1239,
+      "step": 1287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.463302752293578e-05,
+      "loss": 3.2657,
+      "step": 1288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4626146788990825e-05,
+      "loss": 3.2697,
+      "step": 1289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4619266055045873e-05,
+      "loss": 3.2456,
+      "step": 1290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4612385321100918e-05,
+      "loss": 3.0583,
+      "step": 1291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4605504587155962e-05,
+      "loss": 3.2122,
+      "step": 1292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.459862385321101e-05,
+      "loss": 2.9125,
+      "step": 1293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4591743119266055e-05,
+      "loss": 3.0389,
+      "step": 1294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4584862385321104e-05,
+      "loss": 2.8405,
+      "step": 1295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4577981651376148e-05,
+      "loss": 2.7485,
+      "step": 1296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4571100917431193e-05,
+      "loss": 2.6293,
+      "step": 1297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.456422018348624e-05,
+      "loss": 2.61,
+      "step": 1298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4557339449541286e-05,
+      "loss": 2.2357,
+      "step": 1299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.455045871559633e-05,
+      "loss": 2.1484,
+      "step": 1300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.454357798165138e-05,
+      "loss": 4.592,
+      "step": 1301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.453669724770642e-05,
+      "loss": 4.3477,
+      "step": 1302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.452981651376147e-05,
+      "loss": 4.1946,
+      "step": 1303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4522935779816513e-05,
+      "loss": 4.1174,
+      "step": 1304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4516055045871558e-05,
+      "loss": 4.0231,
+      "step": 1305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4509174311926606e-05,
+      "loss": 3.9032,
+      "step": 1306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.450229357798165e-05,
+      "loss": 3.9959,
+      "step": 1307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.44954128440367e-05,
+      "loss": 3.9433,
+      "step": 1308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4488532110091744e-05,
+      "loss": 3.9787,
+      "step": 1309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.448165137614679e-05,
+      "loss": 3.9436,
+      "step": 1310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4474770642201837e-05,
+      "loss": 3.7523,
+      "step": 1311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.446788990825688e-05,
+      "loss": 3.9256,
+      "step": 1312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4461009174311926e-05,
+      "loss": 3.9166,
+      "step": 1313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4454128440366975e-05,
+      "loss": 3.8386,
+      "step": 1314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.444724770642202e-05,
+      "loss": 3.7338,
+      "step": 1315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4440366972477067e-05,
+      "loss": 3.8051,
+      "step": 1316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4433486238532112e-05,
+      "loss": 3.7415,
+      "step": 1317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4426605504587154e-05,
+      "loss": 3.7258,
+      "step": 1318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4419724770642202e-05,
+      "loss": 3.7022,
+      "step": 1319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.4412844036697247e-05,
+      "loss": 3.7687,
+      "step": 1320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4405963302752295e-05,
+      "loss": 3.6721,
+      "step": 1321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.439908256880734e-05,
+      "loss": 3.5876,
+      "step": 1322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4392201834862384e-05,
+      "loss": 3.7133,
+      "step": 1323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4385321100917432e-05,
+      "loss": 3.5554,
+      "step": 1324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4378440366972477e-05,
+      "loss": 3.664,
+      "step": 1325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4371559633027525e-05,
+      "loss": 3.6477,
+      "step": 1326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.436467889908257e-05,
+      "loss": 3.6241,
+      "step": 1327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4357798165137615e-05,
+      "loss": 3.6255,
+      "step": 1328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4350917431192663e-05,
+      "loss": 3.4806,
+      "step": 1329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4344036697247708e-05,
+      "loss": 3.4136,
+      "step": 1330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4337155963302753e-05,
+      "loss": 3.424,
+      "step": 1331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.43302752293578e-05,
+      "loss": 3.618,
+      "step": 1332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4323394495412846e-05,
+      "loss": 3.5606,
+      "step": 1333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.431651376146789e-05,
+      "loss": 3.3669,
+      "step": 1334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.4309633027522935e-05,
+      "loss": 3.4653,
+      "step": 1335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.430275229357798e-05,
+      "loss": 3.4803,
+      "step": 1336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4295871559633028e-05,
+      "loss": 3.235,
+      "step": 1337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4288990825688073e-05,
+      "loss": 3.3147,
+      "step": 1338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.428211009174312e-05,
+      "loss": 3.3446,
+      "step": 1339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4275229357798166e-05,
+      "loss": 3.2564,
+      "step": 1340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.426834862385321e-05,
+      "loss": 3.1659,
+      "step": 1341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.426146788990826e-05,
+      "loss": 3.1153,
+      "step": 1342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4254587155963303e-05,
+      "loss": 3.0939,
+      "step": 1343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4247706422018348e-05,
+      "loss": 2.9229,
+      "step": 1344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4240825688073396e-05,
+      "loss": 2.9258,
+      "step": 1345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.423394495412844e-05,
+      "loss": 2.7025,
+      "step": 1346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.422706422018349e-05,
+      "loss": 2.7989,
+      "step": 1347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4220183486238534e-05,
+      "loss": 2.5108,
+      "step": 1348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.421330275229358e-05,
+      "loss": 2.3328,
+      "step": 1349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4206422018348624e-05,
+      "loss": 2.3285,
+      "step": 1350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.419954128440367e-05,
+      "loss": 4.8696,
+      "step": 1351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.4192660550458717e-05,
+      "loss": 4.3979,
+      "step": 1352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.418577981651376e-05,
+      "loss": 4.2494,
+      "step": 1353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4178899082568806e-05,
+      "loss": 4.1156,
+      "step": 1354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4172018348623854e-05,
+      "loss": 3.9304,
+      "step": 1355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.41651376146789e-05,
+      "loss": 3.9746,
+      "step": 1356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4158256880733947e-05,
+      "loss": 3.8103,
+      "step": 1357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4151376146788992e-05,
+      "loss": 3.9214,
+      "step": 1358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4144495412844037e-05,
+      "loss": 3.8287,
+      "step": 1359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4137614678899085e-05,
+      "loss": 3.838,
+      "step": 1360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.413073394495413e-05,
+      "loss": 3.9451,
+      "step": 1361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4123853211009174e-05,
+      "loss": 3.8318,
+      "step": 1362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4116972477064223e-05,
+      "loss": 3.7878,
+      "step": 1363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4110091743119267e-05,
+      "loss": 3.9019,
+      "step": 1364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4103211009174315e-05,
+      "loss": 3.8596,
+      "step": 1365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4096330275229357e-05,
+      "loss": 3.6885,
+      "step": 1366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.40894495412844e-05,
+      "loss": 3.7466,
+      "step": 1367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.408256880733945e-05,
+      "loss": 3.8246,
+      "step": 1368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.4075688073394495e-05,
+      "loss": 3.4659,
+      "step": 1369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4068807339449543e-05,
+      "loss": 3.7257,
+      "step": 1370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4061926605504587e-05,
+      "loss": 3.7745,
+      "step": 1371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4055045871559632e-05,
+      "loss": 3.6234,
+      "step": 1372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.404816513761468e-05,
+      "loss": 3.6291,
+      "step": 1373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4041284403669725e-05,
+      "loss": 3.6946,
+      "step": 1374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.403440366972477e-05,
+      "loss": 3.6744,
+      "step": 1375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4027522935779818e-05,
+      "loss": 3.6959,
+      "step": 1376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4020642201834863e-05,
+      "loss": 3.6046,
+      "step": 1377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.401376146788991e-05,
+      "loss": 3.4543,
+      "step": 1378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4006880733944956e-05,
+      "loss": 3.4634,
+      "step": 1379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.4e-05,
+      "loss": 3.5337,
+      "step": 1380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.399311926605505e-05,
+      "loss": 3.6098,
+      "step": 1381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.398623853211009e-05,
+      "loss": 3.4776,
+      "step": 1382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.397935779816514e-05,
+      "loss": 3.2452,
+      "step": 1383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3972477064220183e-05,
+      "loss": 3.2739,
+      "step": 1384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3965596330275228e-05,
+      "loss": 3.422,
+      "step": 1385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3958715596330276e-05,
+      "loss": 3.2299,
+      "step": 1386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.395183486238532e-05,
+      "loss": 3.2096,
+      "step": 1387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3944954128440366e-05,
+      "loss": 3.1842,
+      "step": 1388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3938073394495414e-05,
+      "loss": 3.3082,
+      "step": 1389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.393119266055046e-05,
+      "loss": 3.2946,
+      "step": 1390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3924311926605507e-05,
+      "loss": 3.2161,
+      "step": 1391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.391743119266055e-05,
+      "loss": 3.0696,
+      "step": 1392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3910550458715596e-05,
+      "loss": 2.9346,
+      "step": 1393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3903669724770644e-05,
+      "loss": 2.9735,
+      "step": 1394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.389678899082569e-05,
+      "loss": 2.9075,
+      "step": 1395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3889908256880737e-05,
+      "loss": 2.7203,
+      "step": 1396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3883027522935782e-05,
+      "loss": 2.5974,
+      "step": 1397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3876146788990823e-05,
+      "loss": 2.5324,
+      "step": 1398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.386926605504587e-05,
+      "loss": 2.4269,
+      "step": 1399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3862385321100916e-05,
+      "loss": 2.165,
+      "step": 1400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3855504587155965e-05,
+      "loss": 4.4961,
+      "step": 1401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.384862385321101e-05,
+      "loss": 4.2463,
+      "step": 1402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3841743119266054e-05,
+      "loss": 4.2125,
+      "step": 1403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3834862385321102e-05,
+      "loss": 4.0898,
+      "step": 1404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3827981651376147e-05,
+      "loss": 3.8851,
+      "step": 1405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3821100917431192e-05,
+      "loss": 3.835,
+      "step": 1406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.381422018348624e-05,
+      "loss": 3.8333,
+      "step": 1407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3807339449541285e-05,
+      "loss": 3.8626,
+      "step": 1408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3800458715596333e-05,
+      "loss": 3.8788,
+      "step": 1409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3793577981651378e-05,
+      "loss": 3.7103,
+      "step": 1410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3786697247706422e-05,
+      "loss": 3.8224,
+      "step": 1411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.377981651376147e-05,
+      "loss": 3.7108,
+      "step": 1412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3772935779816515e-05,
+      "loss": 3.8111,
+      "step": 1413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.376605504587156e-05,
+      "loss": 3.6052,
+      "step": 1414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3759174311926605e-05,
+      "loss": 3.768,
+      "step": 1415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.375229357798165e-05,
+      "loss": 3.7261,
+      "step": 1416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3745412844036698e-05,
+      "loss": 3.7803,
+      "step": 1417
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3738532110091743e-05,
+      "loss": 3.79,
+      "step": 1418
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3731651376146787e-05,
+      "loss": 3.5853,
+      "step": 1419
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3724770642201836e-05,
+      "loss": 3.6816,
+      "step": 1420
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.371788990825688e-05,
+      "loss": 3.5913,
+      "step": 1421
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.371100917431193e-05,
+      "loss": 3.5655,
+      "step": 1422
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3704128440366973e-05,
+      "loss": 3.608,
+      "step": 1423
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3697247706422018e-05,
+      "loss": 3.5837,
+      "step": 1424
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3690366972477066e-05,
+      "loss": 3.6352,
+      "step": 1425
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.368348623853211e-05,
+      "loss": 3.5505,
+      "step": 1426
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.367660550458716e-05,
+      "loss": 3.5986,
+      "step": 1427
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3669724770642204e-05,
+      "loss": 3.4852,
+      "step": 1428
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.366284403669725e-05,
+      "loss": 3.65,
+      "step": 1429
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3655963302752293e-05,
+      "loss": 3.4722,
+      "step": 1430
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3649082568807338e-05,
+      "loss": 3.4306,
+      "step": 1431
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3642201834862386e-05,
+      "loss": 3.4088,
+      "step": 1432
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.363532110091743e-05,
+      "loss": 3.3661,
+      "step": 1433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3628440366972476e-05,
+      "loss": 3.5432,
+      "step": 1434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3621559633027524e-05,
+      "loss": 3.4629,
+      "step": 1435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.361467889908257e-05,
+      "loss": 3.404,
+      "step": 1436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3607798165137614e-05,
+      "loss": 3.1121,
+      "step": 1437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3600917431192662e-05,
+      "loss": 3.2623,
+      "step": 1438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3594036697247707e-05,
+      "loss": 3.1566,
+      "step": 1439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3587155963302755e-05,
+      "loss": 3.0189,
+      "step": 1440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.35802752293578e-05,
+      "loss": 2.9821,
+      "step": 1441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3573394495412844e-05,
+      "loss": 2.9894,
+      "step": 1442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3566513761467892e-05,
+      "loss": 2.9422,
+      "step": 1443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3559633027522937e-05,
+      "loss": 2.8746,
+      "step": 1444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3552752293577985e-05,
+      "loss": 2.848,
+      "step": 1445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3545871559633027e-05,
+      "loss": 2.635,
+      "step": 1446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.353899082568807e-05,
+      "loss": 2.456,
+      "step": 1447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.353211009174312e-05,
+      "loss": 2.4307,
+      "step": 1448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3525229357798164e-05,
+      "loss": 2.2347,
+      "step": 1449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.351834862385321e-05,
+      "loss": 2.2021,
+      "step": 1450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3511467889908257e-05,
+      "loss": 4.5119,
+      "step": 1451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3504587155963302e-05,
+      "loss": 4.3281,
+      "step": 1452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.349770642201835e-05,
+      "loss": 4.2756,
+      "step": 1453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3490825688073395e-05,
+      "loss": 3.9256,
+      "step": 1454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.348394495412844e-05,
+      "loss": 3.8621,
+      "step": 1455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3477064220183488e-05,
+      "loss": 3.9062,
+      "step": 1456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3470183486238533e-05,
+      "loss": 3.9663,
+      "step": 1457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.346330275229358e-05,
+      "loss": 3.8374,
+      "step": 1458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3456422018348626e-05,
+      "loss": 3.8223,
+      "step": 1459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.344954128440367e-05,
+      "loss": 3.925,
+      "step": 1460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.344266055045872e-05,
+      "loss": 3.788,
+      "step": 1461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.343577981651376e-05,
+      "loss": 3.9209,
+      "step": 1462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3428899082568808e-05,
+      "loss": 3.6784,
+      "step": 1463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3422018348623853e-05,
+      "loss": 3.5394,
+      "step": 1464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3415137614678898e-05,
+      "loss": 3.7465,
+      "step": 1465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3408256880733946e-05,
+      "loss": 3.6699,
+      "step": 1466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.340137614678899e-05,
+      "loss": 3.5663,
+      "step": 1467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3394495412844035e-05,
+      "loss": 3.653,
+      "step": 1468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3387614678899084e-05,
+      "loss": 3.7491,
+      "step": 1469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.338073394495413e-05,
+      "loss": 3.6034,
+      "step": 1470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3373853211009176e-05,
+      "loss": 3.5902,
+      "step": 1471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.336697247706422e-05,
+      "loss": 3.7989,
+      "step": 1472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3360091743119266e-05,
+      "loss": 3.5565,
+      "step": 1473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3353211009174314e-05,
+      "loss": 3.6887,
+      "step": 1474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.334633027522936e-05,
+      "loss": 3.4496,
+      "step": 1475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3339449541284407e-05,
+      "loss": 3.4525,
+      "step": 1476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3332568807339452e-05,
+      "loss": 3.4953,
+      "step": 1477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3325688073394493e-05,
+      "loss": 3.4592,
+      "step": 1478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.331880733944954e-05,
+      "loss": 3.5576,
+      "step": 1479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3311926605504586e-05,
+      "loss": 3.3052,
+      "step": 1480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330504587155963e-05,
+      "loss": 3.2738,
+      "step": 1481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.329816513761468e-05,
+      "loss": 3.4464,
+      "step": 1482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3291284403669724e-05,
+      "loss": 3.2706,
+      "step": 1483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3284403669724772e-05,
+      "loss": 3.2341,
+      "step": 1484
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3277522935779817e-05,
+      "loss": 3.3155,
+      "step": 1485
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327064220183486e-05,
+      "loss": 3.2247,
+      "step": 1486
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.326376146788991e-05,
+      "loss": 3.1707,
+      "step": 1487
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3256880733944955e-05,
+      "loss": 3.2067,
+      "step": 1488
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 3.0924,
+      "step": 1489
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3243119266055047e-05,
+      "loss": 2.926,
+      "step": 1490
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3236238532110092e-05,
+      "loss": 3.0387,
+      "step": 1491
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.322935779816514e-05,
+      "loss": 2.7168,
+      "step": 1492
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3222477064220185e-05,
+      "loss": 2.993,
+      "step": 1493
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.321559633027523e-05,
+      "loss": 2.854,
+      "step": 1494
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3208715596330275e-05,
+      "loss": 2.7578,
+      "step": 1495
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.320183486238532e-05,
+      "loss": 2.7039,
+      "step": 1496
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3194954128440368e-05,
+      "loss": 2.3932,
+      "step": 1497
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3188073394495412e-05,
+      "loss": 2.3995,
+      "step": 1498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3181192660550457e-05,
+      "loss": 2.3981,
+      "step": 1499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3174311926605505e-05,
+      "loss": 2.2152,
+      "step": 1500
+    },
+    {
+      "epoch": 0.93,
+      "eval_bleu": 0.0,
+      "eval_loss": 3.9444446563720703,
+      "eval_runtime": 2571.7557,
+      "eval_samples_per_second": 5.739,
+      "eval_steps_per_second": 0.717,
+      "step": 1500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.316743119266055e-05,
+      "loss": 4.5045,
+      "step": 1501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3160550458715598e-05,
+      "loss": 4.4275,
+      "step": 1502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3153669724770643e-05,
+      "loss": 4.1468,
+      "step": 1503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3146788990825688e-05,
+      "loss": 4.1388,
+      "step": 1504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3139908256880736e-05,
+      "loss": 3.9561,
+      "step": 1505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.313302752293578e-05,
+      "loss": 3.7509,
+      "step": 1506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.312614678899083e-05,
+      "loss": 3.8607,
+      "step": 1507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3119266055045874e-05,
+      "loss": 3.9173,
+      "step": 1508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.311238532110092e-05,
+      "loss": 3.8576,
+      "step": 1509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3105504587155963e-05,
+      "loss": 3.8649,
+      "step": 1510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3098623853211008e-05,
+      "loss": 3.7506,
+      "step": 1511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3091743119266053e-05,
+      "loss": 3.8002,
+      "step": 1512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.30848623853211e-05,
+      "loss": 3.7694,
+      "step": 1513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3077981651376146e-05,
+      "loss": 3.6663,
+      "step": 1514
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3071100917431194e-05,
+      "loss": 3.5367,
+      "step": 1515
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.306422018348624e-05,
+      "loss": 3.6211,
+      "step": 1516
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3057339449541283e-05,
+      "loss": 3.6132,
+      "step": 1517
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.305045871559633e-05,
+      "loss": 3.7383,
+      "step": 1518
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3043577981651376e-05,
+      "loss": 3.5845,
+      "step": 1519
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3036697247706425e-05,
+      "loss": 3.7685,
+      "step": 1520
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.302981651376147e-05,
+      "loss": 3.5512,
+      "step": 1521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3022935779816514e-05,
+      "loss": 3.761,
+      "step": 1522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3016055045871562e-05,
+      "loss": 3.6506,
+      "step": 1523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3009174311926607e-05,
+      "loss": 3.5463,
+      "step": 1524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3002293577981652e-05,
+      "loss": 3.5885,
+      "step": 1525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2995412844036697e-05,
+      "loss": 3.6125,
+      "step": 1526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.298853211009174e-05,
+      "loss": 3.5687,
+      "step": 1527
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.298165137614679e-05,
+      "loss": 3.5502,
+      "step": 1528
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2974770642201834e-05,
+      "loss": 3.4581,
+      "step": 1529
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.296788990825688e-05,
+      "loss": 3.4186,
+      "step": 1530
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2961009174311927e-05,
+      "loss": 3.3705,
+      "step": 1531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2954128440366972e-05,
+      "loss": 3.5034,
+      "step": 1532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.294724770642202e-05,
+      "loss": 3.2403,
+      "step": 1533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2940366972477065e-05,
+      "loss": 3.307,
+      "step": 1534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.293348623853211e-05,
+      "loss": 3.1666,
+      "step": 1535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2926605504587158e-05,
+      "loss": 3.1993,
+      "step": 1536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2919724770642203e-05,
+      "loss": 3.3693,
+      "step": 1537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.291284403669725e-05,
+      "loss": 3.2454,
+      "step": 1538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2905963302752296e-05,
+      "loss": 2.9069,
+      "step": 1539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.289908256880734e-05,
+      "loss": 2.9224,
+      "step": 1540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.289220183486239e-05,
+      "loss": 2.9794,
+      "step": 1541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.288532110091743e-05,
+      "loss": 2.9093,
+      "step": 1542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2878440366972475e-05,
+      "loss": 2.7868,
+      "step": 1543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2871559633027523e-05,
+      "loss": 2.6584,
+      "step": 1544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2864678899082568e-05,
+      "loss": 2.744,
+      "step": 1545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.2857798165137616e-05,
+      "loss": 2.63,
+      "step": 1546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.285091743119266e-05,
+      "loss": 2.3973,
+      "step": 1547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2844036697247705e-05,
+      "loss": 2.3384,
+      "step": 1548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2837155963302753e-05,
+      "loss": 2.222,
+      "step": 1549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2830275229357798e-05,
+      "loss": 2.0857,
+      "step": 1550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2823394495412846e-05,
+      "loss": 4.5337,
+      "step": 1551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.281651376146789e-05,
+      "loss": 4.4446,
+      "step": 1552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2809633027522936e-05,
+      "loss": 4.0854,
+      "step": 1553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2802752293577984e-05,
+      "loss": 4.0876,
+      "step": 1554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.279587155963303e-05,
+      "loss": 3.9287,
+      "step": 1555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2788990825688074e-05,
+      "loss": 3.9667,
+      "step": 1556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2782110091743122e-05,
+      "loss": 3.823,
+      "step": 1557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2775229357798163e-05,
+      "loss": 3.6891,
+      "step": 1558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.276834862385321e-05,
+      "loss": 3.7817,
+      "step": 1559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2761467889908256e-05,
+      "loss": 3.7291,
+      "step": 1560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.27545871559633e-05,
+      "loss": 3.8518,
+      "step": 1561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.274770642201835e-05,
+      "loss": 3.7277,
+      "step": 1562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2740825688073394e-05,
+      "loss": 3.7964,
+      "step": 1563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2733944954128442e-05,
+      "loss": 3.5873,
+      "step": 1564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2727064220183487e-05,
+      "loss": 3.7128,
+      "step": 1565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.272018348623853e-05,
+      "loss": 3.6148,
+      "step": 1566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.271330275229358e-05,
+      "loss": 3.7163,
+      "step": 1567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2706422018348624e-05,
+      "loss": 3.5822,
+      "step": 1568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2699541284403673e-05,
+      "loss": 3.6566,
+      "step": 1569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2692660550458717e-05,
+      "loss": 3.6833,
+      "step": 1570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2685779816513762e-05,
+      "loss": 3.5389,
+      "step": 1571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.267889908256881e-05,
+      "loss": 3.5073,
+      "step": 1572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2672018348623855e-05,
+      "loss": 3.4217,
+      "step": 1573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2665137614678896e-05,
+      "loss": 3.5031,
+      "step": 1574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2658256880733945e-05,
+      "loss": 3.4914,
+      "step": 1575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.265137614678899e-05,
+      "loss": 3.5753,
+      "step": 1576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2644495412844037e-05,
+      "loss": 3.3675,
+      "step": 1577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2637614678899082e-05,
+      "loss": 3.4845,
+      "step": 1578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2630733944954127e-05,
+      "loss": 3.2304,
+      "step": 1579
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2623853211009175e-05,
+      "loss": 3.2734,
+      "step": 1580
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.261697247706422e-05,
+      "loss": 3.2365,
+      "step": 1581
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2610091743119268e-05,
+      "loss": 3.2411,
+      "step": 1582
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2603211009174313e-05,
+      "loss": 3.2773,
+      "step": 1583
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2596330275229358e-05,
+      "loss": 3.3307,
+      "step": 1584
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2589449541284406e-05,
+      "loss": 3.1079,
+      "step": 1585
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.258256880733945e-05,
+      "loss": 3.1251,
+      "step": 1586
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2575688073394495e-05,
+      "loss": 3.0983,
+      "step": 1587
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2568807339449544e-05,
+      "loss": 3.0887,
+      "step": 1588
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.256192660550459e-05,
+      "loss": 3.0837,
+      "step": 1589
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2555045871559633e-05,
+      "loss": 2.968,
+      "step": 1590
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2548165137614678e-05,
+      "loss": 2.8127,
+      "step": 1591
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2541284403669723e-05,
+      "loss": 2.9253,
+      "step": 1592
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.253440366972477e-05,
+      "loss": 2.9571,
+      "step": 1593
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2527522935779816e-05,
+      "loss": 2.777,
+      "step": 1594
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2520642201834864e-05,
+      "loss": 2.4987,
+      "step": 1595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.251376146788991e-05,
+      "loss": 2.4314,
+      "step": 1596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2506880733944953e-05,
+      "loss": 2.3759,
+      "step": 1597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.25e-05,
+      "loss": 2.1874,
+      "step": 1598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2493119266055046e-05,
+      "loss": 2.0956,
+      "step": 1599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2486238532110094e-05,
+      "loss": 2.1749,
+      "step": 1600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.247935779816514e-05,
+      "loss": 4.4612,
+      "step": 1601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2472477064220184e-05,
+      "loss": 3.9922,
+      "step": 1602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2465596330275232e-05,
+      "loss": 3.9675,
+      "step": 1603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2458715596330277e-05,
+      "loss": 3.8366,
+      "step": 1604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.245183486238532e-05,
+      "loss": 3.7913,
+      "step": 1605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.244495412844037e-05,
+      "loss": 3.7334,
+      "step": 1606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.243807339449541e-05,
+      "loss": 3.5052,
+      "step": 1607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.243119266055046e-05,
+      "loss": 3.5914,
+      "step": 1608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2424311926605504e-05,
+      "loss": 3.6788,
+      "step": 1609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.241743119266055e-05,
+      "loss": 3.4714,
+      "step": 1610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2410550458715597e-05,
+      "loss": 3.4446,
+      "step": 1611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2403669724770642e-05,
+      "loss": 3.5588,
+      "step": 1612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.239678899082569e-05,
+      "loss": 3.4161,
+      "step": 1613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2389908256880735e-05,
+      "loss": 3.2797,
+      "step": 1614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.238302752293578e-05,
+      "loss": 3.1281,
+      "step": 1615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2376146788990828e-05,
+      "loss": 3.0048,
+      "step": 1616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2369266055045872e-05,
+      "loss": 2.9144,
+      "step": 1617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2362385321100917e-05,
+      "loss": 2.8673,
+      "step": 1618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2355504587155965e-05,
+      "loss": 2.3912,
+      "step": 1619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.234862385321101e-05,
+      "loss": 2.2024,
+      "step": 1620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2341743119266058e-05,
+      "loss": 4.56,
+      "step": 1621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2334862385321103e-05,
+      "loss": 4.2357,
+      "step": 1622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2327981651376144e-05,
+      "loss": 4.0313,
+      "step": 1623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2321100917431193e-05,
+      "loss": 4.1243,
+      "step": 1624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2314220183486237e-05,
+      "loss": 3.803,
+      "step": 1625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2307339449541286e-05,
+      "loss": 3.4489,
+      "step": 1626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.230045871559633e-05,
+      "loss": 3.7082,
+      "step": 1627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2293577981651375e-05,
+      "loss": 3.6517,
+      "step": 1628
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2286697247706423e-05,
+      "loss": 3.7157,
+      "step": 1629
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2279816513761468e-05,
+      "loss": 3.5589,
+      "step": 1630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2272935779816516e-05,
+      "loss": 3.5505,
+      "step": 1631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.226605504587156e-05,
+      "loss": 3.4998,
+      "step": 1632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2259174311926606e-05,
+      "loss": 3.5316,
+      "step": 1633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2252293577981654e-05,
+      "loss": 3.5267,
+      "step": 1634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.22454128440367e-05,
+      "loss": 3.526,
+      "step": 1635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2238532110091743e-05,
+      "loss": 3.4707,
+      "step": 1636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.223165137614679e-05,
+      "loss": 3.4092,
+      "step": 1637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2224770642201836e-05,
+      "loss": 3.507,
+      "step": 1638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.221788990825688e-05,
+      "loss": 3.3493,
+      "step": 1639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2211009174311926e-05,
+      "loss": 3.3852,
+      "step": 1640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.220412844036697e-05,
+      "loss": 3.3935,
+      "step": 1641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.219724770642202e-05,
+      "loss": 3.3901,
+      "step": 1642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2190366972477064e-05,
+      "loss": 3.3328,
+      "step": 1643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.2183486238532112e-05,
+      "loss": 3.4765,
+      "step": 1644
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2176605504587157e-05,
+      "loss": 3.4612,
+      "step": 1645
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.21697247706422e-05,
+      "loss": 3.3267,
+      "step": 1646
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.216284403669725e-05,
+      "loss": 3.2772,
+      "step": 1647
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2155963302752294e-05,
+      "loss": 3.2646,
+      "step": 1648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.214908256880734e-05,
+      "loss": 3.3178,
+      "step": 1649
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2142201834862387e-05,
+      "loss": 3.1696,
+      "step": 1650
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2135321100917432e-05,
+      "loss": 3.2865,
+      "step": 1651
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.212844036697248e-05,
+      "loss": 3.2339,
+      "step": 1652
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2121559633027525e-05,
+      "loss": 3.106,
+      "step": 1653
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.211467889908257e-05,
+      "loss": 3.0468,
+      "step": 1654
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2107798165137614e-05,
+      "loss": 3.1457,
+      "step": 1655
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.210091743119266e-05,
+      "loss": 3.0602,
+      "step": 1656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2094036697247707e-05,
+      "loss": 2.9287,
+      "step": 1657
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2087155963302752e-05,
+      "loss": 2.9742,
+      "step": 1658
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2080275229357797e-05,
+      "loss": 2.9453,
+      "step": 1659
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2073394495412845e-05,
+      "loss": 2.9503,
+      "step": 1660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.206651376146789e-05,
+      "loss": 2.9596,
+      "step": 1661
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2059633027522938e-05,
+      "loss": 2.6819,
+      "step": 1662
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2052752293577983e-05,
+      "loss": 2.6427,
+      "step": 1663
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2045871559633028e-05,
+      "loss": 2.5944,
+      "step": 1664
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2038990825688076e-05,
+      "loss": 2.3338,
+      "step": 1665
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.203211009174312e-05,
+      "loss": 2.4154,
+      "step": 1666
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2025229357798165e-05,
+      "loss": 2.1834,
+      "step": 1667
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2018348623853213e-05,
+      "loss": 2.1656,
+      "step": 1668
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2011467889908258e-05,
+      "loss": 2.0295,
+      "step": 1669
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.2004587155963306e-05,
+      "loss": 2.1093,
+      "step": 1670
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.1997706422018348e-05,
+      "loss": 4.2849,
+      "step": 1671
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.1990825688073392e-05,
+      "loss": 4.03,
+      "step": 1672
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.198394495412844e-05,
+      "loss": 4.0115,
+      "step": 1673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.1977064220183485e-05,
+      "loss": 3.9228,
+      "step": 1674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.1970183486238534e-05,
+      "loss": 3.7942,
+      "step": 1675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.196330275229358e-05,
+      "loss": 3.7149,
+      "step": 1676
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1956422018348623e-05,
+      "loss": 3.4985,
+      "step": 1677
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.194954128440367e-05,
+      "loss": 3.6547,
+      "step": 1678
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1942660550458716e-05,
+      "loss": 3.6034,
+      "step": 1679
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.193577981651376e-05,
+      "loss": 3.5378,
+      "step": 1680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.192889908256881e-05,
+      "loss": 3.6404,
+      "step": 1681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1922018348623854e-05,
+      "loss": 3.5737,
+      "step": 1682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1915137614678902e-05,
+      "loss": 3.4444,
+      "step": 1683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1908256880733947e-05,
+      "loss": 3.6921,
+      "step": 1684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.190137614678899e-05,
+      "loss": 3.509,
+      "step": 1685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.189449541284404e-05,
+      "loss": 3.4852,
+      "step": 1686
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.188761467889908e-05,
+      "loss": 3.5038,
+      "step": 1687
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.188073394495413e-05,
+      "loss": 3.4266,
+      "step": 1688
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1873853211009174e-05,
+      "loss": 3.4473,
+      "step": 1689
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.186697247706422e-05,
+      "loss": 3.3783,
+      "step": 1690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.1860091743119267e-05,
+      "loss": 3.4614,
+      "step": 1691
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.185321100917431e-05,
+      "loss": 3.2721,
+      "step": 1692
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.184633027522936e-05,
+      "loss": 3.3977,
+      "step": 1693
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1839449541284405e-05,
+      "loss": 3.3287,
+      "step": 1694
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.183256880733945e-05,
+      "loss": 3.2619,
+      "step": 1695
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1825688073394497e-05,
+      "loss": 3.2887,
+      "step": 1696
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1818807339449542e-05,
+      "loss": 3.2643,
+      "step": 1697
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1811926605504587e-05,
+      "loss": 3.1692,
+      "step": 1698
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1805045871559635e-05,
+      "loss": 3.3485,
+      "step": 1699
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.179816513761468e-05,
+      "loss": 3.4201,
+      "step": 1700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1791284403669728e-05,
+      "loss": 3.2611,
+      "step": 1701
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1784403669724773e-05,
+      "loss": 3.0801,
+      "step": 1702
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1777522935779814e-05,
+      "loss": 3.0397,
+      "step": 1703
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1770642201834862e-05,
+      "loss": 2.9998,
+      "step": 1704
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1763761467889907e-05,
+      "loss": 3.0227,
+      "step": 1705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1756880733944955e-05,
+      "loss": 2.8953,
+      "step": 1706
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.175e-05,
+      "loss": 2.848,
+      "step": 1707
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1743119266055045e-05,
+      "loss": 2.852,
+      "step": 1708
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.1736238532110093e-05,
+      "loss": 2.8454,
+      "step": 1709
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1729357798165138e-05,
+      "loss": 2.7893,
+      "step": 1710
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1722477064220183e-05,
+      "loss": 2.708,
+      "step": 1711
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.171559633027523e-05,
+      "loss": 2.5183,
+      "step": 1712
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1708715596330276e-05,
+      "loss": 2.506,
+      "step": 1713
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1701834862385324e-05,
+      "loss": 2.5679,
+      "step": 1714
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.169495412844037e-05,
+      "loss": 2.3361,
+      "step": 1715
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1688073394495413e-05,
+      "loss": 2.3786,
+      "step": 1716
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.168119266055046e-05,
+      "loss": 2.2877,
+      "step": 1717
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1674311926605506e-05,
+      "loss": 2.2128,
+      "step": 1718
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.166743119266055e-05,
+      "loss": 1.9993,
+      "step": 1719
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1660550458715596e-05,
+      "loss": 2.0325,
+      "step": 1720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.165366972477064e-05,
+      "loss": 4.3171,
+      "step": 1721
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.164678899082569e-05,
+      "loss": 4.0878,
+      "step": 1722
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1639908256880733e-05,
+      "loss": 3.9876,
+      "step": 1723
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.163302752293578e-05,
+      "loss": 3.8516,
+      "step": 1724
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.1626146788990826e-05,
+      "loss": 3.7069,
+      "step": 1725
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.161926605504587e-05,
+      "loss": 3.702,
+      "step": 1726
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.161238532110092e-05,
+      "loss": 3.6594,
+      "step": 1727
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1605504587155964e-05,
+      "loss": 3.745,
+      "step": 1728
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.159862385321101e-05,
+      "loss": 3.5483,
+      "step": 1729
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1591743119266057e-05,
+      "loss": 3.5725,
+      "step": 1730
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1584862385321102e-05,
+      "loss": 3.3474,
+      "step": 1731
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.157798165137615e-05,
+      "loss": 3.5036,
+      "step": 1732
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1571100917431195e-05,
+      "loss": 3.5824,
+      "step": 1733
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.156422018348624e-05,
+      "loss": 3.5531,
+      "step": 1734
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1557339449541284e-05,
+      "loss": 3.465,
+      "step": 1735
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.155045871559633e-05,
+      "loss": 3.4196,
+      "step": 1736
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1543577981651377e-05,
+      "loss": 3.5226,
+      "step": 1737
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1536697247706422e-05,
+      "loss": 3.4664,
+      "step": 1738
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1529816513761467e-05,
+      "loss": 3.3708,
+      "step": 1739
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.1522935779816515e-05,
+      "loss": 3.3802,
+      "step": 1740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.151605504587156e-05,
+      "loss": 3.2033,
+      "step": 1741
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1509174311926604e-05,
+      "loss": 3.2314,
+      "step": 1742
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1502293577981653e-05,
+      "loss": 3.3027,
+      "step": 1743
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1495412844036697e-05,
+      "loss": 3.1097,
+      "step": 1744
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1488532110091746e-05,
+      "loss": 3.1831,
+      "step": 1745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.148165137614679e-05,
+      "loss": 3.2118,
+      "step": 1746
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1474770642201835e-05,
+      "loss": 3.2635,
+      "step": 1747
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1467889908256883e-05,
+      "loss": 3.2431,
+      "step": 1748
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1461009174311928e-05,
+      "loss": 3.1801,
+      "step": 1749
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1454128440366976e-05,
+      "loss": 3.2763,
+      "step": 1750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1447247706422018e-05,
+      "loss": 3.2694,
+      "step": 1751
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1440366972477062e-05,
+      "loss": 3.1029,
+      "step": 1752
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.143348623853211e-05,
+      "loss": 3.1707,
+      "step": 1753
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1426605504587155e-05,
+      "loss": 2.8702,
+      "step": 1754
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1419724770642203e-05,
+      "loss": 2.9361,
+      "step": 1755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1412844036697248e-05,
+      "loss": 2.9277,
+      "step": 1756
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.1405963302752293e-05,
+      "loss": 2.861,
+      "step": 1757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.139908256880734e-05,
+      "loss": 2.8309,
+      "step": 1758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1392201834862386e-05,
+      "loss": 2.8278,
+      "step": 1759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.138532110091743e-05,
+      "loss": 2.7286,
+      "step": 1760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.137844036697248e-05,
+      "loss": 2.6938,
+      "step": 1761
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1371559633027524e-05,
+      "loss": 2.5536,
+      "step": 1762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1364678899082572e-05,
+      "loss": 2.5837,
+      "step": 1763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1357798165137616e-05,
+      "loss": 2.3951,
+      "step": 1764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.135091743119266e-05,
+      "loss": 2.4133,
+      "step": 1765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.134403669724771e-05,
+      "loss": 2.4354,
+      "step": 1766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.133715596330275e-05,
+      "loss": 2.3021,
+      "step": 1767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.13302752293578e-05,
+      "loss": 2.2406,
+      "step": 1768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1323394495412844e-05,
+      "loss": 1.9991,
+      "step": 1769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.131651376146789e-05,
+      "loss": 1.915,
+      "step": 1770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1309633027522937e-05,
+      "loss": 4.3853,
+      "step": 1771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.130275229357798e-05,
+      "loss": 4.0791,
+      "step": 1772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.1295871559633026e-05,
+      "loss": 3.8389,
+      "step": 1773
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1288990825688074e-05,
+      "loss": 3.7865,
+      "step": 1774
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.128211009174312e-05,
+      "loss": 3.5912,
+      "step": 1775
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1275229357798167e-05,
+      "loss": 3.6649,
+      "step": 1776
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1268348623853212e-05,
+      "loss": 3.4776,
+      "step": 1777
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1261467889908257e-05,
+      "loss": 3.5646,
+      "step": 1778
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1254587155963305e-05,
+      "loss": 3.5606,
+      "step": 1779
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.124770642201835e-05,
+      "loss": 3.4486,
+      "step": 1780
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1240825688073398e-05,
+      "loss": 3.4674,
+      "step": 1781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1233944954128443e-05,
+      "loss": 3.55,
+      "step": 1782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1227064220183484e-05,
+      "loss": 3.3093,
+      "step": 1783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1220183486238532e-05,
+      "loss": 3.4488,
+      "step": 1784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1213302752293577e-05,
+      "loss": 3.345,
+      "step": 1785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1206422018348622e-05,
+      "loss": 3.3616,
+      "step": 1786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.119954128440367e-05,
+      "loss": 3.3964,
+      "step": 1787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1192660550458715e-05,
+      "loss": 3.5607,
+      "step": 1788
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1185779816513763e-05,
+      "loss": 3.4116,
+      "step": 1789
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.1178899082568808e-05,
+      "loss": 3.481,
+      "step": 1790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1172018348623852e-05,
+      "loss": 3.145,
+      "step": 1791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.11651376146789e-05,
+      "loss": 3.2784,
+      "step": 1792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1158256880733945e-05,
+      "loss": 3.2396,
+      "step": 1793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1151376146788994e-05,
+      "loss": 3.2218,
+      "step": 1794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1144495412844038e-05,
+      "loss": 3.2215,
+      "step": 1795
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1137614678899083e-05,
+      "loss": 3.1001,
+      "step": 1796
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.113073394495413e-05,
+      "loss": 3.2988,
+      "step": 1797
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1123853211009176e-05,
+      "loss": 3.1483,
+      "step": 1798
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.111697247706422e-05,
+      "loss": 3.0891,
+      "step": 1799
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1110091743119266e-05,
+      "loss": 3.0918,
+      "step": 1800
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.110321100917431e-05,
+      "loss": 2.9396,
+      "step": 1801
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.109633027522936e-05,
+      "loss": 3.0674,
+      "step": 1802
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1089449541284403e-05,
+      "loss": 3.0782,
+      "step": 1803
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1082568807339448e-05,
+      "loss": 2.9479,
+      "step": 1804
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.1075688073394496e-05,
+      "loss": 2.9885,
+      "step": 1805
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.106880733944954e-05,
+      "loss": 2.8956,
+      "step": 1806
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.106192660550459e-05,
+      "loss": 2.9211,
+      "step": 1807
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1055045871559634e-05,
+      "loss": 2.8075,
+      "step": 1808
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.104816513761468e-05,
+      "loss": 2.7153,
+      "step": 1809
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1041284403669727e-05,
+      "loss": 2.8019,
+      "step": 1810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.103440366972477e-05,
+      "loss": 2.8246,
+      "step": 1811
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.102752293577982e-05,
+      "loss": 2.6477,
+      "step": 1812
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1020642201834865e-05,
+      "loss": 2.607,
+      "step": 1813
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.101376146788991e-05,
+      "loss": 2.4074,
+      "step": 1814
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1006880733944954e-05,
+      "loss": 2.4281,
+      "step": 1815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.1e-05,
+      "loss": 2.2098,
+      "step": 1816
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0993119266055044e-05,
+      "loss": 2.0472,
+      "step": 1817
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0986238532110092e-05,
+      "loss": 2.0226,
+      "step": 1818
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0979357798165137e-05,
+      "loss": 1.9457,
+      "step": 1819
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0972477064220185e-05,
+      "loss": 1.9686,
+      "step": 1820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.096559633027523e-05,
+      "loss": 4.3445,
+      "step": 1821
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.0958715596330274e-05,
+      "loss": 4.2072,
+      "step": 1822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0951834862385322e-05,
+      "loss": 3.8076,
+      "step": 1823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0944954128440367e-05,
+      "loss": 3.7612,
+      "step": 1824
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0938073394495415e-05,
+      "loss": 3.7045,
+      "step": 1825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.093119266055046e-05,
+      "loss": 3.5224,
+      "step": 1826
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0924311926605505e-05,
+      "loss": 3.6928,
+      "step": 1827
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0917431192660553e-05,
+      "loss": 3.5411,
+      "step": 1828
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0910550458715598e-05,
+      "loss": 3.6313,
+      "step": 1829
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0903669724770646e-05,
+      "loss": 3.4739,
+      "step": 1830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0896788990825687e-05,
+      "loss": 3.5698,
+      "step": 1831
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0889908256880732e-05,
+      "loss": 3.3549,
+      "step": 1832
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.088302752293578e-05,
+      "loss": 3.5835,
+      "step": 1833
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0876146788990825e-05,
+      "loss": 3.3359,
+      "step": 1834
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.086926605504587e-05,
+      "loss": 3.4236,
+      "step": 1835
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0862385321100918e-05,
+      "loss": 3.4058,
+      "step": 1836
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.0855504587155963e-05,
+      "loss": 3.3915,
+      "step": 1837
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.084862385321101e-05,
+      "loss": 3.3921,
+      "step": 1838
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0841743119266056e-05,
+      "loss": 3.3025,
+      "step": 1839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.08348623853211e-05,
+      "loss": 3.3117,
+      "step": 1840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.082798165137615e-05,
+      "loss": 3.3225,
+      "step": 1841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0821100917431193e-05,
+      "loss": 3.117,
+      "step": 1842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.081422018348624e-05,
+      "loss": 3.3825,
+      "step": 1843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0807339449541286e-05,
+      "loss": 3.1748,
+      "step": 1844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.080045871559633e-05,
+      "loss": 3.1001,
+      "step": 1845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.079357798165138e-05,
+      "loss": 3.1559,
+      "step": 1846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.078669724770642e-05,
+      "loss": 3.1812,
+      "step": 1847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0779816513761465e-05,
+      "loss": 3.227,
+      "step": 1848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0772935779816514e-05,
+      "loss": 3.0485,
+      "step": 1849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.076605504587156e-05,
+      "loss": 2.9059,
+      "step": 1850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0759174311926607e-05,
+      "loss": 2.9325,
+      "step": 1851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.075229357798165e-05,
+      "loss": 3.0682,
+      "step": 1852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0745412844036696e-05,
+      "loss": 3.0524,
+      "step": 1853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.0738532110091744e-05,
+      "loss": 2.979,
+      "step": 1854
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.073165137614679e-05,
+      "loss": 2.9501,
+      "step": 1855
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0724770642201837e-05,
+      "loss": 2.9475,
+      "step": 1856
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0717889908256882e-05,
+      "loss": 2.7012,
+      "step": 1857
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0711009174311927e-05,
+      "loss": 2.7731,
+      "step": 1858
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0704128440366975e-05,
+      "loss": 2.5024,
+      "step": 1859
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.069724770642202e-05,
+      "loss": 2.6762,
+      "step": 1860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0690366972477068e-05,
+      "loss": 2.5281,
+      "step": 1861
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0683486238532113e-05,
+      "loss": 2.5162,
+      "step": 1862
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0676605504587154e-05,
+      "loss": 2.3649,
+      "step": 1863
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0669724770642202e-05,
+      "loss": 2.5129,
+      "step": 1864
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0662844036697247e-05,
+      "loss": 2.2824,
+      "step": 1865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.065596330275229e-05,
+      "loss": 2.1896,
+      "step": 1866
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.064908256880734e-05,
+      "loss": 2.1827,
+      "step": 1867
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0642201834862385e-05,
+      "loss": 2.0944,
+      "step": 1868
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0635321100917433e-05,
+      "loss": 1.9178,
+      "step": 1869
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0628440366972478e-05,
+      "loss": 1.7555,
+      "step": 1870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.0621559633027522e-05,
+      "loss": 4.221,
+      "step": 1871
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.061467889908257e-05,
+      "loss": 3.9816,
+      "step": 1872
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0607798165137615e-05,
+      "loss": 3.877,
+      "step": 1873
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0600917431192663e-05,
+      "loss": 3.7306,
+      "step": 1874
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0594036697247708e-05,
+      "loss": 3.6222,
+      "step": 1875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0587155963302753e-05,
+      "loss": 3.5057,
+      "step": 1876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.05802752293578e-05,
+      "loss": 3.4415,
+      "step": 1877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0573394495412846e-05,
+      "loss": 3.465,
+      "step": 1878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0566513761467887e-05,
+      "loss": 3.5041,
+      "step": 1879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0559633027522935e-05,
+      "loss": 3.3332,
+      "step": 1880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.055275229357798e-05,
+      "loss": 3.484,
+      "step": 1881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.054587155963303e-05,
+      "loss": 3.2811,
+      "step": 1882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0538990825688073e-05,
+      "loss": 3.3235,
+      "step": 1883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0532110091743118e-05,
+      "loss": 3.3601,
+      "step": 1884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.0525229357798166e-05,
+      "loss": 3.4134,
+      "step": 1885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.051834862385321e-05,
+      "loss": 3.2901,
+      "step": 1886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.051146788990826e-05,
+      "loss": 3.1901,
+      "step": 1887
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0504587155963304e-05,
+      "loss": 3.3423,
+      "step": 1888
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.049770642201835e-05,
+      "loss": 3.2383,
+      "step": 1889
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0490825688073397e-05,
+      "loss": 3.438,
+      "step": 1890
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.048394495412844e-05,
+      "loss": 3.2928,
+      "step": 1891
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.047706422018349e-05,
+      "loss": 3.1044,
+      "step": 1892
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0470183486238534e-05,
+      "loss": 3.3237,
+      "step": 1893
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.046330275229358e-05,
+      "loss": 3.1234,
+      "step": 1894
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0456422018348624e-05,
+      "loss": 3.2069,
+      "step": 1895
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.044954128440367e-05,
+      "loss": 3.0226,
+      "step": 1896
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0442660550458713e-05,
+      "loss": 3.0998,
+      "step": 1897
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.043577981651376e-05,
+      "loss": 3.0967,
+      "step": 1898
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0428899082568806e-05,
+      "loss": 3.0464,
+      "step": 1899
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0422018348623855e-05,
+      "loss": 2.9729,
+      "step": 1900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.04151376146789e-05,
+      "loss": 2.9293,
+      "step": 1901
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0408256880733944e-05,
+      "loss": 2.9274,
+      "step": 1902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.0401376146788992e-05,
+      "loss": 2.9572,
+      "step": 1903
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0394495412844037e-05,
+      "loss": 2.9434,
+      "step": 1904
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0387614678899085e-05,
+      "loss": 2.862,
+      "step": 1905
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.038073394495413e-05,
+      "loss": 2.7632,
+      "step": 1906
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0373853211009175e-05,
+      "loss": 2.6748,
+      "step": 1907
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0366972477064223e-05,
+      "loss": 2.6927,
+      "step": 1908
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0360091743119268e-05,
+      "loss": 2.6441,
+      "step": 1909
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0353211009174312e-05,
+      "loss": 2.6898,
+      "step": 1910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0346330275229357e-05,
+      "loss": 2.7418,
+      "step": 1911
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0339449541284402e-05,
+      "loss": 2.5593,
+      "step": 1912
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.033256880733945e-05,
+      "loss": 2.3802,
+      "step": 1913
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0325688073394495e-05,
+      "loss": 2.365,
+      "step": 1914
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.031880733944954e-05,
+      "loss": 2.27,
+      "step": 1915
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0311926605504588e-05,
+      "loss": 2.2151,
+      "step": 1916
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0305045871559633e-05,
+      "loss": 2.0475,
+      "step": 1917
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.029816513761468e-05,
+      "loss": 2.0997,
+      "step": 1918
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.0291284403669726e-05,
+      "loss": 1.9755,
+      "step": 1919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.028440366972477e-05,
+      "loss": 1.9038,
+      "step": 1920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.027752293577982e-05,
+      "loss": 4.2254,
+      "step": 1921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0270642201834863e-05,
+      "loss": 4.0253,
+      "step": 1922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0263761467889908e-05,
+      "loss": 3.6867,
+      "step": 1923
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0256880733944956e-05,
+      "loss": 3.8363,
+      "step": 1924
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.025e-05,
+      "loss": 3.5074,
+      "step": 1925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.024311926605505e-05,
+      "loss": 3.444,
+      "step": 1926
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.023623853211009e-05,
+      "loss": 3.3458,
+      "step": 1927
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0229357798165135e-05,
+      "loss": 3.5093,
+      "step": 1928
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0222477064220183e-05,
+      "loss": 3.3069,
+      "step": 1929
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0215596330275228e-05,
+      "loss": 3.3729,
+      "step": 1930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0208715596330276e-05,
+      "loss": 3.5257,
+      "step": 1931
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.020183486238532e-05,
+      "loss": 3.4145,
+      "step": 1932
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0194954128440366e-05,
+      "loss": 3.2117,
+      "step": 1933
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.0188073394495414e-05,
+      "loss": 3.1499,
+      "step": 1934
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.018119266055046e-05,
+      "loss": 3.2662,
+      "step": 1935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0174311926605507e-05,
+      "loss": 3.3242,
+      "step": 1936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0167431192660552e-05,
+      "loss": 3.2014,
+      "step": 1937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0160550458715597e-05,
+      "loss": 3.1266,
+      "step": 1938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0153669724770645e-05,
+      "loss": 3.0914,
+      "step": 1939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.014678899082569e-05,
+      "loss": 3.2039,
+      "step": 1940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0139908256880734e-05,
+      "loss": 3.1286,
+      "step": 1941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0133027522935782e-05,
+      "loss": 3.0706,
+      "step": 1942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0126146788990824e-05,
+      "loss": 3.0664,
+      "step": 1943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0119266055045872e-05,
+      "loss": 2.8612,
+      "step": 1944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0112385321100917e-05,
+      "loss": 3.0156,
+      "step": 1945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.010550458715596e-05,
+      "loss": 3.1129,
+      "step": 1946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.009862385321101e-05,
+      "loss": 3.0549,
+      "step": 1947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0091743119266054e-05,
+      "loss": 3.1571,
+      "step": 1948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0084862385321103e-05,
+      "loss": 2.8925,
+      "step": 1949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0077981651376147e-05,
+      "loss": 2.7876,
+      "step": 1950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.0071100917431192e-05,
+      "loss": 2.9418,
+      "step": 1951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.006422018348624e-05,
+      "loss": 2.931,
+      "step": 1952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0057339449541285e-05,
+      "loss": 2.8233,
+      "step": 1953
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.005045871559633e-05,
+      "loss": 2.9627,
+      "step": 1954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0043577981651378e-05,
+      "loss": 2.6953,
+      "step": 1955
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0036697247706423e-05,
+      "loss": 2.8986,
+      "step": 1956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.002981651376147e-05,
+      "loss": 2.6618,
+      "step": 1957
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0022935779816516e-05,
+      "loss": 2.7029,
+      "step": 1958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0016055045871557e-05,
+      "loss": 2.7214,
+      "step": 1959
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.0009174311926605e-05,
+      "loss": 2.5522,
+      "step": 1960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.000229357798165e-05,
+      "loss": 2.5026,
+      "step": 1961
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9995412844036698e-05,
+      "loss": 2.5262,
+      "step": 1962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9988532110091743e-05,
+      "loss": 2.3597,
+      "step": 1963
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9981651376146788e-05,
+      "loss": 2.2282,
+      "step": 1964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9974770642201836e-05,
+      "loss": 2.1923,
+      "step": 1965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.996788990825688e-05,
+      "loss": 2.1719,
+      "step": 1966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.996100917431193e-05,
+      "loss": 2.0063,
+      "step": 1967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9954128440366974e-05,
+      "loss": 1.9643,
+      "step": 1968
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.994724770642202e-05,
+      "loss": 1.826,
+      "step": 1969
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9940366972477066e-05,
+      "loss": 1.741,
+      "step": 1970
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.993348623853211e-05,
+      "loss": 4.2385,
+      "step": 1971
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9926605504587156e-05,
+      "loss": 3.8491,
+      "step": 1972
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9919724770642204e-05,
+      "loss": 3.7065,
+      "step": 1973
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.991284403669725e-05,
+      "loss": 3.7071,
+      "step": 1974
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9905963302752294e-05,
+      "loss": 3.4183,
+      "step": 1975
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.989908256880734e-05,
+      "loss": 3.4277,
+      "step": 1976
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9892201834862383e-05,
+      "loss": 3.4475,
+      "step": 1977
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.988532110091743e-05,
+      "loss": 3.3344,
+      "step": 1978
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9878440366972476e-05,
+      "loss": 3.2213,
+      "step": 1979
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9871559633027524e-05,
+      "loss": 3.3913,
+      "step": 1980
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.986467889908257e-05,
+      "loss": 3.479,
+      "step": 1981
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9857798165137614e-05,
+      "loss": 3.4053,
+      "step": 1982
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9850917431192662e-05,
+      "loss": 3.1771,
+      "step": 1983
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9844036697247707e-05,
+      "loss": 3.4076,
+      "step": 1984
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.983715596330275e-05,
+      "loss": 3.247,
+      "step": 1985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.98302752293578e-05,
+      "loss": 3.1897,
+      "step": 1986
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9823394495412845e-05,
+      "loss": 3.2305,
+      "step": 1987
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9816513761467893e-05,
+      "loss": 3.1475,
+      "step": 1988
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9809633027522937e-05,
+      "loss": 3.1631,
+      "step": 1989
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9802752293577982e-05,
+      "loss": 3.0711,
+      "step": 1990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9795871559633027e-05,
+      "loss": 3.052,
+      "step": 1991
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9788990825688072e-05,
+      "loss": 3.0575,
+      "step": 1992
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.978211009174312e-05,
+      "loss": 3.0212,
+      "step": 1993
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9775229357798165e-05,
+      "loss": 3.2529,
+      "step": 1994
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.976834862385321e-05,
+      "loss": 2.9749,
+      "step": 1995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9761467889908258e-05,
+      "loss": 2.9995,
+      "step": 1996
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9754587155963302e-05,
+      "loss": 2.9967,
+      "step": 1997
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.974770642201835e-05,
+      "loss": 2.9247,
+      "step": 1998
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9740825688073395e-05,
+      "loss": 2.8755,
+      "step": 1999
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.973394495412844e-05,
+      "loss": 2.79,
+      "step": 2000
+    },
+    {
+      "epoch": 1.23,
+      "eval_bleu": 5.205762599108201e-14,
+      "eval_loss": 3.2045702934265137,
+      "eval_runtime": 2521.8943,
+      "eval_samples_per_second": 5.853,
+      "eval_steps_per_second": 0.732,
+      "step": 2000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9727064220183488e-05,
+      "loss": 2.8875,
+      "step": 2001
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9720183486238533e-05,
+      "loss": 2.9485,
+      "step": 2002
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9713302752293578e-05,
+      "loss": 2.6901,
+      "step": 2003
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9706422018348626e-05,
+      "loss": 2.803,
+      "step": 2004
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.969954128440367e-05,
+      "loss": 2.7995,
+      "step": 2005
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.969266055045872e-05,
+      "loss": 2.666,
+      "step": 2006
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.968577981651376e-05,
+      "loss": 2.5341,
+      "step": 2007
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9678899082568805e-05,
+      "loss": 2.6527,
+      "step": 2008
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9672018348623853e-05,
+      "loss": 2.6372,
+      "step": 2009
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9665137614678898e-05,
+      "loss": 2.6773,
+      "step": 2010
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9658256880733946e-05,
+      "loss": 2.6454,
+      "step": 2011
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.965137614678899e-05,
+      "loss": 2.5151,
+      "step": 2012
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9644495412844036e-05,
+      "loss": 2.4919,
+      "step": 2013
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9637614678899084e-05,
+      "loss": 2.3686,
+      "step": 2014
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.963073394495413e-05,
+      "loss": 2.3085,
+      "step": 2015
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9623853211009173e-05,
+      "loss": 2.2148,
+      "step": 2016
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.961697247706422e-05,
+      "loss": 2.042,
+      "step": 2017
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9610091743119266e-05,
+      "loss": 2.036,
+      "step": 2018
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9603211009174315e-05,
+      "loss": 1.891,
+      "step": 2019
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.959633027522936e-05,
+      "loss": 1.7034,
+      "step": 2020
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9589449541284404e-05,
+      "loss": 4.2759,
+      "step": 2021
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9582568807339452e-05,
+      "loss": 3.904,
+      "step": 2022
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9575688073394494e-05,
+      "loss": 3.8006,
+      "step": 2023
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9568807339449542e-05,
+      "loss": 3.4912,
+      "step": 2024
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9561926605504587e-05,
+      "loss": 3.4223,
+      "step": 2025
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.955504587155963e-05,
+      "loss": 3.4935,
+      "step": 2026
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.954816513761468e-05,
+      "loss": 3.4017,
+      "step": 2027
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9541284403669724e-05,
+      "loss": 3.3047,
+      "step": 2028
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9534403669724772e-05,
+      "loss": 3.4297,
+      "step": 2029
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9527522935779817e-05,
+      "loss": 3.3388,
+      "step": 2030
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9520642201834862e-05,
+      "loss": 3.2905,
+      "step": 2031
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.951376146788991e-05,
+      "loss": 3.2107,
+      "step": 2032
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9506880733944955e-05,
+      "loss": 3.2937,
+      "step": 2033
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.95e-05,
+      "loss": 3.059,
+      "step": 2034
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9493119266055048e-05,
+      "loss": 3.2659,
+      "step": 2035
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9486238532110093e-05,
+      "loss": 3.3604,
+      "step": 2036
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.947935779816514e-05,
+      "loss": 3.127,
+      "step": 2037
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9472477064220186e-05,
+      "loss": 3.1097,
+      "step": 2038
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9465596330275227e-05,
+      "loss": 2.9874,
+      "step": 2039
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9458715596330275e-05,
+      "loss": 2.9898,
+      "step": 2040
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.945183486238532e-05,
+      "loss": 3.0955,
+      "step": 2041
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9444954128440368e-05,
+      "loss": 3.1291,
+      "step": 2042
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9438073394495413e-05,
+      "loss": 2.9918,
+      "step": 2043
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9431192660550458e-05,
+      "loss": 2.9654,
+      "step": 2044
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9424311926605506e-05,
+      "loss": 2.959,
+      "step": 2045
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.941743119266055e-05,
+      "loss": 2.8797,
+      "step": 2046
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9410550458715595e-05,
+      "loss": 2.9048,
+      "step": 2047
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9403669724770643e-05,
+      "loss": 2.9275,
+      "step": 2048
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9396788990825688e-05,
+      "loss": 2.8946,
+      "step": 2049
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9389908256880736e-05,
+      "loss": 2.8638,
+      "step": 2050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.938302752293578e-05,
+      "loss": 2.779,
+      "step": 2051
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9376146788990826e-05,
+      "loss": 2.8118,
+      "step": 2052
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9369266055045874e-05,
+      "loss": 2.7865,
+      "step": 2053
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.936238532110092e-05,
+      "loss": 2.6685,
+      "step": 2054
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9355504587155964e-05,
+      "loss": 2.541,
+      "step": 2055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.934862385321101e-05,
+      "loss": 2.562,
+      "step": 2056
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9341743119266053e-05,
+      "loss": 2.6673,
+      "step": 2057
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.93348623853211e-05,
+      "loss": 2.6375,
+      "step": 2058
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9327981651376146e-05,
+      "loss": 2.4431,
+      "step": 2059
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9321100917431194e-05,
+      "loss": 2.5719,
+      "step": 2060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.931422018348624e-05,
+      "loss": 2.3813,
+      "step": 2061
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9307339449541284e-05,
+      "loss": 2.4129,
+      "step": 2062
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9300458715596332e-05,
+      "loss": 2.264,
+      "step": 2063
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9293577981651377e-05,
+      "loss": 2.202,
+      "step": 2064
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.928669724770642e-05,
+      "loss": 2.1973,
+      "step": 2065
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.927981651376147e-05,
+      "loss": 1.9707,
+      "step": 2066
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9272935779816514e-05,
+      "loss": 1.7715,
+      "step": 2067
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9266055045871563e-05,
+      "loss": 1.9416,
+      "step": 2068
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9259174311926607e-05,
+      "loss": 1.7655,
+      "step": 2069
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9252293577981652e-05,
+      "loss": 1.5732,
+      "step": 2070
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9245412844036697e-05,
+      "loss": 4.0865,
+      "step": 2071
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.923853211009174e-05,
+      "loss": 3.7101,
+      "step": 2072
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.923165137614679e-05,
+      "loss": 3.5888,
+      "step": 2073
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9224770642201835e-05,
+      "loss": 3.4214,
+      "step": 2074
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.921788990825688e-05,
+      "loss": 3.3289,
+      "step": 2075
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9211009174311927e-05,
+      "loss": 3.2303,
+      "step": 2076
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9204128440366972e-05,
+      "loss": 3.3854,
+      "step": 2077
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9197247706422017e-05,
+      "loss": 3.3019,
+      "step": 2078
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9190366972477065e-05,
+      "loss": 3.3903,
+      "step": 2079
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.918348623853211e-05,
+      "loss": 3.2709,
+      "step": 2080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9176605504587158e-05,
+      "loss": 3.1025,
+      "step": 2081
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9169724770642203e-05,
+      "loss": 3.3375,
+      "step": 2082
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9162844036697248e-05,
+      "loss": 3.1793,
+      "step": 2083
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9155963302752296e-05,
+      "loss": 3.2112,
+      "step": 2084
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.914908256880734e-05,
+      "loss": 3.2022,
+      "step": 2085
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.914220183486239e-05,
+      "loss": 3.0281,
+      "step": 2086
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.913532110091743e-05,
+      "loss": 2.976,
+      "step": 2087
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9128440366972475e-05,
+      "loss": 3.1226,
+      "step": 2088
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9121559633027523e-05,
+      "loss": 3.1417,
+      "step": 2089
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9114678899082568e-05,
+      "loss": 2.9836,
+      "step": 2090
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9107798165137616e-05,
+      "loss": 3.0219,
+      "step": 2091
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.910091743119266e-05,
+      "loss": 3.0381,
+      "step": 2092
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9094036697247706e-05,
+      "loss": 2.8538,
+      "step": 2093
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9087155963302754e-05,
+      "loss": 2.8651,
+      "step": 2094
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.90802752293578e-05,
+      "loss": 2.9072,
+      "step": 2095
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9073394495412843e-05,
+      "loss": 2.9567,
+      "step": 2096
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.906651376146789e-05,
+      "loss": 3.1007,
+      "step": 2097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9059633027522936e-05,
+      "loss": 2.82,
+      "step": 2098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9052752293577984e-05,
+      "loss": 2.7403,
+      "step": 2099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.904587155963303e-05,
+      "loss": 2.7928,
+      "step": 2100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9038990825688074e-05,
+      "loss": 2.7215,
+      "step": 2101
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9032110091743122e-05,
+      "loss": 2.6853,
+      "step": 2102
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9025229357798163e-05,
+      "loss": 2.6453,
+      "step": 2103
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.901834862385321e-05,
+      "loss": 2.784,
+      "step": 2104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9011467889908256e-05,
+      "loss": 2.523,
+      "step": 2105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.90045871559633e-05,
+      "loss": 2.7375,
+      "step": 2106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.899770642201835e-05,
+      "loss": 2.6481,
+      "step": 2107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8990825688073394e-05,
+      "loss": 2.5335,
+      "step": 2108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.898394495412844e-05,
+      "loss": 2.6238,
+      "step": 2109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8977064220183487e-05,
+      "loss": 2.4203,
+      "step": 2110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8970183486238532e-05,
+      "loss": 2.3357,
+      "step": 2111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.896330275229358e-05,
+      "loss": 2.4566,
+      "step": 2112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.8956422018348625e-05,
+      "loss": 2.3007,
+      "step": 2113
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.894954128440367e-05,
+      "loss": 2.2106,
+      "step": 2114
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8942660550458718e-05,
+      "loss": 2.0403,
+      "step": 2115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8935779816513762e-05,
+      "loss": 2.0395,
+      "step": 2116
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.892889908256881e-05,
+      "loss": 1.9583,
+      "step": 2117
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8922018348623855e-05,
+      "loss": 1.8823,
+      "step": 2118
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8915137614678897e-05,
+      "loss": 1.7667,
+      "step": 2119
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8908256880733945e-05,
+      "loss": 1.6001,
+      "step": 2120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.890137614678899e-05,
+      "loss": 4.0476,
+      "step": 2121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8894495412844038e-05,
+      "loss": 3.6031,
+      "step": 2122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8887614678899083e-05,
+      "loss": 3.6476,
+      "step": 2123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8880733944954127e-05,
+      "loss": 3.394,
+      "step": 2124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8873853211009176e-05,
+      "loss": 3.207,
+      "step": 2125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.886697247706422e-05,
+      "loss": 3.138,
+      "step": 2126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8860091743119265e-05,
+      "loss": 3.2913,
+      "step": 2127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8853211009174313e-05,
+      "loss": 3.0331,
+      "step": 2128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8846330275229358e-05,
+      "loss": 3.2702,
+      "step": 2129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.8839449541284406e-05,
+      "loss": 3.2543,
+      "step": 2130
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.883256880733945e-05,
+      "loss": 3.1269,
+      "step": 2131
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8825688073394496e-05,
+      "loss": 3.0883,
+      "step": 2132
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8818807339449544e-05,
+      "loss": 3.0264,
+      "step": 2133
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.881192660550459e-05,
+      "loss": 3.1884,
+      "step": 2134
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8805045871559633e-05,
+      "loss": 3.1361,
+      "step": 2135
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8798165137614678e-05,
+      "loss": 3.0321,
+      "step": 2136
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8791284403669723e-05,
+      "loss": 3.0146,
+      "step": 2137
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.878440366972477e-05,
+      "loss": 3.0114,
+      "step": 2138
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8777522935779816e-05,
+      "loss": 2.9579,
+      "step": 2139
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.877064220183486e-05,
+      "loss": 2.9197,
+      "step": 2140
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.876376146788991e-05,
+      "loss": 2.9739,
+      "step": 2141
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8756880733944954e-05,
+      "loss": 2.8959,
+      "step": 2142
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 3.0636,
+      "step": 2143
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8743119266055047e-05,
+      "loss": 3.0544,
+      "step": 2144
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.873623853211009e-05,
+      "loss": 2.7991,
+      "step": 2145
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.872935779816514e-05,
+      "loss": 2.7726,
+      "step": 2146
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8722477064220184e-05,
+      "loss": 2.9793,
+      "step": 2147
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8715596330275232e-05,
+      "loss": 2.8213,
+      "step": 2148
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8708715596330277e-05,
+      "loss": 2.8906,
+      "step": 2149
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8701834862385322e-05,
+      "loss": 2.8705,
+      "step": 2150
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.869495412844037e-05,
+      "loss": 2.7925,
+      "step": 2151
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.868807339449541e-05,
+      "loss": 2.7316,
+      "step": 2152
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.868119266055046e-05,
+      "loss": 2.6434,
+      "step": 2153
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8674311926605504e-05,
+      "loss": 2.6698,
+      "step": 2154
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.866743119266055e-05,
+      "loss": 2.4718,
+      "step": 2155
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8660550458715597e-05,
+      "loss": 2.6476,
+      "step": 2156
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8653669724770642e-05,
+      "loss": 2.5058,
+      "step": 2157
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8646788990825687e-05,
+      "loss": 2.3585,
+      "step": 2158
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8639908256880735e-05,
+      "loss": 2.3191,
+      "step": 2159
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.863302752293578e-05,
+      "loss": 2.3016,
+      "step": 2160
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8626146788990828e-05,
+      "loss": 2.3778,
+      "step": 2161
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8619266055045873e-05,
+      "loss": 2.3643,
+      "step": 2162
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8612385321100918e-05,
+      "loss": 2.207,
+      "step": 2163
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8605504587155966e-05,
+      "loss": 2.1567,
+      "step": 2164
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.859862385321101e-05,
+      "loss": 2.2012,
+      "step": 2165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.859174311926606e-05,
+      "loss": 2.0841,
+      "step": 2166
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8584862385321103e-05,
+      "loss": 1.8812,
+      "step": 2167
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8577981651376145e-05,
+      "loss": 1.7372,
+      "step": 2168
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8571100917431193e-05,
+      "loss": 1.7146,
+      "step": 2169
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8564220183486238e-05,
+      "loss": 1.6167,
+      "step": 2170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8557339449541282e-05,
+      "loss": 3.9565,
+      "step": 2171
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.855045871559633e-05,
+      "loss": 3.6182,
+      "step": 2172
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8543577981651375e-05,
+      "loss": 3.507,
+      "step": 2173
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8536697247706424e-05,
+      "loss": 3.3837,
+      "step": 2174
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.852981651376147e-05,
+      "loss": 3.3371,
+      "step": 2175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8522935779816513e-05,
+      "loss": 3.2526,
+      "step": 2176
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.851605504587156e-05,
+      "loss": 3.0577,
+      "step": 2177
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8509174311926606e-05,
+      "loss": 3.2311,
+      "step": 2178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8502293577981654e-05,
+      "loss": 3.1893,
+      "step": 2179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.84954128440367e-05,
+      "loss": 3.2017,
+      "step": 2180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8488532110091744e-05,
+      "loss": 3.1053,
+      "step": 2181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8481651376146792e-05,
+      "loss": 3.0092,
+      "step": 2182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8474770642201837e-05,
+      "loss": 3.0627,
+      "step": 2183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8467889908256878e-05,
+      "loss": 3.1523,
+      "step": 2184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8461009174311926e-05,
+      "loss": 2.9013,
+      "step": 2185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.845412844036697e-05,
+      "loss": 3.1466,
+      "step": 2186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.844724770642202e-05,
+      "loss": 3.0549,
+      "step": 2187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8440366972477064e-05,
+      "loss": 2.8666,
+      "step": 2188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.843348623853211e-05,
+      "loss": 3.0291,
+      "step": 2189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8426605504587157e-05,
+      "loss": 2.9408,
+      "step": 2190
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.84197247706422e-05,
+      "loss": 2.8882,
+      "step": 2191
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.841284403669725e-05,
+      "loss": 2.8491,
+      "step": 2192
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8405963302752295e-05,
+      "loss": 2.9867,
+      "step": 2193
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.839908256880734e-05,
+      "loss": 2.846,
+      "step": 2194
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8392201834862387e-05,
+      "loss": 2.7523,
+      "step": 2195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8385321100917432e-05,
+      "loss": 2.7874,
+      "step": 2196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.837844036697248e-05,
+      "loss": 2.7847,
+      "step": 2197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8371559633027525e-05,
+      "loss": 2.595,
+      "step": 2198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.836467889908257e-05,
+      "loss": 2.6692,
+      "step": 2199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8357798165137615e-05,
+      "loss": 2.7014,
+      "step": 2200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.835091743119266e-05,
+      "loss": 2.7399,
+      "step": 2201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8344036697247704e-05,
+      "loss": 2.5325,
+      "step": 2202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8337155963302752e-05,
+      "loss": 2.593,
+      "step": 2203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8330275229357797e-05,
+      "loss": 2.5653,
+      "step": 2204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8323394495412845e-05,
+      "loss": 2.4783,
+      "step": 2205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.831651376146789e-05,
+      "loss": 2.5416,
+      "step": 2206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8309633027522935e-05,
+      "loss": 2.4414,
+      "step": 2207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8302752293577983e-05,
+      "loss": 2.4409,
+      "step": 2208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8295871559633028e-05,
+      "loss": 2.4249,
+      "step": 2209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.8288990825688076e-05,
+      "loss": 2.4046,
+      "step": 2210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.828211009174312e-05,
+      "loss": 2.4241,
+      "step": 2211
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8275229357798166e-05,
+      "loss": 2.4416,
+      "step": 2212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8268348623853214e-05,
+      "loss": 2.1854,
+      "step": 2213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.826146788990826e-05,
+      "loss": 2.0856,
+      "step": 2214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8254587155963303e-05,
+      "loss": 2.2396,
+      "step": 2215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8247706422018348e-05,
+      "loss": 2.0913,
+      "step": 2216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8240825688073393e-05,
+      "loss": 1.9313,
+      "step": 2217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.823394495412844e-05,
+      "loss": 1.6869,
+      "step": 2218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8227064220183486e-05,
+      "loss": 1.514,
+      "step": 2219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.822018348623853e-05,
+      "loss": 1.7076,
+      "step": 2220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.821330275229358e-05,
+      "loss": 4.0691,
+      "step": 2221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8206422018348623e-05,
+      "loss": 3.5557,
+      "step": 2222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.819954128440367e-05,
+      "loss": 3.6104,
+      "step": 2223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8192660550458716e-05,
+      "loss": 3.4079,
+      "step": 2224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.818577981651376e-05,
+      "loss": 3.1731,
+      "step": 2225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.817889908256881e-05,
+      "loss": 3.0763,
+      "step": 2226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8172018348623854e-05,
+      "loss": 3.0102,
+      "step": 2227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8165137614678902e-05,
+      "loss": 3.1165,
+      "step": 2228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8158256880733947e-05,
+      "loss": 3.067,
+      "step": 2229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8151376146788992e-05,
+      "loss": 3.0956,
+      "step": 2230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.814449541284404e-05,
+      "loss": 3.1999,
+      "step": 2231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.813761467889908e-05,
+      "loss": 2.8917,
+      "step": 2232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8130733944954126e-05,
+      "loss": 3.0519,
+      "step": 2233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8123853211009174e-05,
+      "loss": 3.0095,
+      "step": 2234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.811697247706422e-05,
+      "loss": 2.9691,
+      "step": 2235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8110091743119267e-05,
+      "loss": 2.9347,
+      "step": 2236
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8103211009174312e-05,
+      "loss": 2.9571,
+      "step": 2237
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8096330275229357e-05,
+      "loss": 2.9372,
+      "step": 2238
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8089449541284405e-05,
+      "loss": 2.9027,
+      "step": 2239
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.808256880733945e-05,
+      "loss": 3.0103,
+      "step": 2240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8075688073394498e-05,
+      "loss": 2.9592,
+      "step": 2241
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8068807339449543e-05,
+      "loss": 2.8663,
+      "step": 2242
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.8061926605504587e-05,
+      "loss": 2.7978,
+      "step": 2243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8055045871559636e-05,
+      "loss": 2.7948,
+      "step": 2244
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.804816513761468e-05,
+      "loss": 2.6917,
+      "step": 2245
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8041284403669725e-05,
+      "loss": 2.6407,
+      "step": 2246
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8034403669724773e-05,
+      "loss": 2.7942,
+      "step": 2247
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8027522935779815e-05,
+      "loss": 2.7887,
+      "step": 2248
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8020642201834863e-05,
+      "loss": 2.8262,
+      "step": 2249
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8013761467889908e-05,
+      "loss": 2.5444,
+      "step": 2250
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8006880733944952e-05,
+      "loss": 2.5951,
+      "step": 2251
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8e-05,
+      "loss": 2.5413,
+      "step": 2252
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7993119266055045e-05,
+      "loss": 2.4398,
+      "step": 2253
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7986238532110093e-05,
+      "loss": 2.4925,
+      "step": 2254
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7979357798165138e-05,
+      "loss": 2.423,
+      "step": 2255
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7972477064220183e-05,
+      "loss": 2.551,
+      "step": 2256
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.796559633027523e-05,
+      "loss": 2.4847,
+      "step": 2257
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7958715596330276e-05,
+      "loss": 2.4481,
+      "step": 2258
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.7951834862385324e-05,
+      "loss": 2.4132,
+      "step": 2259
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.794495412844037e-05,
+      "loss": 2.4144,
+      "step": 2260
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7938073394495414e-05,
+      "loss": 2.2489,
+      "step": 2261
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7931192660550462e-05,
+      "loss": 2.1427,
+      "step": 2262
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7924311926605506e-05,
+      "loss": 2.1172,
+      "step": 2263
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7917431192660548e-05,
+      "loss": 2.0935,
+      "step": 2264
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7910550458715596e-05,
+      "loss": 2.1327,
+      "step": 2265
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.790366972477064e-05,
+      "loss": 2.0425,
+      "step": 2266
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.789678899082569e-05,
+      "loss": 1.8651,
+      "step": 2267
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7889908256880734e-05,
+      "loss": 1.8843,
+      "step": 2268
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.788302752293578e-05,
+      "loss": 1.7389,
+      "step": 2269
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7876146788990827e-05,
+      "loss": 1.6538,
+      "step": 2270
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.786926605504587e-05,
+      "loss": 3.6895,
+      "step": 2271
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.786238532110092e-05,
+      "loss": 3.6053,
+      "step": 2272
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7855504587155964e-05,
+      "loss": 3.5534,
+      "step": 2273
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.784862385321101e-05,
+      "loss": 3.3142,
+      "step": 2274
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7841743119266057e-05,
+      "loss": 3.1558,
+      "step": 2275
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.7834862385321102e-05,
+      "loss": 2.9261,
+      "step": 2276
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7827981651376147e-05,
+      "loss": 3.1063,
+      "step": 2277
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7821100917431195e-05,
+      "loss": 3.1501,
+      "step": 2278
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.781422018348624e-05,
+      "loss": 3.1805,
+      "step": 2279
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7807339449541285e-05,
+      "loss": 2.9488,
+      "step": 2280
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.780045871559633e-05,
+      "loss": 2.9808,
+      "step": 2281
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7793577981651374e-05,
+      "loss": 2.954,
+      "step": 2282
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7786697247706422e-05,
+      "loss": 2.7545,
+      "step": 2283
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7779816513761467e-05,
+      "loss": 2.9088,
+      "step": 2284
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7772935779816515e-05,
+      "loss": 2.884,
+      "step": 2285
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.776605504587156e-05,
+      "loss": 2.7847,
+      "step": 2286
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7759174311926605e-05,
+      "loss": 2.9458,
+      "step": 2287
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7752293577981653e-05,
+      "loss": 2.9267,
+      "step": 2288
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7745412844036698e-05,
+      "loss": 2.7675,
+      "step": 2289
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7738532110091746e-05,
+      "loss": 2.7124,
+      "step": 2290
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.773165137614679e-05,
+      "loss": 2.7515,
+      "step": 2291
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.7724770642201835e-05,
+      "loss": 2.8148,
+      "step": 2292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7717889908256884e-05,
+      "loss": 2.936,
+      "step": 2293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7711009174311928e-05,
+      "loss": 2.8541,
+      "step": 2294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7704128440366973e-05,
+      "loss": 2.8093,
+      "step": 2295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7697247706422018e-05,
+      "loss": 2.683,
+      "step": 2296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7690366972477063e-05,
+      "loss": 2.6727,
+      "step": 2297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.768348623853211e-05,
+      "loss": 2.5834,
+      "step": 2298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7676605504587156e-05,
+      "loss": 2.7381,
+      "step": 2299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.76697247706422e-05,
+      "loss": 2.7015,
+      "step": 2300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.766284403669725e-05,
+      "loss": 2.583,
+      "step": 2301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7655963302752293e-05,
+      "loss": 2.7104,
+      "step": 2302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.764908256880734e-05,
+      "loss": 2.4593,
+      "step": 2303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7642201834862386e-05,
+      "loss": 2.4972,
+      "step": 2304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.763532110091743e-05,
+      "loss": 2.5114,
+      "step": 2305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.762844036697248e-05,
+      "loss": 2.517,
+      "step": 2306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.7621559633027524e-05,
+      "loss": 2.4922,
+      "step": 2307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.761467889908257e-05,
+      "loss": 2.2675,
+      "step": 2308
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7607798165137617e-05,
+      "loss": 2.4809,
+      "step": 2309
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.760091743119266e-05,
+      "loss": 2.2994,
+      "step": 2310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.759403669724771e-05,
+      "loss": 2.11,
+      "step": 2311
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.758715596330275e-05,
+      "loss": 2.1587,
+      "step": 2312
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7580275229357796e-05,
+      "loss": 2.2199,
+      "step": 2313
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7573394495412844e-05,
+      "loss": 2.1393,
+      "step": 2314
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.756651376146789e-05,
+      "loss": 1.9868,
+      "step": 2315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7559633027522937e-05,
+      "loss": 1.9642,
+      "step": 2316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7552752293577982e-05,
+      "loss": 1.9327,
+      "step": 2317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7545871559633027e-05,
+      "loss": 1.7893,
+      "step": 2318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7538990825688075e-05,
+      "loss": 1.6797,
+      "step": 2319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.753211009174312e-05,
+      "loss": 1.5419,
+      "step": 2320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7525229357798164e-05,
+      "loss": 3.7387,
+      "step": 2321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7518348623853212e-05,
+      "loss": 3.5873,
+      "step": 2322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7511467889908257e-05,
+      "loss": 3.5164,
+      "step": 2323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.7504587155963305e-05,
+      "loss": 3.2564,
+      "step": 2324
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.749770642201835e-05,
+      "loss": 3.215,
+      "step": 2325
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7490825688073395e-05,
+      "loss": 3.124,
+      "step": 2326
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7483944954128443e-05,
+      "loss": 3.0672,
+      "step": 2327
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7477064220183484e-05,
+      "loss": 3.1956,
+      "step": 2328
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7470183486238533e-05,
+      "loss": 3.0054,
+      "step": 2329
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7463302752293577e-05,
+      "loss": 3.0312,
+      "step": 2330
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7456422018348622e-05,
+      "loss": 2.9069,
+      "step": 2331
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.744954128440367e-05,
+      "loss": 3.0448,
+      "step": 2332
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7442660550458715e-05,
+      "loss": 2.8179,
+      "step": 2333
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7435779816513763e-05,
+      "loss": 2.7749,
+      "step": 2334
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7428899082568808e-05,
+      "loss": 2.9422,
+      "step": 2335
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7422018348623853e-05,
+      "loss": 2.8516,
+      "step": 2336
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.74151376146789e-05,
+      "loss": 2.8028,
+      "step": 2337
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.7408256880733946e-05,
+      "loss": 2.8338,
+      "step": 2338
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.740137614678899e-05,
+      "loss": 2.8893,
+      "step": 2339
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.739449541284404e-05,
+      "loss": 2.7432,
+      "step": 2340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7387614678899083e-05,
+      "loss": 2.835,
+      "step": 2341
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.738073394495413e-05,
+      "loss": 2.8679,
+      "step": 2342
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7373853211009176e-05,
+      "loss": 2.7043,
+      "step": 2343
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7366972477064218e-05,
+      "loss": 2.8013,
+      "step": 2344
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7360091743119266e-05,
+      "loss": 2.6812,
+      "step": 2345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.735321100917431e-05,
+      "loss": 2.6025,
+      "step": 2346
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.734633027522936e-05,
+      "loss": 2.5941,
+      "step": 2347
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7339449541284404e-05,
+      "loss": 2.6523,
+      "step": 2348
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.733256880733945e-05,
+      "loss": 2.6599,
+      "step": 2349
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7325688073394497e-05,
+      "loss": 2.5956,
+      "step": 2350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.731880733944954e-05,
+      "loss": 2.6013,
+      "step": 2351
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7311926605504586e-05,
+      "loss": 2.5559,
+      "step": 2352
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7305045871559634e-05,
+      "loss": 2.5989,
+      "step": 2353
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.729816513761468e-05,
+      "loss": 2.417,
+      "step": 2354
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7291284403669727e-05,
+      "loss": 2.3346,
+      "step": 2355
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7284403669724772e-05,
+      "loss": 2.3204,
+      "step": 2356
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.7277522935779817e-05,
+      "loss": 2.3709,
+      "step": 2357
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7270642201834865e-05,
+      "loss": 2.2367,
+      "step": 2358
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.726376146788991e-05,
+      "loss": 2.366,
+      "step": 2359
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7256880733944954e-05,
+      "loss": 2.323,
+      "step": 2360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.725e-05,
+      "loss": 2.2969,
+      "step": 2361
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7243119266055044e-05,
+      "loss": 2.1768,
+      "step": 2362
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7236238532110092e-05,
+      "loss": 2.2035,
+      "step": 2363
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7229357798165137e-05,
+      "loss": 2.1329,
+      "step": 2364
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7222477064220185e-05,
+      "loss": 1.9364,
+      "step": 2365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.721559633027523e-05,
+      "loss": 1.9164,
+      "step": 2366
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7208715596330275e-05,
+      "loss": 1.9278,
+      "step": 2367
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7201834862385323e-05,
+      "loss": 1.7599,
+      "step": 2368
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7194954128440368e-05,
+      "loss": 1.8271,
+      "step": 2369
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7188073394495412e-05,
+      "loss": 1.4309,
+      "step": 2370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.718119266055046e-05,
+      "loss": 3.6143,
+      "step": 2371
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7174311926605505e-05,
+      "loss": 3.5081,
+      "step": 2372
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.7167431192660553e-05,
+      "loss": 3.2772,
+      "step": 2373
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7160550458715598e-05,
+      "loss": 3.1653,
+      "step": 2374
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7153669724770643e-05,
+      "loss": 3.1121,
+      "step": 2375
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7146788990825688e-05,
+      "loss": 2.9467,
+      "step": 2376
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7139908256880732e-05,
+      "loss": 2.9783,
+      "step": 2377
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.713302752293578e-05,
+      "loss": 3.0164,
+      "step": 2378
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7126146788990825e-05,
+      "loss": 2.8725,
+      "step": 2379
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.711926605504587e-05,
+      "loss": 2.7697,
+      "step": 2380
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.711238532110092e-05,
+      "loss": 3.0274,
+      "step": 2381
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7105504587155963e-05,
+      "loss": 2.9482,
+      "step": 2382
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7098623853211008e-05,
+      "loss": 2.8594,
+      "step": 2383
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7091743119266056e-05,
+      "loss": 2.8141,
+      "step": 2384
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.70848623853211e-05,
+      "loss": 3.0067,
+      "step": 2385
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.707798165137615e-05,
+      "loss": 2.8851,
+      "step": 2386
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7071100917431194e-05,
+      "loss": 2.8348,
+      "step": 2387
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.706422018348624e-05,
+      "loss": 2.7786,
+      "step": 2388
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.7057339449541287e-05,
+      "loss": 2.7945,
+      "step": 2389
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.705045871559633e-05,
+      "loss": 2.7949,
+      "step": 2390
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.704357798165138e-05,
+      "loss": 2.8667,
+      "step": 2391
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.703669724770642e-05,
+      "loss": 2.7218,
+      "step": 2392
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7029816513761466e-05,
+      "loss": 2.6663,
+      "step": 2393
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7022935779816514e-05,
+      "loss": 2.5554,
+      "step": 2394
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.701605504587156e-05,
+      "loss": 2.6293,
+      "step": 2395
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.7009174311926607e-05,
+      "loss": 2.6471,
+      "step": 2396
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.700229357798165e-05,
+      "loss": 2.5977,
+      "step": 2397
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6995412844036696e-05,
+      "loss": 2.6609,
+      "step": 2398
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6988532110091745e-05,
+      "loss": 2.4733,
+      "step": 2399
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.698165137614679e-05,
+      "loss": 2.5571,
+      "step": 2400
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6974770642201834e-05,
+      "loss": 2.5636,
+      "step": 2401
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6967889908256882e-05,
+      "loss": 2.4898,
+      "step": 2402
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6961009174311927e-05,
+      "loss": 2.5082,
+      "step": 2403
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6954128440366975e-05,
+      "loss": 2.5251,
+      "step": 2404
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.694724770642202e-05,
+      "loss": 2.4161,
+      "step": 2405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6940366972477065e-05,
+      "loss": 2.3068,
+      "step": 2406
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6933486238532113e-05,
+      "loss": 2.3008,
+      "step": 2407
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6926605504587154e-05,
+      "loss": 2.2989,
+      "step": 2408
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6919724770642202e-05,
+      "loss": 2.3236,
+      "step": 2409
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6912844036697247e-05,
+      "loss": 2.2641,
+      "step": 2410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6905963302752292e-05,
+      "loss": 2.1856,
+      "step": 2411
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.689908256880734e-05,
+      "loss": 2.1537,
+      "step": 2412
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6892201834862385e-05,
+      "loss": 2.0756,
+      "step": 2413
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.688532110091743e-05,
+      "loss": 1.9763,
+      "step": 2414
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6878440366972478e-05,
+      "loss": 1.8948,
+      "step": 2415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6871559633027523e-05,
+      "loss": 1.9951,
+      "step": 2416
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.686467889908257e-05,
+      "loss": 1.8476,
+      "step": 2417
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6857798165137616e-05,
+      "loss": 1.8206,
+      "step": 2418
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.685091743119266e-05,
+      "loss": 1.6405,
+      "step": 2419
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.684403669724771e-05,
+      "loss": 1.499,
+      "step": 2420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6837155963302753e-05,
+      "loss": 3.6689,
+      "step": 2421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.68302752293578e-05,
+      "loss": 3.525,
+      "step": 2422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6823394495412846e-05,
+      "loss": 3.2471,
+      "step": 2423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6816513761467888e-05,
+      "loss": 3.1919,
+      "step": 2424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6809633027522936e-05,
+      "loss": 3.1353,
+      "step": 2425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.680275229357798e-05,
+      "loss": 2.9964,
+      "step": 2426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.679587155963303e-05,
+      "loss": 2.966,
+      "step": 2427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6788990825688073e-05,
+      "loss": 2.9392,
+      "step": 2428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6782110091743118e-05,
+      "loss": 2.9599,
+      "step": 2429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6775229357798166e-05,
+      "loss": 2.8625,
+      "step": 2430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.676834862385321e-05,
+      "loss": 2.9558,
+      "step": 2431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6761467889908256e-05,
+      "loss": 2.7767,
+      "step": 2432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6754587155963304e-05,
+      "loss": 2.7589,
+      "step": 2433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.674770642201835e-05,
+      "loss": 2.7847,
+      "step": 2434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6740825688073397e-05,
+      "loss": 2.768,
+      "step": 2435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6733944954128442e-05,
+      "loss": 2.9217,
+      "step": 2436
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6727064220183487e-05,
+      "loss": 2.8286,
+      "step": 2437
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6720183486238535e-05,
+      "loss": 2.5518,
+      "step": 2438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.671330275229358e-05,
+      "loss": 2.6689,
+      "step": 2439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6706422018348624e-05,
+      "loss": 2.6661,
+      "step": 2440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.669954128440367e-05,
+      "loss": 2.728,
+      "step": 2441
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6692660550458714e-05,
+      "loss": 2.8434,
+      "step": 2442
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6685779816513762e-05,
+      "loss": 2.6161,
+      "step": 2443
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6678899082568807e-05,
+      "loss": 2.7382,
+      "step": 2444
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.667201834862385e-05,
+      "loss": 2.6457,
+      "step": 2445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.66651376146789e-05,
+      "loss": 2.7219,
+      "step": 2446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6658256880733944e-05,
+      "loss": 2.7301,
+      "step": 2447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6651376146788993e-05,
+      "loss": 2.4807,
+      "step": 2448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6644495412844037e-05,
+      "loss": 2.6805,
+      "step": 2449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6637614678899082e-05,
+      "loss": 2.4713,
+      "step": 2450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.663073394495413e-05,
+      "loss": 2.4091,
+      "step": 2451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6623853211009175e-05,
+      "loss": 2.3903,
+      "step": 2452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6616972477064223e-05,
+      "loss": 2.3278,
+      "step": 2453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6610091743119268e-05,
+      "loss": 2.3429,
+      "step": 2454
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6603211009174313e-05,
+      "loss": 2.6918,
+      "step": 2455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6596330275229358e-05,
+      "loss": 2.4331,
+      "step": 2456
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6589449541284402e-05,
+      "loss": 2.4159,
+      "step": 2457
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.658256880733945e-05,
+      "loss": 2.1994,
+      "step": 2458
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6575688073394495e-05,
+      "loss": 2.3323,
+      "step": 2459
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.656880733944954e-05,
+      "loss": 2.1256,
+      "step": 2460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6561926605504588e-05,
+      "loss": 2.1702,
+      "step": 2461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6555045871559633e-05,
+      "loss": 2.1323,
+      "step": 2462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6548165137614678e-05,
+      "loss": 1.9149,
+      "step": 2463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6541284403669726e-05,
+      "loss": 1.9034,
+      "step": 2464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.653440366972477e-05,
+      "loss": 1.9079,
+      "step": 2465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.652752293577982e-05,
+      "loss": 1.8296,
+      "step": 2466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6520642201834864e-05,
+      "loss": 1.7222,
+      "step": 2467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.651376146788991e-05,
+      "loss": 1.6675,
+      "step": 2468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6506880733944956e-05,
+      "loss": 1.6972,
+      "step": 2469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.65e-05,
+      "loss": 1.6516,
+      "step": 2470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.649311926605505e-05,
+      "loss": 3.6805,
+      "step": 2471
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.648623853211009e-05,
+      "loss": 3.4384,
+      "step": 2472
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6479357798165136e-05,
+      "loss": 3.2486,
+      "step": 2473
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6472477064220184e-05,
+      "loss": 3.0171,
+      "step": 2474
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.646559633027523e-05,
+      "loss": 2.9203,
+      "step": 2475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6458715596330273e-05,
+      "loss": 2.9607,
+      "step": 2476
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.645183486238532e-05,
+      "loss": 2.8038,
+      "step": 2477
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6444954128440366e-05,
+      "loss": 2.7537,
+      "step": 2478
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6438073394495414e-05,
+      "loss": 2.958,
+      "step": 2479
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.643119266055046e-05,
+      "loss": 2.7642,
+      "step": 2480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6424311926605504e-05,
+      "loss": 2.7598,
+      "step": 2481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6417431192660552e-05,
+      "loss": 3.0984,
+      "step": 2482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6410550458715597e-05,
+      "loss": 2.7864,
+      "step": 2483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6403669724770645e-05,
+      "loss": 2.7807,
+      "step": 2484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.639678899082569e-05,
+      "loss": 2.6692,
+      "step": 2485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6389908256880735e-05,
+      "loss": 2.7546,
+      "step": 2486
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6383027522935783e-05,
+      "loss": 2.6941,
+      "step": 2487
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6376146788990824e-05,
+      "loss": 2.7656,
+      "step": 2488
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6369266055045872e-05,
+      "loss": 2.7346,
+      "step": 2489
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6362385321100917e-05,
+      "loss": 2.7349,
+      "step": 2490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6355504587155962e-05,
+      "loss": 2.6959,
+      "step": 2491
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.634862385321101e-05,
+      "loss": 2.6463,
+      "step": 2492
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6341743119266055e-05,
+      "loss": 2.6792,
+      "step": 2493
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.63348623853211e-05,
+      "loss": 2.6362,
+      "step": 2494
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6327981651376148e-05,
+      "loss": 2.6544,
+      "step": 2495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6321100917431192e-05,
+      "loss": 2.5107,
+      "step": 2496
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.631422018348624e-05,
+      "loss": 2.538,
+      "step": 2497
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6307339449541285e-05,
+      "loss": 2.4926,
+      "step": 2498
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.630045871559633e-05,
+      "loss": 2.4483,
+      "step": 2499
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6293577981651378e-05,
+      "loss": 2.569,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "eval_bleu": 1.448718755598576e-13,
+      "eval_loss": 2.6812455654144287,
+      "eval_runtime": 2547.7793,
+      "eval_samples_per_second": 5.793,
+      "eval_steps_per_second": 0.724,
+      "step": 2500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6286697247706423e-05,
+      "loss": 2.4748,
+      "step": 2501
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.627981651376147e-05,
+      "loss": 2.4199,
+      "step": 2502
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6272935779816516e-05,
+      "loss": 2.3294,
+      "step": 2503
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6266055045871557e-05,
+      "loss": 2.4431,
+      "step": 2504
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6259174311926606e-05,
+      "loss": 2.4163,
+      "step": 2505
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.625229357798165e-05,
+      "loss": 2.2569,
+      "step": 2506
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6245412844036695e-05,
+      "loss": 2.2772,
+      "step": 2507
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6238532110091743e-05,
+      "loss": 2.3977,
+      "step": 2508
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6231651376146788e-05,
+      "loss": 2.2201,
+      "step": 2509
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6224770642201836e-05,
+      "loss": 2.2794,
+      "step": 2510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.621788990825688e-05,
+      "loss": 2.2358,
+      "step": 2511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6211009174311926e-05,
+      "loss": 2.1152,
+      "step": 2512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6204128440366974e-05,
+      "loss": 2.1671,
+      "step": 2513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.619724770642202e-05,
+      "loss": 1.8169,
+      "step": 2514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6190366972477067e-05,
+      "loss": 2.0514,
+      "step": 2515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.618348623853211e-05,
+      "loss": 1.9191,
+      "step": 2516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6176605504587156e-05,
+      "loss": 1.8616,
+      "step": 2517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6169724770642205e-05,
+      "loss": 1.682,
+      "step": 2518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.616284403669725e-05,
+      "loss": 1.6308,
+      "step": 2519
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6155963302752294e-05,
+      "loss": 1.4481,
+      "step": 2520
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.614908256880734e-05,
+      "loss": 3.5734,
+      "step": 2521
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6142201834862384e-05,
+      "loss": 3.2565,
+      "step": 2522
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6135321100917432e-05,
+      "loss": 3.1813,
+      "step": 2523
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6128440366972477e-05,
+      "loss": 3.0011,
+      "step": 2524
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.612155963302752e-05,
+      "loss": 3.0285,
+      "step": 2525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.611467889908257e-05,
+      "loss": 2.7496,
+      "step": 2526
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6107798165137614e-05,
+      "loss": 2.994,
+      "step": 2527
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6100917431192662e-05,
+      "loss": 2.8791,
+      "step": 2528
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6094036697247707e-05,
+      "loss": 2.874,
+      "step": 2529
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6087155963302752e-05,
+      "loss": 2.6926,
+      "step": 2530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.60802752293578e-05,
+      "loss": 2.8209,
+      "step": 2531
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6073394495412845e-05,
+      "loss": 2.7699,
+      "step": 2532
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6066513761467893e-05,
+      "loss": 2.757,
+      "step": 2533
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6059633027522938e-05,
+      "loss": 2.7453,
+      "step": 2534
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6052752293577983e-05,
+      "loss": 2.8405,
+      "step": 2535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6045871559633027e-05,
+      "loss": 2.6875,
+      "step": 2536
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6038990825688072e-05,
+      "loss": 2.8003,
+      "step": 2537
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6032110091743117e-05,
+      "loss": 2.7266,
+      "step": 2538
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6025229357798165e-05,
+      "loss": 2.6127,
+      "step": 2539
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.601834862385321e-05,
+      "loss": 2.8184,
+      "step": 2540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6011467889908258e-05,
+      "loss": 2.6871,
+      "step": 2541
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6004587155963303e-05,
+      "loss": 2.6702,
+      "step": 2542
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5997706422018348e-05,
+      "loss": 2.6447,
+      "step": 2543
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5990825688073396e-05,
+      "loss": 2.6835,
+      "step": 2544
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.598394495412844e-05,
+      "loss": 2.3757,
+      "step": 2545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.597706422018349e-05,
+      "loss": 2.4525,
+      "step": 2546
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5970183486238533e-05,
+      "loss": 2.4762,
+      "step": 2547
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5963302752293578e-05,
+      "loss": 2.4553,
+      "step": 2548
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5956422018348626e-05,
+      "loss": 2.2984,
+      "step": 2549
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.594954128440367e-05,
+      "loss": 2.3666,
+      "step": 2550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.5942660550458716e-05,
+      "loss": 2.3694,
+      "step": 2551
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.593577981651376e-05,
+      "loss": 2.3104,
+      "step": 2552
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5928899082568805e-05,
+      "loss": 2.3869,
+      "step": 2553
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5922018348623854e-05,
+      "loss": 2.2961,
+      "step": 2554
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.59151376146789e-05,
+      "loss": 2.3895,
+      "step": 2555
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5908256880733943e-05,
+      "loss": 2.317,
+      "step": 2556
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.590137614678899e-05,
+      "loss": 2.1834,
+      "step": 2557
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5894495412844036e-05,
+      "loss": 2.2361,
+      "step": 2558
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5887614678899084e-05,
+      "loss": 2.0873,
+      "step": 2559
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.588073394495413e-05,
+      "loss": 2.1551,
+      "step": 2560
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5873853211009174e-05,
+      "loss": 2.113,
+      "step": 2561
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5866972477064222e-05,
+      "loss": 1.9283,
+      "step": 2562
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5860091743119267e-05,
+      "loss": 2.0706,
+      "step": 2563
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5853211009174315e-05,
+      "loss": 2.1794,
+      "step": 2564
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.584633027522936e-05,
+      "loss": 2.0126,
+      "step": 2565
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5839449541284404e-05,
+      "loss": 1.8213,
+      "step": 2566
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5832568807339453e-05,
+      "loss": 1.8782,
+      "step": 2567
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5825688073394494e-05,
+      "loss": 1.6299,
+      "step": 2568
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.581880733944954e-05,
+      "loss": 1.5828,
+      "step": 2569
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5811926605504587e-05,
+      "loss": 1.432,
+      "step": 2570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.580504587155963e-05,
+      "loss": 3.5831,
+      "step": 2571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.579816513761468e-05,
+      "loss": 3.1797,
+      "step": 2572
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5791284403669725e-05,
+      "loss": 3.1354,
+      "step": 2573
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.578440366972477e-05,
+      "loss": 2.9805,
+      "step": 2574
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5777522935779818e-05,
+      "loss": 2.8388,
+      "step": 2575
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5770642201834862e-05,
+      "loss": 2.9647,
+      "step": 2576
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.576376146788991e-05,
+      "loss": 2.8283,
+      "step": 2577
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5756880733944955e-05,
+      "loss": 2.8195,
+      "step": 2578
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.575e-05,
+      "loss": 2.8317,
+      "step": 2579
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5743119266055048e-05,
+      "loss": 2.5203,
+      "step": 2580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5736238532110093e-05,
+      "loss": 2.7201,
+      "step": 2581
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5729357798165138e-05,
+      "loss": 2.8984,
+      "step": 2582
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5722477064220186e-05,
+      "loss": 2.6984,
+      "step": 2583
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5715596330275227e-05,
+      "loss": 2.7491,
+      "step": 2584
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5708715596330275e-05,
+      "loss": 2.7533,
+      "step": 2585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.570183486238532e-05,
+      "loss": 2.7296,
+      "step": 2586
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5694954128440365e-05,
+      "loss": 2.6288,
+      "step": 2587
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5688073394495413e-05,
+      "loss": 2.5928,
+      "step": 2588
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5681192660550458e-05,
+      "loss": 2.6059,
+      "step": 2589
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5674311926605506e-05,
+      "loss": 2.7595,
+      "step": 2590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.566743119266055e-05,
+      "loss": 2.5988,
+      "step": 2591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5660550458715596e-05,
+      "loss": 2.5625,
+      "step": 2592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5653669724770644e-05,
+      "loss": 2.475,
+      "step": 2593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.564678899082569e-05,
+      "loss": 2.5737,
+      "step": 2594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5639908256880737e-05,
+      "loss": 2.581,
+      "step": 2595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.563302752293578e-05,
+      "loss": 2.5516,
+      "step": 2596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5626146788990826e-05,
+      "loss": 2.4543,
+      "step": 2597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5619266055045874e-05,
+      "loss": 2.5543,
+      "step": 2598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.561238532110092e-05,
+      "loss": 2.4139,
+      "step": 2599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.560550458715596e-05,
+      "loss": 2.471,
+      "step": 2600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.559862385321101e-05,
+      "loss": 2.327,
+      "step": 2601
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5591743119266053e-05,
+      "loss": 2.5306,
+      "step": 2602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.55848623853211e-05,
+      "loss": 2.2596,
+      "step": 2603
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5577981651376146e-05,
+      "loss": 2.278,
+      "step": 2604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.557110091743119e-05,
+      "loss": 2.2782,
+      "step": 2605
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.556422018348624e-05,
+      "loss": 2.3388,
+      "step": 2606
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5557339449541284e-05,
+      "loss": 2.3224,
+      "step": 2607
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5550458715596332e-05,
+      "loss": 2.082,
+      "step": 2608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5543577981651377e-05,
+      "loss": 2.1797,
+      "step": 2609
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5536697247706422e-05,
+      "loss": 2.0622,
+      "step": 2610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.552981651376147e-05,
+      "loss": 2.2249,
+      "step": 2611
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5522935779816515e-05,
+      "loss": 1.9892,
+      "step": 2612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.551605504587156e-05,
+      "loss": 1.8331,
+      "step": 2613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5509174311926608e-05,
+      "loss": 2.068,
+      "step": 2614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5502293577981652e-05,
+      "loss": 1.9844,
+      "step": 2615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5495412844036697e-05,
+      "loss": 1.8131,
+      "step": 2616
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5488532110091742e-05,
+      "loss": 1.6899,
+      "step": 2617
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5481651376146787e-05,
+      "loss": 1.5964,
+      "step": 2618
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5474770642201835e-05,
+      "loss": 1.6063,
+      "step": 2619
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.546788990825688e-05,
+      "loss": 1.4219,
+      "step": 2620
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5461009174311928e-05,
+      "loss": 3.4508,
+      "step": 2621
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5454128440366973e-05,
+      "loss": 3.1922,
+      "step": 2622
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5447247706422017e-05,
+      "loss": 2.8854,
+      "step": 2623
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5440366972477066e-05,
+      "loss": 3.0534,
+      "step": 2624
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.543348623853211e-05,
+      "loss": 2.8355,
+      "step": 2625
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.542660550458716e-05,
+      "loss": 2.7808,
+      "step": 2626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5419724770642203e-05,
+      "loss": 2.7304,
+      "step": 2627
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5412844036697248e-05,
+      "loss": 2.8729,
+      "step": 2628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5405963302752296e-05,
+      "loss": 2.8827,
+      "step": 2629
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.539908256880734e-05,
+      "loss": 2.7435,
+      "step": 2630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5392201834862386e-05,
+      "loss": 2.6402,
+      "step": 2631
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.538532110091743e-05,
+      "loss": 2.7565,
+      "step": 2632
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5378440366972475e-05,
+      "loss": 2.7051,
+      "step": 2633
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5371559633027523e-05,
+      "loss": 2.5258,
+      "step": 2634
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5364678899082568e-05,
+      "loss": 2.6826,
+      "step": 2635
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5357798165137613e-05,
+      "loss": 2.5826,
+      "step": 2636
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.535091743119266e-05,
+      "loss": 2.6191,
+      "step": 2637
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5344036697247706e-05,
+      "loss": 2.5905,
+      "step": 2638
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5337155963302754e-05,
+      "loss": 2.6235,
+      "step": 2639
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.53302752293578e-05,
+      "loss": 2.4967,
+      "step": 2640
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5323394495412844e-05,
+      "loss": 2.5358,
+      "step": 2641
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5316513761467892e-05,
+      "loss": 2.5783,
+      "step": 2642
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5309633027522937e-05,
+      "loss": 2.5374,
+      "step": 2643
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.530275229357798e-05,
+      "loss": 2.4671,
+      "step": 2644
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.529587155963303e-05,
+      "loss": 2.6227,
+      "step": 2645
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5288990825688074e-05,
+      "loss": 2.4127,
+      "step": 2646
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5282110091743122e-05,
+      "loss": 2.5049,
+      "step": 2647
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5275229357798164e-05,
+      "loss": 2.5363,
+      "step": 2648
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.526834862385321e-05,
+      "loss": 2.3472,
+      "step": 2649
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5261467889908257e-05,
+      "loss": 2.3782,
+      "step": 2650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5254587155963301e-05,
+      "loss": 2.3721,
+      "step": 2651
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5247706422018348e-05,
+      "loss": 2.1756,
+      "step": 2652
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5240825688073394e-05,
+      "loss": 2.3979,
+      "step": 2653
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5233944954128441e-05,
+      "loss": 2.3226,
+      "step": 2654
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5227064220183486e-05,
+      "loss": 2.3736,
+      "step": 2655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5220183486238532e-05,
+      "loss": 2.1729,
+      "step": 2656
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5213302752293579e-05,
+      "loss": 2.2204,
+      "step": 2657
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5206422018348625e-05,
+      "loss": 2.2249,
+      "step": 2658
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5199541284403672e-05,
+      "loss": 2.2162,
+      "step": 2659
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5192660550458716e-05,
+      "loss": 2.1415,
+      "step": 2660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5185779816513763e-05,
+      "loss": 1.9807,
+      "step": 2661
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.517889908256881e-05,
+      "loss": 1.8545,
+      "step": 2662
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5172018348623856e-05,
+      "loss": 1.9308,
+      "step": 2663
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5165137614678897e-05,
+      "loss": 1.8903,
+      "step": 2664
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5158256880733944e-05,
+      "loss": 1.7782,
+      "step": 2665
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.515137614678899e-05,
+      "loss": 1.6815,
+      "step": 2666
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5144495412844036e-05,
+      "loss": 1.6589,
+      "step": 2667
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5137614678899083e-05,
+      "loss": 1.6394,
+      "step": 2668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5130733944954128e-05,
+      "loss": 1.5067,
+      "step": 2669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5123853211009174e-05,
+      "loss": 1.4108,
+      "step": 2670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.511697247706422e-05,
+      "loss": 3.4137,
+      "step": 2671
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5110091743119267e-05,
+      "loss": 3.3684,
+      "step": 2672
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5103211009174312e-05,
+      "loss": 3.0364,
+      "step": 2673
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5096330275229358e-05,
+      "loss": 2.9746,
+      "step": 2674
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5089449541284405e-05,
+      "loss": 2.7717,
+      "step": 2675
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5082568807339451e-05,
+      "loss": 2.7595,
+      "step": 2676
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5075688073394496e-05,
+      "loss": 2.9061,
+      "step": 2677
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5068807339449542e-05,
+      "loss": 2.7553,
+      "step": 2678
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5061926605504589e-05,
+      "loss": 2.9772,
+      "step": 2679
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5055045871559632e-05,
+      "loss": 2.7699,
+      "step": 2680
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5048165137614679e-05,
+      "loss": 2.5734,
+      "step": 2681
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5041284403669723e-05,
+      "loss": 2.7035,
+      "step": 2682
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.503440366972477e-05,
+      "loss": 2.6422,
+      "step": 2683
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5027522935779816e-05,
+      "loss": 2.789,
+      "step": 2684
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5020642201834863e-05,
+      "loss": 2.6551,
+      "step": 2685
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5013761467889907e-05,
+      "loss": 2.721,
+      "step": 2686
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5006880733944954e-05,
+      "loss": 2.4718,
+      "step": 2687
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5e-05,
+      "loss": 2.6105,
+      "step": 2688
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4993119266055047e-05,
+      "loss": 2.5689,
+      "step": 2689
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4986238532110093e-05,
+      "loss": 2.4748,
+      "step": 2690
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4979357798165138e-05,
+      "loss": 2.4915,
+      "step": 2691
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4972477064220185e-05,
+      "loss": 2.5606,
+      "step": 2692
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.496559633027523e-05,
+      "loss": 2.4461,
+      "step": 2693
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4958715596330276e-05,
+      "loss": 2.5335,
+      "step": 2694
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.495183486238532e-05,
+      "loss": 2.6517,
+      "step": 2695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4944954128440367e-05,
+      "loss": 2.5127,
+      "step": 2696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.4938073394495413e-05,
+      "loss": 2.38,
+      "step": 2697
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.493119266055046e-05,
+      "loss": 2.4457,
+      "step": 2698
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4924311926605506e-05,
+      "loss": 2.4534,
+      "step": 2699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4917431192660551e-05,
+      "loss": 2.4709,
+      "step": 2700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4910550458715596e-05,
+      "loss": 2.3339,
+      "step": 2701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4903669724770642e-05,
+      "loss": 2.318,
+      "step": 2702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4896788990825689e-05,
+      "loss": 2.2709,
+      "step": 2703
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4889908256880734e-05,
+      "loss": 2.1571,
+      "step": 2704
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.488302752293578e-05,
+      "loss": 2.1276,
+      "step": 2705
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4876146788990827e-05,
+      "loss": 2.4123,
+      "step": 2706
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4869266055045873e-05,
+      "loss": 2.255,
+      "step": 2707
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4862385321100918e-05,
+      "loss": 2.0954,
+      "step": 2708
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4855504587155963e-05,
+      "loss": 2.0177,
+      "step": 2709
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4848623853211009e-05,
+      "loss": 1.9596,
+      "step": 2710
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4841743119266056e-05,
+      "loss": 1.9975,
+      "step": 2711
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4834862385321102e-05,
+      "loss": 2.0821,
+      "step": 2712
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.4827981651376147e-05,
+      "loss": 2.1313,
+      "step": 2713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4821100917431193e-05,
+      "loss": 1.9048,
+      "step": 2714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.481422018348624e-05,
+      "loss": 1.9256,
+      "step": 2715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4807339449541286e-05,
+      "loss": 1.747,
+      "step": 2716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.480045871559633e-05,
+      "loss": 1.5417,
+      "step": 2717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4793577981651376e-05,
+      "loss": 1.5457,
+      "step": 2718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4786697247706422e-05,
+      "loss": 1.4545,
+      "step": 2719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4779816513761469e-05,
+      "loss": 1.5234,
+      "step": 2720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4772935779816515e-05,
+      "loss": 3.4627,
+      "step": 2721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.476605504587156e-05,
+      "loss": 3.2405,
+      "step": 2722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4759174311926606e-05,
+      "loss": 2.9264,
+      "step": 2723
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4752293577981653e-05,
+      "loss": 2.9154,
+      "step": 2724
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4745412844036698e-05,
+      "loss": 2.7967,
+      "step": 2725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4738532110091742e-05,
+      "loss": 2.8633,
+      "step": 2726
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4731651376146789e-05,
+      "loss": 2.7275,
+      "step": 2727
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4724770642201835e-05,
+      "loss": 2.7148,
+      "step": 2728
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.4717889908256882e-05,
+      "loss": 2.9467,
+      "step": 2729
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4711009174311927e-05,
+      "loss": 2.7409,
+      "step": 2730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4704128440366973e-05,
+      "loss": 2.7502,
+      "step": 2731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.469724770642202e-05,
+      "loss": 2.7589,
+      "step": 2732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4690366972477064e-05,
+      "loss": 2.7219,
+      "step": 2733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.468348623853211e-05,
+      "loss": 2.7015,
+      "step": 2734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4676605504587155e-05,
+      "loss": 2.7633,
+      "step": 2735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4669724770642202e-05,
+      "loss": 2.6205,
+      "step": 2736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4662844036697248e-05,
+      "loss": 2.6494,
+      "step": 2737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4655963302752295e-05,
+      "loss": 2.6662,
+      "step": 2738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.464908256880734e-05,
+      "loss": 2.5482,
+      "step": 2739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4642201834862386e-05,
+      "loss": 2.644,
+      "step": 2740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4635321100917431e-05,
+      "loss": 2.3397,
+      "step": 2741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4628440366972477e-05,
+      "loss": 2.5752,
+      "step": 2742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4621559633027524e-05,
+      "loss": 2.4876,
+      "step": 2743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4614678899082569e-05,
+      "loss": 2.5306,
+      "step": 2744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.4607798165137615e-05,
+      "loss": 2.4259,
+      "step": 2745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4600917431192662e-05,
+      "loss": 2.4461,
+      "step": 2746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4594036697247708e-05,
+      "loss": 2.3251,
+      "step": 2747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4587155963302753e-05,
+      "loss": 2.4803,
+      "step": 2748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4580275229357798e-05,
+      "loss": 2.2513,
+      "step": 2749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4573394495412844e-05,
+      "loss": 2.4194,
+      "step": 2750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.456651376146789e-05,
+      "loss": 2.2233,
+      "step": 2751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4559633027522937e-05,
+      "loss": 2.306,
+      "step": 2752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4552752293577982e-05,
+      "loss": 2.3841,
+      "step": 2753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4545871559633028e-05,
+      "loss": 2.131,
+      "step": 2754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4538990825688075e-05,
+      "loss": 2.2118,
+      "step": 2755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4532110091743121e-05,
+      "loss": 2.3629,
+      "step": 2756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4525229357798164e-05,
+      "loss": 2.185,
+      "step": 2757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.451834862385321e-05,
+      "loss": 2.115,
+      "step": 2758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4511467889908257e-05,
+      "loss": 2.0509,
+      "step": 2759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4504587155963304e-05,
+      "loss": 2.0482,
+      "step": 2760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4497706422018348e-05,
+      "loss": 1.9013,
+      "step": 2761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.4490825688073395e-05,
+      "loss": 2.0503,
+      "step": 2762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4483944954128441e-05,
+      "loss": 1.8546,
+      "step": 2763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4477064220183488e-05,
+      "loss": 1.9758,
+      "step": 2764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4470183486238533e-05,
+      "loss": 1.7113,
+      "step": 2765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4463302752293577e-05,
+      "loss": 1.552,
+      "step": 2766
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4456422018348624e-05,
+      "loss": 1.8108,
+      "step": 2767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.444954128440367e-05,
+      "loss": 1.7141,
+      "step": 2768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4442660550458717e-05,
+      "loss": 1.5113,
+      "step": 2769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4435779816513761e-05,
+      "loss": 1.3695,
+      "step": 2770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4428899082568808e-05,
+      "loss": 3.3743,
+      "step": 2771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4422018348623854e-05,
+      "loss": 3.1657,
+      "step": 2772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.44151376146789e-05,
+      "loss": 2.8047,
+      "step": 2773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4408256880733946e-05,
+      "loss": 2.7099,
+      "step": 2774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.440137614678899e-05,
+      "loss": 2.7988,
+      "step": 2775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4394495412844037e-05,
+      "loss": 2.5724,
+      "step": 2776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.4387614678899083e-05,
+      "loss": 2.6011,
+      "step": 2777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.438073394495413e-05,
+      "loss": 2.6557,
+      "step": 2778
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4373853211009175e-05,
+      "loss": 2.6613,
+      "step": 2779
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4366972477064221e-05,
+      "loss": 2.6057,
+      "step": 2780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4360091743119266e-05,
+      "loss": 2.6579,
+      "step": 2781
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4353211009174312e-05,
+      "loss": 2.7009,
+      "step": 2782
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4346330275229359e-05,
+      "loss": 2.549,
+      "step": 2783
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4339449541284404e-05,
+      "loss": 2.6545,
+      "step": 2784
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.433256880733945e-05,
+      "loss": 2.5939,
+      "step": 2785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4325688073394496e-05,
+      "loss": 2.4487,
+      "step": 2786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4318807339449543e-05,
+      "loss": 2.5801,
+      "step": 2787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4311926605504588e-05,
+      "loss": 2.7621,
+      "step": 2788
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4305045871559632e-05,
+      "loss": 2.4606,
+      "step": 2789
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4298165137614679e-05,
+      "loss": 2.4613,
+      "step": 2790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4291284403669725e-05,
+      "loss": 2.4676,
+      "step": 2791
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.428440366972477e-05,
+      "loss": 2.3937,
+      "step": 2792
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4277522935779817e-05,
+      "loss": 2.5138,
+      "step": 2793
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.4270642201834863e-05,
+      "loss": 2.3879,
+      "step": 2794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.426376146788991e-05,
+      "loss": 2.5767,
+      "step": 2795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4256880733944956e-05,
+      "loss": 2.4291,
+      "step": 2796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 2.3746,
+      "step": 2797
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4243119266055046e-05,
+      "loss": 2.6246,
+      "step": 2798
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4236238532110092e-05,
+      "loss": 2.3215,
+      "step": 2799
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4229357798165138e-05,
+      "loss": 2.422,
+      "step": 2800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4222477064220183e-05,
+      "loss": 2.174,
+      "step": 2801
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.421559633027523e-05,
+      "loss": 2.322,
+      "step": 2802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4208715596330276e-05,
+      "loss": 2.223,
+      "step": 2803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4201834862385323e-05,
+      "loss": 2.1753,
+      "step": 2804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4194954128440367e-05,
+      "loss": 2.2133,
+      "step": 2805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4188073394495412e-05,
+      "loss": 2.1521,
+      "step": 2806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4181192660550459e-05,
+      "loss": 2.1413,
+      "step": 2807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4174311926605505e-05,
+      "loss": 2.0932,
+      "step": 2808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4167431192660552e-05,
+      "loss": 2.0478,
+      "step": 2809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.4160550458715596e-05,
+      "loss": 1.9908,
+      "step": 2810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4153669724770643e-05,
+      "loss": 2.0011,
+      "step": 2811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.414678899082569e-05,
+      "loss": 1.827,
+      "step": 2812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4139908256880734e-05,
+      "loss": 1.9547,
+      "step": 2813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.413302752293578e-05,
+      "loss": 1.8594,
+      "step": 2814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4126146788990825e-05,
+      "loss": 1.7927,
+      "step": 2815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4119266055045872e-05,
+      "loss": 1.6374,
+      "step": 2816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4112385321100918e-05,
+      "loss": 1.7043,
+      "step": 2817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4105504587155965e-05,
+      "loss": 1.5083,
+      "step": 2818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.409862385321101e-05,
+      "loss": 1.3917,
+      "step": 2819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4091743119266056e-05,
+      "loss": 1.682,
+      "step": 2820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.40848623853211e-05,
+      "loss": 3.2782,
+      "step": 2821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4077981651376147e-05,
+      "loss": 3.1333,
+      "step": 2822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4071100917431192e-05,
+      "loss": 2.953,
+      "step": 2823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4064220183486238e-05,
+      "loss": 2.8578,
+      "step": 2824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4057339449541285e-05,
+      "loss": 2.6731,
+      "step": 2825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.4050458715596331e-05,
+      "loss": 2.6259,
+      "step": 2826
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4043577981651378e-05,
+      "loss": 2.7913,
+      "step": 2827
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4036697247706423e-05,
+      "loss": 2.6696,
+      "step": 2828
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4029816513761467e-05,
+      "loss": 2.7547,
+      "step": 2829
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4022935779816514e-05,
+      "loss": 2.5509,
+      "step": 2830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.401605504587156e-05,
+      "loss": 2.4786,
+      "step": 2831
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4009174311926605e-05,
+      "loss": 2.6349,
+      "step": 2832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.4002293577981652e-05,
+      "loss": 2.5055,
+      "step": 2833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3995412844036698e-05,
+      "loss": 2.339,
+      "step": 2834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3988532110091744e-05,
+      "loss": 2.5601,
+      "step": 2835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3981651376146791e-05,
+      "loss": 2.5807,
+      "step": 2836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3974770642201834e-05,
+      "loss": 2.6846,
+      "step": 2837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.396788990825688e-05,
+      "loss": 2.4524,
+      "step": 2838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3961009174311927e-05,
+      "loss": 2.5231,
+      "step": 2839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3954128440366973e-05,
+      "loss": 2.4482,
+      "step": 2840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3947247706422018e-05,
+      "loss": 2.4243,
+      "step": 2841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3940366972477065e-05,
+      "loss": 2.4645,
+      "step": 2842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3933486238532111e-05,
+      "loss": 2.3849,
+      "step": 2843
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3926605504587158e-05,
+      "loss": 2.6005,
+      "step": 2844
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3919724770642202e-05,
+      "loss": 2.3232,
+      "step": 2845
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3912844036697247e-05,
+      "loss": 2.2848,
+      "step": 2846
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3905963302752294e-05,
+      "loss": 2.5224,
+      "step": 2847
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.389908256880734e-05,
+      "loss": 2.287,
+      "step": 2848
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3892201834862387e-05,
+      "loss": 2.2758,
+      "step": 2849
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3885321100917431e-05,
+      "loss": 2.2333,
+      "step": 2850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3878440366972478e-05,
+      "loss": 2.3837,
+      "step": 2851
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3871559633027524e-05,
+      "loss": 2.2391,
+      "step": 2852
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3864678899082569e-05,
+      "loss": 2.283,
+      "step": 2853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3857798165137614e-05,
+      "loss": 2.2234,
+      "step": 2854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.385091743119266e-05,
+      "loss": 2.1756,
+      "step": 2855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3844036697247707e-05,
+      "loss": 1.9754,
+      "step": 2856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3837155963302753e-05,
+      "loss": 2.1553,
+      "step": 2857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.38302752293578e-05,
+      "loss": 2.0373,
+      "step": 2858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.3823394495412844e-05,
+      "loss": 1.9749,
+      "step": 2859
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3816513761467891e-05,
+      "loss": 1.9547,
+      "step": 2860
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3809633027522936e-05,
+      "loss": 1.9455,
+      "step": 2861
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3802752293577982e-05,
+      "loss": 1.8361,
+      "step": 2862
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3795871559633027e-05,
+      "loss": 1.8596,
+      "step": 2863
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3788990825688073e-05,
+      "loss": 2.0573,
+      "step": 2864
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.378211009174312e-05,
+      "loss": 1.8749,
+      "step": 2865
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3775229357798166e-05,
+      "loss": 1.7497,
+      "step": 2866
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3768348623853213e-05,
+      "loss": 1.7243,
+      "step": 2867
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3761467889908258e-05,
+      "loss": 1.4406,
+      "step": 2868
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3754587155963302e-05,
+      "loss": 1.4597,
+      "step": 2869
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3747706422018349e-05,
+      "loss": 1.5117,
+      "step": 2870
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3740825688073395e-05,
+      "loss": 3.4703,
+      "step": 2871
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.373394495412844e-05,
+      "loss": 3.1261,
+      "step": 2872
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3727064220183486e-05,
+      "loss": 2.92,
+      "step": 2873
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.3720183486238533e-05,
+      "loss": 2.6432,
+      "step": 2874
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.371330275229358e-05,
+      "loss": 2.71,
+      "step": 2875
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3706422018348624e-05,
+      "loss": 2.7699,
+      "step": 2876
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3699541284403669e-05,
+      "loss": 2.7135,
+      "step": 2877
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3692660550458715e-05,
+      "loss": 2.5663,
+      "step": 2878
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3685779816513762e-05,
+      "loss": 2.6016,
+      "step": 2879
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3678899082568808e-05,
+      "loss": 2.5623,
+      "step": 2880
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3672018348623853e-05,
+      "loss": 2.5552,
+      "step": 2881
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.36651376146789e-05,
+      "loss": 2.5718,
+      "step": 2882
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3658256880733946e-05,
+      "loss": 2.4715,
+      "step": 2883
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3651376146788992e-05,
+      "loss": 2.4225,
+      "step": 2884
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3644495412844036e-05,
+      "loss": 2.5825,
+      "step": 2885
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3637614678899082e-05,
+      "loss": 2.5032,
+      "step": 2886
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3630733944954129e-05,
+      "loss": 2.5283,
+      "step": 2887
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3623853211009175e-05,
+      "loss": 2.4033,
+      "step": 2888
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3616972477064221e-05,
+      "loss": 2.3952,
+      "step": 2889
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3610091743119266e-05,
+      "loss": 2.4645,
+      "step": 2890
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.3603211009174313e-05,
+      "loss": 2.5607,
+      "step": 2891
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3596330275229359e-05,
+      "loss": 2.5435,
+      "step": 2892
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3589449541284404e-05,
+      "loss": 2.477,
+      "step": 2893
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3582568807339449e-05,
+      "loss": 2.4841,
+      "step": 2894
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3575688073394495e-05,
+      "loss": 2.2027,
+      "step": 2895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3568807339449542e-05,
+      "loss": 2.2889,
+      "step": 2896
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3561926605504588e-05,
+      "loss": 2.3684,
+      "step": 2897
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3555045871559635e-05,
+      "loss": 2.3061,
+      "step": 2898
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.354816513761468e-05,
+      "loss": 2.1622,
+      "step": 2899
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3541284403669726e-05,
+      "loss": 2.2372,
+      "step": 2900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.353440366972477e-05,
+      "loss": 2.1794,
+      "step": 2901
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3527522935779817e-05,
+      "loss": 2.2728,
+      "step": 2902
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3520642201834862e-05,
+      "loss": 2.1175,
+      "step": 2903
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3513761467889908e-05,
+      "loss": 2.2185,
+      "step": 2904
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3506880733944955e-05,
+      "loss": 2.2166,
+      "step": 2905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.2319,
+      "step": 2906
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.3493119266055046e-05,
+      "loss": 2.0092,
+      "step": 2907
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3486238532110092e-05,
+      "loss": 2.1144,
+      "step": 2908
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3479357798165137e-05,
+      "loss": 1.9269,
+      "step": 2909
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3472477064220184e-05,
+      "loss": 2.0476,
+      "step": 2910
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.346559633027523e-05,
+      "loss": 1.9823,
+      "step": 2911
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3458715596330275e-05,
+      "loss": 1.8872,
+      "step": 2912
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3451834862385321e-05,
+      "loss": 1.8575,
+      "step": 2913
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3444954128440368e-05,
+      "loss": 1.7426,
+      "step": 2914
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3438073394495414e-05,
+      "loss": 1.8052,
+      "step": 2915
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3431192660550459e-05,
+      "loss": 1.616,
+      "step": 2916
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3424311926605504e-05,
+      "loss": 1.529,
+      "step": 2917
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.341743119266055e-05,
+      "loss": 1.6757,
+      "step": 2918
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3410550458715597e-05,
+      "loss": 1.2858,
+      "step": 2919
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3403669724770643e-05,
+      "loss": 1.5484,
+      "step": 2920
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3396788990825688e-05,
+      "loss": 3.4052,
+      "step": 2921
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3389908256880734e-05,
+      "loss": 3.0176,
+      "step": 2922
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3383027522935781e-05,
+      "loss": 2.8299,
+      "step": 2923
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.3376146788990827e-05,
+      "loss": 2.6678,
+      "step": 2924
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.336926605504587e-05,
+      "loss": 2.8276,
+      "step": 2925
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3362385321100917e-05,
+      "loss": 2.6767,
+      "step": 2926
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3355504587155963e-05,
+      "loss": 2.666,
+      "step": 2927
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.334862385321101e-05,
+      "loss": 2.6676,
+      "step": 2928
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3341743119266055e-05,
+      "loss": 2.5886,
+      "step": 2929
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3334862385321101e-05,
+      "loss": 2.767,
+      "step": 2930
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3327981651376148e-05,
+      "loss": 2.6808,
+      "step": 2931
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3321100917431194e-05,
+      "loss": 2.6306,
+      "step": 2932
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3314220183486239e-05,
+      "loss": 2.7255,
+      "step": 2933
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3307339449541284e-05,
+      "loss": 2.5518,
+      "step": 2934
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.330045871559633e-05,
+      "loss": 2.5159,
+      "step": 2935
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3293577981651377e-05,
+      "loss": 2.4584,
+      "step": 2936
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3286697247706423e-05,
+      "loss": 2.5979,
+      "step": 2937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3279816513761468e-05,
+      "loss": 2.4542,
+      "step": 2938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.3272935779816514e-05,
+      "loss": 2.3852,
+      "step": 2939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.326605504587156e-05,
+      "loss": 2.2678,
+      "step": 2940
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3259174311926605e-05,
+      "loss": 2.5971,
+      "step": 2941
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3252293577981652e-05,
+      "loss": 2.2834,
+      "step": 2942
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3245412844036697e-05,
+      "loss": 2.4862,
+      "step": 2943
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3238532110091743e-05,
+      "loss": 2.4527,
+      "step": 2944
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.323165137614679e-05,
+      "loss": 2.3268,
+      "step": 2945
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3224770642201836e-05,
+      "loss": 2.4645,
+      "step": 2946
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3217889908256881e-05,
+      "loss": 2.3773,
+      "step": 2947
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3211009174311927e-05,
+      "loss": 2.2755,
+      "step": 2948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3204128440366972e-05,
+      "loss": 2.3679,
+      "step": 2949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3197247706422019e-05,
+      "loss": 2.2787,
+      "step": 2950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3190366972477065e-05,
+      "loss": 2.2488,
+      "step": 2951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.318348623853211e-05,
+      "loss": 2.336,
+      "step": 2952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3176605504587156e-05,
+      "loss": 2.4236,
+      "step": 2953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3169724770642203e-05,
+      "loss": 2.2183,
+      "step": 2954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.316284403669725e-05,
+      "loss": 2.2357,
+      "step": 2955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.3155963302752294e-05,
+      "loss": 2.0351,
+      "step": 2956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3149082568807339e-05,
+      "loss": 2.0037,
+      "step": 2957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3142201834862385e-05,
+      "loss": 2.0366,
+      "step": 2958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3135321100917432e-05,
+      "loss": 1.9411,
+      "step": 2959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3128440366972476e-05,
+      "loss": 1.8521,
+      "step": 2960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3121559633027523e-05,
+      "loss": 1.9356,
+      "step": 2961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.311467889908257e-05,
+      "loss": 2.0139,
+      "step": 2962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3107798165137616e-05,
+      "loss": 1.7847,
+      "step": 2963
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3100917431192662e-05,
+      "loss": 1.8232,
+      "step": 2964
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3094036697247705e-05,
+      "loss": 1.778,
+      "step": 2965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3087155963302752e-05,
+      "loss": 1.4748,
+      "step": 2966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3080275229357798e-05,
+      "loss": 1.521,
+      "step": 2967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3073394495412845e-05,
+      "loss": 1.4889,
+      "step": 2968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.306651376146789e-05,
+      "loss": 1.4884,
+      "step": 2969
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3059633027522936e-05,
+      "loss": 1.51,
+      "step": 2970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3052752293577983e-05,
+      "loss": 3.3206,
+      "step": 2971
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.3045871559633029e-05,
+      "loss": 2.9253,
+      "step": 2972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3038990825688074e-05,
+      "loss": 2.8491,
+      "step": 2973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3032110091743119e-05,
+      "loss": 2.6323,
+      "step": 2974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3025229357798165e-05,
+      "loss": 2.7448,
+      "step": 2975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3018348623853211e-05,
+      "loss": 2.6932,
+      "step": 2976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3011467889908258e-05,
+      "loss": 2.6353,
+      "step": 2977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.3004587155963303e-05,
+      "loss": 2.6757,
+      "step": 2978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2997706422018349e-05,
+      "loss": 2.5732,
+      "step": 2979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2990825688073396e-05,
+      "loss": 2.5755,
+      "step": 2980
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.298394495412844e-05,
+      "loss": 2.5756,
+      "step": 2981
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2977064220183487e-05,
+      "loss": 2.4928,
+      "step": 2982
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2970183486238532e-05,
+      "loss": 2.5819,
+      "step": 2983
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2963302752293578e-05,
+      "loss": 2.6818,
+      "step": 2984
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2956422018348625e-05,
+      "loss": 2.5527,
+      "step": 2985
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2949541284403671e-05,
+      "loss": 2.463,
+      "step": 2986
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2942660550458716e-05,
+      "loss": 2.5635,
+      "step": 2987
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2935779816513762e-05,
+      "loss": 2.5764,
+      "step": 2988
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2928899082568807e-05,
+      "loss": 2.3739,
+      "step": 2989
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2922018348623853e-05,
+      "loss": 2.4959,
+      "step": 2990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2915137614678898e-05,
+      "loss": 2.3974,
+      "step": 2991
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2908256880733945e-05,
+      "loss": 2.5056,
+      "step": 2992
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2901376146788991e-05,
+      "loss": 2.5027,
+      "step": 2993
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2894495412844038e-05,
+      "loss": 2.4639,
+      "step": 2994
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2887614678899084e-05,
+      "loss": 2.448,
+      "step": 2995
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2880733944954129e-05,
+      "loss": 2.3703,
+      "step": 2996
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2873853211009174e-05,
+      "loss": 2.3763,
+      "step": 2997
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.286697247706422e-05,
+      "loss": 2.4714,
+      "step": 2998
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2860091743119267e-05,
+      "loss": 2.3097,
+      "step": 2999
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2853211009174311e-05,
+      "loss": 2.322,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "eval_bleu": 1.9304498489031548e-13,
+      "eval_loss": 2.408076047897339,
+      "eval_runtime": 2592.9468,
+      "eval_samples_per_second": 5.692,
+      "eval_steps_per_second": 0.712,
+      "step": 3000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2846330275229358e-05,
+      "loss": 2.3288,
+      "step": 3001
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2839449541284404e-05,
+      "loss": 2.215,
+      "step": 3002
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.283256880733945e-05,
+      "loss": 2.083,
+      "step": 3003
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2825688073394497e-05,
+      "loss": 1.9948,
+      "step": 3004
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.281880733944954e-05,
+      "loss": 2.2531,
+      "step": 3005
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2811926605504587e-05,
+      "loss": 2.0511,
+      "step": 3006
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2805045871559633e-05,
+      "loss": 2.1926,
+      "step": 3007
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.279816513761468e-05,
+      "loss": 1.9659,
+      "step": 3008
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2791284403669724e-05,
+      "loss": 2.0675,
+      "step": 3009
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2784403669724771e-05,
+      "loss": 2.0515,
+      "step": 3010
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2777522935779817e-05,
+      "loss": 1.8359,
+      "step": 3011
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2770642201834864e-05,
+      "loss": 1.8535,
+      "step": 3012
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2763761467889909e-05,
+      "loss": 1.8507,
+      "step": 3013
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2756880733944953e-05,
+      "loss": 1.739,
+      "step": 3014
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.275e-05,
+      "loss": 1.7355,
+      "step": 3015
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2743119266055046e-05,
+      "loss": 1.6592,
+      "step": 3016
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2736238532110093e-05,
+      "loss": 1.5465,
+      "step": 3017
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2729357798165138e-05,
+      "loss": 1.5753,
+      "step": 3018
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2722477064220184e-05,
+      "loss": 1.3659,
+      "step": 3019
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.271559633027523e-05,
+      "loss": 1.3845,
+      "step": 3020
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2708715596330275e-05,
+      "loss": 3.3431,
+      "step": 3021
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.270183486238532e-05,
+      "loss": 2.8944,
+      "step": 3022
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2694954128440367e-05,
+      "loss": 2.8238,
+      "step": 3023
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2688073394495413e-05,
+      "loss": 2.739,
+      "step": 3024
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.268119266055046e-05,
+      "loss": 2.4485,
+      "step": 3025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2674311926605506e-05,
+      "loss": 2.4478,
+      "step": 3026
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.266743119266055e-05,
+      "loss": 2.6612,
+      "step": 3027
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2660550458715597e-05,
+      "loss": 2.4487,
+      "step": 3028
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2653669724770642e-05,
+      "loss": 2.4835,
+      "step": 3029
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2646788990825688e-05,
+      "loss": 2.5644,
+      "step": 3030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2639908256880733e-05,
+      "loss": 2.3917,
+      "step": 3031
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.263302752293578e-05,
+      "loss": 2.5056,
+      "step": 3032
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2626146788990826e-05,
+      "loss": 2.4112,
+      "step": 3033
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2619266055045873e-05,
+      "loss": 2.2492,
+      "step": 3034
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2612385321100919e-05,
+      "loss": 2.3584,
+      "step": 3035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2605504587155964e-05,
+      "loss": 2.393,
+      "step": 3036
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2598623853211009e-05,
+      "loss": 2.6865,
+      "step": 3037
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2591743119266055e-05,
+      "loss": 2.4825,
+      "step": 3038
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2584862385321102e-05,
+      "loss": 2.5632,
+      "step": 3039
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2577981651376146e-05,
+      "loss": 2.3153,
+      "step": 3040
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2571100917431193e-05,
+      "loss": 2.3363,
+      "step": 3041
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.256422018348624e-05,
+      "loss": 2.304,
+      "step": 3042
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2557339449541286e-05,
+      "loss": 2.5048,
+      "step": 3043
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.255045871559633e-05,
+      "loss": 2.291,
+      "step": 3044
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2543577981651375e-05,
+      "loss": 2.3893,
+      "step": 3045
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2536697247706422e-05,
+      "loss": 2.3866,
+      "step": 3046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2529816513761468e-05,
+      "loss": 2.3302,
+      "step": 3047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2522935779816515e-05,
+      "loss": 2.1537,
+      "step": 3048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.251605504587156e-05,
+      "loss": 2.2312,
+      "step": 3049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2509174311926606e-05,
+      "loss": 2.2714,
+      "step": 3050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2502293577981652e-05,
+      "loss": 2.3226,
+      "step": 3051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2495412844036699e-05,
+      "loss": 2.188,
+      "step": 3052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2488532110091742e-05,
+      "loss": 2.1637,
+      "step": 3053
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2481651376146788e-05,
+      "loss": 2.1937,
+      "step": 3054
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2474770642201835e-05,
+      "loss": 1.9789,
+      "step": 3055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2467889908256881e-05,
+      "loss": 2.0782,
+      "step": 3056
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2461009174311928e-05,
+      "loss": 1.9831,
+      "step": 3057
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2454128440366973e-05,
+      "loss": 2.0571,
+      "step": 3058
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2447247706422019e-05,
+      "loss": 2.0079,
+      "step": 3059
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2440366972477065e-05,
+      "loss": 1.819,
+      "step": 3060
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.243348623853211e-05,
+      "loss": 1.9509,
+      "step": 3061
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2426605504587155e-05,
+      "loss": 1.9455,
+      "step": 3062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2419724770642201e-05,
+      "loss": 1.7811,
+      "step": 3063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2412844036697248e-05,
+      "loss": 1.9359,
+      "step": 3064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2405963302752294e-05,
+      "loss": 1.8575,
+      "step": 3065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2399082568807341e-05,
+      "loss": 1.7008,
+      "step": 3066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2392201834862386e-05,
+      "loss": 1.6792,
+      "step": 3067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2385321100917432e-05,
+      "loss": 1.5675,
+      "step": 3068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2378440366972477e-05,
+      "loss": 1.4848,
+      "step": 3069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2371559633027523e-05,
+      "loss": 1.3532,
+      "step": 3070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2364678899082568e-05,
+      "loss": 3.1958,
+      "step": 3071
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2357798165137615e-05,
+      "loss": 2.9836,
+      "step": 3072
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2350917431192661e-05,
+      "loss": 2.7372,
+      "step": 3073
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2344036697247708e-05,
+      "loss": 2.7378,
+      "step": 3074
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2337155963302752e-05,
+      "loss": 2.8076,
+      "step": 3075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2330275229357799e-05,
+      "loss": 2.4573,
+      "step": 3076
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2323394495412844e-05,
+      "loss": 2.378,
+      "step": 3077
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.231651376146789e-05,
+      "loss": 2.5274,
+      "step": 3078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2309633027522936e-05,
+      "loss": 2.5824,
+      "step": 3079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2302752293577981e-05,
+      "loss": 2.5781,
+      "step": 3080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2295871559633028e-05,
+      "loss": 2.533,
+      "step": 3081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2288990825688074e-05,
+      "loss": 2.4477,
+      "step": 3082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.228211009174312e-05,
+      "loss": 2.4011,
+      "step": 3083
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2275229357798165e-05,
+      "loss": 2.3943,
+      "step": 3084
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.226834862385321e-05,
+      "loss": 2.4015,
+      "step": 3085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2261467889908257e-05,
+      "loss": 2.4142,
+      "step": 3086
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2254587155963303e-05,
+      "loss": 2.4216,
+      "step": 3087
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.224770642201835e-05,
+      "loss": 2.4345,
+      "step": 3088
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2240825688073394e-05,
+      "loss": 2.3985,
+      "step": 3089
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.223394495412844e-05,
+      "loss": 2.3602,
+      "step": 3090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2227064220183487e-05,
+      "loss": 2.3,
+      "step": 3091
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2220183486238534e-05,
+      "loss": 2.2679,
+      "step": 3092
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2213302752293577e-05,
+      "loss": 2.378,
+      "step": 3093
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2206422018348623e-05,
+      "loss": 2.3487,
+      "step": 3094
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.219954128440367e-05,
+      "loss": 2.416,
+      "step": 3095
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2192660550458716e-05,
+      "loss": 2.198,
+      "step": 3096
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2185779816513763e-05,
+      "loss": 2.3193,
+      "step": 3097
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2178899082568807e-05,
+      "loss": 2.2336,
+      "step": 3098
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2172018348623854e-05,
+      "loss": 2.1397,
+      "step": 3099
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.21651376146789e-05,
+      "loss": 2.1257,
+      "step": 3100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2158256880733945e-05,
+      "loss": 2.2413,
+      "step": 3101
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.215137614678899e-05,
+      "loss": 2.1644,
+      "step": 3102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2144495412844036e-05,
+      "loss": 2.291,
+      "step": 3103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2137614678899083e-05,
+      "loss": 2.0318,
+      "step": 3104
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.213073394495413e-05,
+      "loss": 2.0868,
+      "step": 3105
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2123853211009174e-05,
+      "loss": 1.9583,
+      "step": 3106
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.211697247706422e-05,
+      "loss": 2.0158,
+      "step": 3107
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2110091743119267e-05,
+      "loss": 2.0302,
+      "step": 3108
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2103211009174312e-05,
+      "loss": 2.0003,
+      "step": 3109
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2096330275229358e-05,
+      "loss": 1.9671,
+      "step": 3110
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2089449541284403e-05,
+      "loss": 1.7879,
+      "step": 3111
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.208256880733945e-05,
+      "loss": 1.8751,
+      "step": 3112
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2075688073394496e-05,
+      "loss": 1.8304,
+      "step": 3113
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2068807339449542e-05,
+      "loss": 1.7742,
+      "step": 3114
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2061926605504587e-05,
+      "loss": 1.4739,
+      "step": 3115
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2055045871559634e-05,
+      "loss": 1.6761,
+      "step": 3116
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2048165137614678e-05,
+      "loss": 1.6252,
+      "step": 3117
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2041284403669725e-05,
+      "loss": 1.4457,
+      "step": 3118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2034403669724771e-05,
+      "loss": 1.5433,
+      "step": 3119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2027522935779816e-05,
+      "loss": 1.29,
+      "step": 3120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2020642201834863e-05,
+      "loss": 3.1475,
+      "step": 3121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2013761467889909e-05,
+      "loss": 3.0629,
+      "step": 3122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2006880733944956e-05,
+      "loss": 2.7221,
+      "step": 3123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2e-05,
+      "loss": 2.6695,
+      "step": 3124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1993119266055045e-05,
+      "loss": 2.5284,
+      "step": 3125
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1986238532110092e-05,
+      "loss": 2.4847,
+      "step": 3126
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1979357798165138e-05,
+      "loss": 2.6669,
+      "step": 3127
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1972477064220183e-05,
+      "loss": 2.4682,
+      "step": 3128
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.196559633027523e-05,
+      "loss": 2.5247,
+      "step": 3129
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1958715596330276e-05,
+      "loss": 2.4724,
+      "step": 3130
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1951834862385322e-05,
+      "loss": 2.531,
+      "step": 3131
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1944954128440369e-05,
+      "loss": 2.5517,
+      "step": 3132
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1938073394495412e-05,
+      "loss": 2.4616,
+      "step": 3133
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1931192660550458e-05,
+      "loss": 2.4486,
+      "step": 3134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1924311926605505e-05,
+      "loss": 2.4777,
+      "step": 3135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1917431192660551e-05,
+      "loss": 2.3763,
+      "step": 3136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1910550458715596e-05,
+      "loss": 2.4084,
+      "step": 3137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1903669724770642e-05,
+      "loss": 2.2963,
+      "step": 3138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1896788990825689e-05,
+      "loss": 2.3945,
+      "step": 3139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1889908256880735e-05,
+      "loss": 2.4587,
+      "step": 3140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.188302752293578e-05,
+      "loss": 2.3229,
+      "step": 3141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1876146788990825e-05,
+      "loss": 2.1959,
+      "step": 3142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1869266055045871e-05,
+      "loss": 2.302,
+      "step": 3143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1862385321100918e-05,
+      "loss": 2.4066,
+      "step": 3144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1855504587155964e-05,
+      "loss": 2.3656,
+      "step": 3145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1848623853211009e-05,
+      "loss": 2.3977,
+      "step": 3146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1841743119266055e-05,
+      "loss": 2.3859,
+      "step": 3147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1834862385321102e-05,
+      "loss": 2.3109,
+      "step": 3148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1827981651376147e-05,
+      "loss": 2.3188,
+      "step": 3149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1821100917431193e-05,
+      "loss": 2.1823,
+      "step": 3150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1814220183486238e-05,
+      "loss": 2.1406,
+      "step": 3151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1807339449541284e-05,
+      "loss": 2.0647,
+      "step": 3152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1800458715596331e-05,
+      "loss": 2.2143,
+      "step": 3153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1793577981651377e-05,
+      "loss": 1.9941,
+      "step": 3154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1786697247706422e-05,
+      "loss": 2.1736,
+      "step": 3155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1779816513761469e-05,
+      "loss": 1.9738,
+      "step": 3156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1772935779816513e-05,
+      "loss": 2.1497,
+      "step": 3157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.176605504587156e-05,
+      "loss": 1.9691,
+      "step": 3158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1759174311926605e-05,
+      "loss": 1.8565,
+      "step": 3159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1752293577981651e-05,
+      "loss": 1.9477,
+      "step": 3160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1745412844036698e-05,
+      "loss": 1.9139,
+      "step": 3161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1738532110091744e-05,
+      "loss": 1.7214,
+      "step": 3162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.173165137614679e-05,
+      "loss": 1.776,
+      "step": 3163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1724770642201835e-05,
+      "loss": 1.6602,
+      "step": 3164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.171788990825688e-05,
+      "loss": 1.8281,
+      "step": 3165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1711009174311926e-05,
+      "loss": 1.6933,
+      "step": 3166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1704128440366973e-05,
+      "loss": 1.4674,
+      "step": 3167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1697247706422018e-05,
+      "loss": 1.4204,
+      "step": 3168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1690366972477064e-05,
+      "loss": 1.3342,
+      "step": 3169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.168348623853211e-05,
+      "loss": 1.3162,
+      "step": 3170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1676605504587157e-05,
+      "loss": 3.2124,
+      "step": 3171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1669724770642204e-05,
+      "loss": 2.7789,
+      "step": 3172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1662844036697247e-05,
+      "loss": 2.8417,
+      "step": 3173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1655963302752293e-05,
+      "loss": 2.6451,
+      "step": 3174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.164908256880734e-05,
+      "loss": 2.3678,
+      "step": 3175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1642201834862386e-05,
+      "loss": 2.4429,
+      "step": 3176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.163532110091743e-05,
+      "loss": 2.576,
+      "step": 3177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1628440366972477e-05,
+      "loss": 2.4722,
+      "step": 3178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1621559633027524e-05,
+      "loss": 2.4875,
+      "step": 3179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.161467889908257e-05,
+      "loss": 2.3907,
+      "step": 3180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1607798165137615e-05,
+      "loss": 2.4459,
+      "step": 3181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.160091743119266e-05,
+      "loss": 2.4951,
+      "step": 3182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.1594036697247706e-05,
+      "loss": 2.4547,
+      "step": 3183
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1587155963302753e-05,
+      "loss": 2.5021,
+      "step": 3184
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1580275229357799e-05,
+      "loss": 2.4448,
+      "step": 3185
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1573394495412844e-05,
+      "loss": 2.5216,
+      "step": 3186
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.156651376146789e-05,
+      "loss": 2.4565,
+      "step": 3187
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1559633027522937e-05,
+      "loss": 2.3653,
+      "step": 3188
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1552752293577982e-05,
+      "loss": 2.4444,
+      "step": 3189
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1545871559633026e-05,
+      "loss": 2.3455,
+      "step": 3190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1538990825688073e-05,
+      "loss": 2.3982,
+      "step": 3191
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.153211009174312e-05,
+      "loss": 2.3281,
+      "step": 3192
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1525229357798166e-05,
+      "loss": 2.224,
+      "step": 3193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1518348623853212e-05,
+      "loss": 2.2749,
+      "step": 3194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1511467889908257e-05,
+      "loss": 2.2976,
+      "step": 3195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1504587155963303e-05,
+      "loss": 2.3294,
+      "step": 3196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1497706422018348e-05,
+      "loss": 2.307,
+      "step": 3197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.1490825688073395e-05,
+      "loss": 2.1607,
+      "step": 3198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.148394495412844e-05,
+      "loss": 2.2211,
+      "step": 3199
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1477064220183486e-05,
+      "loss": 2.0939,
+      "step": 3200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1470183486238532e-05,
+      "loss": 2.2334,
+      "step": 3201
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1463302752293579e-05,
+      "loss": 2.0565,
+      "step": 3202
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1456422018348625e-05,
+      "loss": 2.1112,
+      "step": 3203
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.144954128440367e-05,
+      "loss": 2.1934,
+      "step": 3204
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1442660550458715e-05,
+      "loss": 2.0455,
+      "step": 3205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1435779816513761e-05,
+      "loss": 2.0419,
+      "step": 3206
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1428899082568808e-05,
+      "loss": 1.8242,
+      "step": 3207
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1422018348623853e-05,
+      "loss": 1.9201,
+      "step": 3208
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1415137614678899e-05,
+      "loss": 1.8747,
+      "step": 3209
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1408256880733946e-05,
+      "loss": 1.9153,
+      "step": 3210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1401376146788992e-05,
+      "loss": 1.9414,
+      "step": 3211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1394495412844037e-05,
+      "loss": 1.8864,
+      "step": 3212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1387614678899082e-05,
+      "loss": 1.8319,
+      "step": 3213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1380733944954128e-05,
+      "loss": 1.7014,
+      "step": 3214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.1373853211009174e-05,
+      "loss": 1.5888,
+      "step": 3215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1366972477064221e-05,
+      "loss": 1.6771,
+      "step": 3216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1360091743119266e-05,
+      "loss": 1.5093,
+      "step": 3217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1353211009174312e-05,
+      "loss": 1.5035,
+      "step": 3218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1346330275229359e-05,
+      "loss": 1.5151,
+      "step": 3219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1339449541284405e-05,
+      "loss": 1.2909,
+      "step": 3220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1332568807339448e-05,
+      "loss": 3.0641,
+      "step": 3221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1325688073394495e-05,
+      "loss": 2.6125,
+      "step": 3222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1318807339449541e-05,
+      "loss": 2.4671,
+      "step": 3223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1311926605504588e-05,
+      "loss": 2.3907,
+      "step": 3224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1305045871559634e-05,
+      "loss": 2.4169,
+      "step": 3225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1298165137614679e-05,
+      "loss": 2.3145,
+      "step": 3226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1291284403669725e-05,
+      "loss": 2.4442,
+      "step": 3227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1284403669724772e-05,
+      "loss": 2.4531,
+      "step": 3228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1277522935779817e-05,
+      "loss": 2.4119,
+      "step": 3229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1270642201834861e-05,
+      "loss": 2.2431,
+      "step": 3230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1263761467889908e-05,
+      "loss": 2.2835,
+      "step": 3231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1256880733944954e-05,
+      "loss": 2.2747,
+      "step": 3232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.125e-05,
+      "loss": 2.2219,
+      "step": 3233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1243119266055047e-05,
+      "loss": 2.1712,
+      "step": 3234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1236238532110092e-05,
+      "loss": 2.0235,
+      "step": 3235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1229357798165138e-05,
+      "loss": 1.9001,
+      "step": 3236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1222477064220185e-05,
+      "loss": 1.9141,
+      "step": 3237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.121559633027523e-05,
+      "loss": 1.7496,
+      "step": 3238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1208715596330274e-05,
+      "loss": 1.6354,
+      "step": 3239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1201834862385321e-05,
+      "loss": 1.3914,
+      "step": 3240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1194954128440367e-05,
+      "loss": 3.1256,
+      "step": 3241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1188073394495414e-05,
+      "loss": 2.8295,
+      "step": 3242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1181192660550459e-05,
+      "loss": 2.5188,
+      "step": 3243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1174311926605505e-05,
+      "loss": 2.5623,
+      "step": 3244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1167431192660552e-05,
+      "loss": 2.4477,
+      "step": 3245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1160550458715596e-05,
+      "loss": 2.3475,
+      "step": 3246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1153669724770643e-05,
+      "loss": 2.3751,
+      "step": 3247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1146788990825688e-05,
+      "loss": 2.271,
+      "step": 3248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1139908256880734e-05,
+      "loss": 2.3194,
+      "step": 3249
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.113302752293578e-05,
+      "loss": 2.3002,
+      "step": 3250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1126146788990827e-05,
+      "loss": 2.1731,
+      "step": 3251
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1119266055045872e-05,
+      "loss": 2.415,
+      "step": 3252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1112385321100918e-05,
+      "loss": 2.4325,
+      "step": 3253
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1105504587155963e-05,
+      "loss": 2.258,
+      "step": 3254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.109862385321101e-05,
+      "loss": 2.272,
+      "step": 3255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1091743119266056e-05,
+      "loss": 2.3617,
+      "step": 3256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.10848623853211e-05,
+      "loss": 2.2998,
+      "step": 3257
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1077981651376147e-05,
+      "loss": 2.2731,
+      "step": 3258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1071100917431194e-05,
+      "loss": 2.2212,
+      "step": 3259
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.106422018348624e-05,
+      "loss": 2.3501,
+      "step": 3260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1057339449541285e-05,
+      "loss": 2.2205,
+      "step": 3261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.105045871559633e-05,
+      "loss": 2.2342,
+      "step": 3262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1043577981651376e-05,
+      "loss": 2.1092,
+      "step": 3263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1036697247706423e-05,
+      "loss": 2.2034,
+      "step": 3264
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1029816513761469e-05,
+      "loss": 2.0088,
+      "step": 3265
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1022935779816514e-05,
+      "loss": 2.1455,
+      "step": 3266
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.101605504587156e-05,
+      "loss": 2.0286,
+      "step": 3267
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1009174311926607e-05,
+      "loss": 1.9658,
+      "step": 3268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1002293577981653e-05,
+      "loss": 2.0462,
+      "step": 3269
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0995412844036696e-05,
+      "loss": 2.0821,
+      "step": 3270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0988532110091743e-05,
+      "loss": 1.9541,
+      "step": 3271
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.098165137614679e-05,
+      "loss": 1.9241,
+      "step": 3272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0974770642201836e-05,
+      "loss": 2.099,
+      "step": 3273
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.096788990825688e-05,
+      "loss": 1.9575,
+      "step": 3274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0961009174311927e-05,
+      "loss": 1.8803,
+      "step": 3275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0954128440366973e-05,
+      "loss": 2.0764,
+      "step": 3276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.094724770642202e-05,
+      "loss": 1.9026,
+      "step": 3277
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0940366972477065e-05,
+      "loss": 1.759,
+      "step": 3278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.093348623853211e-05,
+      "loss": 1.7704,
+      "step": 3279
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0926605504587156e-05,
+      "loss": 1.8608,
+      "step": 3280
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0919724770642202e-05,
+      "loss": 1.8171,
+      "step": 3281
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0912844036697249e-05,
+      "loss": 1.6767,
+      "step": 3282
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0905963302752294e-05,
+      "loss": 1.7229,
+      "step": 3283
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.089908256880734e-05,
+      "loss": 1.5844,
+      "step": 3284
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0892201834862386e-05,
+      "loss": 1.5684,
+      "step": 3285
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0885321100917431e-05,
+      "loss": 1.6187,
+      "step": 3286
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0878440366972478e-05,
+      "loss": 1.4295,
+      "step": 3287
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0871559633027522e-05,
+      "loss": 1.2962,
+      "step": 3288
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0864678899082569e-05,
+      "loss": 1.1982,
+      "step": 3289
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0857798165137615e-05,
+      "loss": 1.2451,
+      "step": 3290
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0850917431192662e-05,
+      "loss": 3.0011,
+      "step": 3291
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0844036697247707e-05,
+      "loss": 2.8198,
+      "step": 3292
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0837155963302753e-05,
+      "loss": 2.5933,
+      "step": 3293
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0830275229357798e-05,
+      "loss": 2.5188,
+      "step": 3294
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0823394495412844e-05,
+      "loss": 2.4129,
+      "step": 3295
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.081651376146789e-05,
+      "loss": 2.3705,
+      "step": 3296
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0809633027522936e-05,
+      "loss": 2.3477,
+      "step": 3297
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0802752293577982e-05,
+      "loss": 2.5704,
+      "step": 3298
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0795871559633028e-05,
+      "loss": 2.4764,
+      "step": 3299
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0788990825688075e-05,
+      "loss": 2.2785,
+      "step": 3300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.078211009174312e-05,
+      "loss": 2.2584,
+      "step": 3301
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0775229357798165e-05,
+      "loss": 2.3021,
+      "step": 3302
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0768348623853211e-05,
+      "loss": 2.2802,
+      "step": 3303
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0761467889908257e-05,
+      "loss": 2.288,
+      "step": 3304
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0754587155963302e-05,
+      "loss": 2.2469,
+      "step": 3305
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0747706422018349e-05,
+      "loss": 2.2904,
+      "step": 3306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0740825688073395e-05,
+      "loss": 2.1779,
+      "step": 3307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0733944954128442e-05,
+      "loss": 2.2511,
+      "step": 3308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0727064220183488e-05,
+      "loss": 2.1683,
+      "step": 3309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0720183486238531e-05,
+      "loss": 2.2021,
+      "step": 3310
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0713302752293578e-05,
+      "loss": 2.1805,
+      "step": 3311
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0706422018348624e-05,
+      "loss": 2.1272,
+      "step": 3312
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.069954128440367e-05,
+      "loss": 2.1856,
+      "step": 3313
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0692660550458715e-05,
+      "loss": 2.1221,
+      "step": 3314
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0685779816513762e-05,
+      "loss": 2.1791,
+      "step": 3315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0678899082568808e-05,
+      "loss": 2.1448,
+      "step": 3316
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0672018348623855e-05,
+      "loss": 2.0765,
+      "step": 3317
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.06651376146789e-05,
+      "loss": 2.1325,
+      "step": 3318
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0658256880733944e-05,
+      "loss": 1.9862,
+      "step": 3319
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.065137614678899e-05,
+      "loss": 2.0747,
+      "step": 3320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0644495412844037e-05,
+      "loss": 2.094,
+      "step": 3321
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0637614678899084e-05,
+      "loss": 2.002,
+      "step": 3322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0630733944954128e-05,
+      "loss": 2.0404,
+      "step": 3323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0623853211009175e-05,
+      "loss": 1.8783,
+      "step": 3324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0616972477064221e-05,
+      "loss": 1.9287,
+      "step": 3325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0610091743119266e-05,
+      "loss": 1.8173,
+      "step": 3326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0603211009174311e-05,
+      "loss": 1.8406,
+      "step": 3327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0596330275229357e-05,
+      "loss": 1.8763,
+      "step": 3328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0589449541284404e-05,
+      "loss": 1.7558,
+      "step": 3329
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.058256880733945e-05,
+      "loss": 1.7525,
+      "step": 3330
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0575688073394497e-05,
+      "loss": 1.7759,
+      "step": 3331
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0568807339449542e-05,
+      "loss": 1.6849,
+      "step": 3332
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0561926605504588e-05,
+      "loss": 1.4599,
+      "step": 3333
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0555045871559633e-05,
+      "loss": 1.7071,
+      "step": 3334
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.054816513761468e-05,
+      "loss": 1.5825,
+      "step": 3335
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0541284403669724e-05,
+      "loss": 1.4006,
+      "step": 3336
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.053440366972477e-05,
+      "loss": 1.4868,
+      "step": 3337
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0527522935779817e-05,
+      "loss": 1.4068,
+      "step": 3338
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0520642201834863e-05,
+      "loss": 1.2614,
+      "step": 3339
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.051376146788991e-05,
+      "loss": 1.2635,
+      "step": 3340
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0506880733944955e-05,
+      "loss": 3.0229,
+      "step": 3341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.05e-05,
+      "loss": 2.8543,
+      "step": 3342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0493119266055046e-05,
+      "loss": 2.5911,
+      "step": 3343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0486238532110092e-05,
+      "loss": 2.4458,
+      "step": 3344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0479357798165137e-05,
+      "loss": 2.4512,
+      "step": 3345
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0472477064220184e-05,
+      "loss": 2.2889,
+      "step": 3346
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.046559633027523e-05,
+      "loss": 2.4551,
+      "step": 3347
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0458715596330277e-05,
+      "loss": 2.5095,
+      "step": 3348
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0451834862385323e-05,
+      "loss": 2.3216,
+      "step": 3349
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0444954128440366e-05,
+      "loss": 2.2863,
+      "step": 3350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0438073394495413e-05,
+      "loss": 2.4638,
+      "step": 3351
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0431192660550459e-05,
+      "loss": 2.3108,
+      "step": 3352
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0424311926605505e-05,
+      "loss": 2.2415,
+      "step": 3353
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.041743119266055e-05,
+      "loss": 2.359,
+      "step": 3354
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0410550458715597e-05,
+      "loss": 2.4628,
+      "step": 3355
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0403669724770643e-05,
+      "loss": 2.2421,
+      "step": 3356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.039678899082569e-05,
+      "loss": 2.3777,
+      "step": 3357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0389908256880733e-05,
+      "loss": 2.3123,
+      "step": 3358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.038302752293578e-05,
+      "loss": 2.2696,
+      "step": 3359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0376146788990826e-05,
+      "loss": 2.1682,
+      "step": 3360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.0369266055045872e-05,
+      "loss": 2.2232,
+      "step": 3361
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0362385321100919e-05,
+      "loss": 2.1194,
+      "step": 3362
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0355504587155963e-05,
+      "loss": 2.2308,
+      "step": 3363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.034862385321101e-05,
+      "loss": 2.168,
+      "step": 3364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0341743119266056e-05,
+      "loss": 2.0516,
+      "step": 3365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0334862385321101e-05,
+      "loss": 2.1131,
+      "step": 3366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0327981651376146e-05,
+      "loss": 2.2196,
+      "step": 3367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0321100917431192e-05,
+      "loss": 2.0732,
+      "step": 3368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0314220183486239e-05,
+      "loss": 2.0074,
+      "step": 3369
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0307339449541285e-05,
+      "loss": 2.0208,
+      "step": 3370
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0300458715596332e-05,
+      "loss": 1.8328,
+      "step": 3371
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0293577981651376e-05,
+      "loss": 2.1513,
+      "step": 3372
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0286697247706423e-05,
+      "loss": 1.8687,
+      "step": 3373
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0279816513761468e-05,
+      "loss": 1.8634,
+      "step": 3374
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0272935779816514e-05,
+      "loss": 2.0696,
+      "step": 3375
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0266055045871559e-05,
+      "loss": 1.9081,
+      "step": 3376
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.0259174311926605e-05,
+      "loss": 1.846,
+      "step": 3377
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0252293577981652e-05,
+      "loss": 1.8858,
+      "step": 3378
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0245412844036698e-05,
+      "loss": 1.818,
+      "step": 3379
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0238532110091745e-05,
+      "loss": 1.525,
+      "step": 3380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.023165137614679e-05,
+      "loss": 1.837,
+      "step": 3381
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0224770642201834e-05,
+      "loss": 1.701,
+      "step": 3382
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.021788990825688e-05,
+      "loss": 1.5887,
+      "step": 3383
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0211009174311927e-05,
+      "loss": 1.4742,
+      "step": 3384
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0204128440366972e-05,
+      "loss": 1.5631,
+      "step": 3385
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0197247706422019e-05,
+      "loss": 1.3929,
+      "step": 3386
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0190366972477065e-05,
+      "loss": 1.3612,
+      "step": 3387
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0183486238532111e-05,
+      "loss": 1.3227,
+      "step": 3388
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0176605504587156e-05,
+      "loss": 1.4041,
+      "step": 3389
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0169724770642201e-05,
+      "loss": 1.359,
+      "step": 3390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0162844036697247e-05,
+      "loss": 2.9628,
+      "step": 3391
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.0155963302752294e-05,
+      "loss": 2.6095,
+      "step": 3392
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.014908256880734e-05,
+      "loss": 2.6958,
+      "step": 3393
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0142201834862385e-05,
+      "loss": 2.5665,
+      "step": 3394
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0135321100917432e-05,
+      "loss": 2.4665,
+      "step": 3395
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0128440366972478e-05,
+      "loss": 2.3139,
+      "step": 3396
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0121559633027525e-05,
+      "loss": 2.4624,
+      "step": 3397
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0114678899082568e-05,
+      "loss": 2.5064,
+      "step": 3398
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0107798165137614e-05,
+      "loss": 2.4788,
+      "step": 3399
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.010091743119266e-05,
+      "loss": 2.2793,
+      "step": 3400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0094036697247707e-05,
+      "loss": 2.4239,
+      "step": 3401
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0087155963302753e-05,
+      "loss": 2.3329,
+      "step": 3402
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0080275229357798e-05,
+      "loss": 2.198,
+      "step": 3403
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0073394495412845e-05,
+      "loss": 2.3393,
+      "step": 3404
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0066513761467891e-05,
+      "loss": 2.2891,
+      "step": 3405
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0059633027522936e-05,
+      "loss": 2.2639,
+      "step": 3406
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.005275229357798e-05,
+      "loss": 2.1493,
+      "step": 3407
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0045871559633027e-05,
+      "loss": 2.3716,
+      "step": 3408
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.0038990825688074e-05,
+      "loss": 2.2131,
+      "step": 3409
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.003211009174312e-05,
+      "loss": 2.2718,
+      "step": 3410
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0025229357798165e-05,
+      "loss": 2.2526,
+      "step": 3411
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0018348623853211e-05,
+      "loss": 2.2411,
+      "step": 3412
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0011467889908258e-05,
+      "loss": 2.0938,
+      "step": 3413
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.0004587155963303e-05,
+      "loss": 2.1298,
+      "step": 3414
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.997706422018349e-06,
+      "loss": 2.1759,
+      "step": 3415
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.990825688073394e-06,
+      "loss": 2.0968,
+      "step": 3416
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.98394495412844e-06,
+      "loss": 2.1612,
+      "step": 3417
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.977064220183487e-06,
+      "loss": 2.0378,
+      "step": 3418
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.970183486238533e-06,
+      "loss": 2.1172,
+      "step": 3419
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.963302752293578e-06,
+      "loss": 2.0564,
+      "step": 3420
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.956422018348624e-06,
+      "loss": 1.9752,
+      "step": 3421
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.94954128440367e-06,
+      "loss": 1.942,
+      "step": 3422
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.942660550458716e-06,
+      "loss": 1.9759,
+      "step": 3423
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.935779816513762e-06,
+      "loss": 2.11,
+      "step": 3424
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.928899082568807e-06,
+      "loss": 1.9355,
+      "step": 3425
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.922018348623853e-06,
+      "loss": 2.1101,
+      "step": 3426
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.9151376146789e-06,
+      "loss": 1.8355,
+      "step": 3427
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.908256880733946e-06,
+      "loss": 1.912,
+      "step": 3428
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.901376146788991e-06,
+      "loss": 1.7714,
+      "step": 3429
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.894495412844036e-06,
+      "loss": 1.7243,
+      "step": 3430
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.887614678899082e-06,
+      "loss": 1.6834,
+      "step": 3431
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.880733944954129e-06,
+      "loss": 1.5752,
+      "step": 3432
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.873853211009175e-06,
+      "loss": 1.7205,
+      "step": 3433
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.86697247706422e-06,
+      "loss": 1.5559,
+      "step": 3434
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.860091743119267e-06,
+      "loss": 1.5596,
+      "step": 3435
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.853211009174313e-06,
+      "loss": 1.5253,
+      "step": 3436
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.84633027522936e-06,
+      "loss": 1.4814,
+      "step": 3437
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.839449541284403e-06,
+      "loss": 1.3923,
+      "step": 3438
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.832568807339449e-06,
+      "loss": 1.3059,
+      "step": 3439
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.825688073394495e-06,
+      "loss": 1.3109,
+      "step": 3440
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.818807339449542e-06,
+      "loss": 2.9207,
+      "step": 3441
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.811926605504587e-06,
+      "loss": 2.795,
+      "step": 3442
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.805045871559633e-06,
+      "loss": 2.6135,
+      "step": 3443
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.79816513761468e-06,
+      "loss": 2.2935,
+      "step": 3444
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.791284403669726e-06,
+      "loss": 2.3207,
+      "step": 3445
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.784403669724771e-06,
+      "loss": 2.2549,
+      "step": 3446
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.777522935779816e-06,
+      "loss": 2.4528,
+      "step": 3447
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.770642201834862e-06,
+      "loss": 2.3031,
+      "step": 3448
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.763761467889909e-06,
+      "loss": 2.3145,
+      "step": 3449
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.756880733944955e-06,
+      "loss": 2.2353,
+      "step": 3450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.75e-06,
+      "loss": 2.3089,
+      "step": 3451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.743119266055046e-06,
+      "loss": 2.5557,
+      "step": 3452
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.736238532110093e-06,
+      "loss": 2.259,
+      "step": 3453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.729357798165138e-06,
+      "loss": 2.2759,
+      "step": 3454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.722477064220184e-06,
+      "loss": 2.1845,
+      "step": 3455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.715596330275229e-06,
+      "loss": 2.1287,
+      "step": 3456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.708715596330275e-06,
+      "loss": 2.3244,
+      "step": 3457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.701834862385322e-06,
+      "loss": 2.2781,
+      "step": 3458
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.694954128440368e-06,
+      "loss": 2.1327,
+      "step": 3459
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.688073394495413e-06,
+      "loss": 2.255,
+      "step": 3460
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.68119266055046e-06,
+      "loss": 2.2241,
+      "step": 3461
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.674311926605504e-06,
+      "loss": 2.1255,
+      "step": 3462
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.66743119266055e-06,
+      "loss": 2.1333,
+      "step": 3463
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.660550458715597e-06,
+      "loss": 2.0966,
+      "step": 3464
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.653669724770642e-06,
+      "loss": 2.1606,
+      "step": 3465
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.646788990825688e-06,
+      "loss": 2.0207,
+      "step": 3466
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.639908256880735e-06,
+      "loss": 2.1007,
+      "step": 3467
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.633027522935781e-06,
+      "loss": 2.0752,
+      "step": 3468
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.626146788990826e-06,
+      "loss": 2.1167,
+      "step": 3469
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.61926605504587e-06,
+      "loss": 1.9321,
+      "step": 3470
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.612385321100917e-06,
+      "loss": 1.9709,
+      "step": 3471
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.605504587155964e-06,
+      "loss": 1.9851,
+      "step": 3472
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.598623853211009e-06,
+      "loss": 2.0521,
+      "step": 3473
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.591743119266055e-06,
+      "loss": 1.8831,
+      "step": 3474
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.584862385321101e-06,
+      "loss": 1.7811,
+      "step": 3475
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.577981651376148e-06,
+      "loss": 1.7768,
+      "step": 3476
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.571100917431194e-06,
+      "loss": 1.8026,
+      "step": 3477
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.564220183486237e-06,
+      "loss": 1.9444,
+      "step": 3478
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.557339449541284e-06,
+      "loss": 1.7585,
+      "step": 3479
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.55045871559633e-06,
+      "loss": 1.8929,
+      "step": 3480
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.543577981651377e-06,
+      "loss": 1.7608,
+      "step": 3481
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.536697247706422e-06,
+      "loss": 1.7841,
+      "step": 3482
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.529816513761468e-06,
+      "loss": 1.6615,
+      "step": 3483
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.522935779816515e-06,
+      "loss": 1.609,
+      "step": 3484
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.516055045871561e-06,
+      "loss": 1.5291,
+      "step": 3485
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.509174311926606e-06,
+      "loss": 1.4577,
+      "step": 3486
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.50229357798165e-06,
+      "loss": 1.507,
+      "step": 3487
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.495412844036697e-06,
+      "loss": 1.3404,
+      "step": 3488
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.488532110091744e-06,
+      "loss": 1.3275,
+      "step": 3489
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.48165137614679e-06,
+      "loss": 1.4142,
+      "step": 3490
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.474770642201835e-06,
+      "loss": 2.9065,
+      "step": 3491
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.467889908256881e-06,
+      "loss": 2.7592,
+      "step": 3492
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.461009174311928e-06,
+      "loss": 2.5632,
+      "step": 3493
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.454128440366972e-06,
+      "loss": 2.3673,
+      "step": 3494
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.447247706422019e-06,
+      "loss": 2.3029,
+      "step": 3495
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.440366972477064e-06,
+      "loss": 2.2673,
+      "step": 3496
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.43348623853211e-06,
+      "loss": 2.441,
+      "step": 3497
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.426605504587157e-06,
+      "loss": 2.2615,
+      "step": 3498
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.419724770642203e-06,
+      "loss": 2.1983,
+      "step": 3499
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.412844036697248e-06,
+      "loss": 2.3435,
+      "step": 3500
+    },
+    {
+      "epoch": 2.16,
+      "eval_bleu": 5.858561653487833e-14,
+      "eval_loss": 2.26958966255188,
+      "eval_runtime": 2627.4744,
+      "eval_samples_per_second": 5.618,
+      "eval_steps_per_second": 0.702,
+      "step": 3500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.405963302752294e-06,
+      "loss": 2.265,
+      "step": 3501
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.399082568807339e-06,
+      "loss": 2.3039,
+      "step": 3502
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.392201834862386e-06,
+      "loss": 2.4795,
+      "step": 3503
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.38532110091743e-06,
+      "loss": 2.2107,
+      "step": 3504
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.378440366972477e-06,
+      "loss": 2.2518,
+      "step": 3505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.371559633027523e-06,
+      "loss": 2.2298,
+      "step": 3506
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.36467889908257e-06,
+      "loss": 2.2162,
+      "step": 3507
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.357798165137616e-06,
+      "loss": 2.0956,
+      "step": 3508
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.350917431192661e-06,
+      "loss": 2.2544,
+      "step": 3509
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.344036697247706e-06,
+      "loss": 2.1913,
+      "step": 3510
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.337155963302752e-06,
+      "loss": 2.0782,
+      "step": 3511
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.330275229357799e-06,
+      "loss": 2.112,
+      "step": 3512
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.323394495412843e-06,
+      "loss": 2.1582,
+      "step": 3513
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.31651376146789e-06,
+      "loss": 2.2632,
+      "step": 3514
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.309633027522936e-06,
+      "loss": 2.0077,
+      "step": 3515
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.302752293577983e-06,
+      "loss": 2.0639,
+      "step": 3516
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.29587155963303e-06,
+      "loss": 2.1485,
+      "step": 3517
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.288990825688072e-06,
+      "loss": 1.9105,
+      "step": 3518
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.282110091743119e-06,
+      "loss": 2.0165,
+      "step": 3519
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.275229357798165e-06,
+      "loss": 2.1499,
+      "step": 3520
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.268348623853212e-06,
+      "loss": 1.867,
+      "step": 3521
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.261467889908257e-06,
+      "loss": 2.0569,
+      "step": 3522
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.254587155963303e-06,
+      "loss": 1.8038,
+      "step": 3523
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.24770642201835e-06,
+      "loss": 1.8929,
+      "step": 3524
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.240825688073396e-06,
+      "loss": 1.9588,
+      "step": 3525
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.233944954128439e-06,
+      "loss": 1.8703,
+      "step": 3526
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.227064220183485e-06,
+      "loss": 1.722,
+      "step": 3527
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.220183486238532e-06,
+      "loss": 1.7365,
+      "step": 3528
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.213302752293578e-06,
+      "loss": 1.7814,
+      "step": 3529
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.206422018348625e-06,
+      "loss": 1.6683,
+      "step": 3530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.19954128440367e-06,
+      "loss": 1.6491,
+      "step": 3531
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.192660550458716e-06,
+      "loss": 1.7436,
+      "step": 3532
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.185779816513763e-06,
+      "loss": 1.5906,
+      "step": 3533
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.178899082568807e-06,
+      "loss": 1.6304,
+      "step": 3534
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.172018348623852e-06,
+      "loss": 1.6185,
+      "step": 3535
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.165137614678899e-06,
+      "loss": 1.4182,
+      "step": 3536
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.158256880733945e-06,
+      "loss": 1.4423,
+      "step": 3537
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.151376146788992e-06,
+      "loss": 1.3085,
+      "step": 3538
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.144495412844038e-06,
+      "loss": 1.3997,
+      "step": 3539
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.137614678899083e-06,
+      "loss": 1.3658,
+      "step": 3540
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.13073394495413e-06,
+      "loss": 2.8403,
+      "step": 3541
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.123853211009174e-06,
+      "loss": 2.7399,
+      "step": 3542
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.11697247706422e-06,
+      "loss": 2.44,
+      "step": 3543
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.110091743119265e-06,
+      "loss": 2.6152,
+      "step": 3544
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.103211009174312e-06,
+      "loss": 2.4269,
+      "step": 3545
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.096330275229358e-06,
+      "loss": 2.2306,
+      "step": 3546
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.089449541284405e-06,
+      "loss": 2.2947,
+      "step": 3547
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.082568807339451e-06,
+      "loss": 2.3366,
+      "step": 3548
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.075688073394496e-06,
+      "loss": 2.2719,
+      "step": 3549
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.06880733944954e-06,
+      "loss": 2.3677,
+      "step": 3550
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.061926605504587e-06,
+      "loss": 2.2733,
+      "step": 3551
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.055045871559634e-06,
+      "loss": 2.277,
+      "step": 3552
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.048165137614678e-06,
+      "loss": 2.4225,
+      "step": 3553
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.041284403669725e-06,
+      "loss": 2.0818,
+      "step": 3554
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.034403669724771e-06,
+      "loss": 2.2529,
+      "step": 3555
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.027522935779818e-06,
+      "loss": 2.1871,
+      "step": 3556
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.020642201834863e-06,
+      "loss": 2.2238,
+      "step": 3557
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.013761467889907e-06,
+      "loss": 2.1433,
+      "step": 3558
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.006880733944954e-06,
+      "loss": 2.0797,
+      "step": 3559
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9e-06,
+      "loss": 2.3583,
+      "step": 3560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.993119266055047e-06,
+      "loss": 2.385,
+      "step": 3561
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.986238532110091e-06,
+      "loss": 2.2546,
+      "step": 3562
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.979357798165138e-06,
+      "loss": 2.078,
+      "step": 3563
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.972477064220184e-06,
+      "loss": 2.2531,
+      "step": 3564
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.965596330275231e-06,
+      "loss": 2.0859,
+      "step": 3565
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.958715596330274e-06,
+      "loss": 2.1305,
+      "step": 3566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.95183486238532e-06,
+      "loss": 2.0597,
+      "step": 3567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.944954128440367e-06,
+      "loss": 2.0801,
+      "step": 3568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.938073394495413e-06,
+      "loss": 1.9775,
+      "step": 3569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.93119266055046e-06,
+      "loss": 1.885,
+      "step": 3570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.924311926605505e-06,
+      "loss": 1.888,
+      "step": 3571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.917431192660551e-06,
+      "loss": 1.8204,
+      "step": 3572
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.910550458715598e-06,
+      "loss": 2.0922,
+      "step": 3573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.903669724770642e-06,
+      "loss": 1.8709,
+      "step": 3574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.896788990825687e-06,
+      "loss": 1.891,
+      "step": 3575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.889908256880734e-06,
+      "loss": 1.8748,
+      "step": 3576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.88302752293578e-06,
+      "loss": 1.801,
+      "step": 3577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.876146788990826e-06,
+      "loss": 1.8167,
+      "step": 3578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.869266055045873e-06,
+      "loss": 1.8623,
+      "step": 3579
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.862385321100918e-06,
+      "loss": 1.7622,
+      "step": 3580
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.855504587155964e-06,
+      "loss": 1.869,
+      "step": 3581
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.848623853211009e-06,
+      "loss": 1.6647,
+      "step": 3582
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.841743119266055e-06,
+      "loss": 1.5536,
+      "step": 3583
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.8348623853211e-06,
+      "loss": 1.6789,
+      "step": 3584
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.827981651376147e-06,
+      "loss": 1.6513,
+      "step": 3585
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.821100917431193e-06,
+      "loss": 1.7243,
+      "step": 3586
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.81422018348624e-06,
+      "loss": 1.3369,
+      "step": 3587
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.807339449541284e-06,
+      "loss": 1.3385,
+      "step": 3588
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.80045871559633e-06,
+      "loss": 1.3247,
+      "step": 3589
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.793577981651376e-06,
+      "loss": 1.2549,
+      "step": 3590
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.786697247706422e-06,
+      "loss": 2.8131,
+      "step": 3591
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.779816513761469e-06,
+      "loss": 2.6642,
+      "step": 3592
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.772935779816513e-06,
+      "loss": 2.5129,
+      "step": 3593
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.76605504587156e-06,
+      "loss": 2.4579,
+      "step": 3594
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.759174311926606e-06,
+      "loss": 2.2337,
+      "step": 3595
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.752293577981653e-06,
+      "loss": 2.1635,
+      "step": 3596
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.745412844036697e-06,
+      "loss": 2.3752,
+      "step": 3597
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.738532110091742e-06,
+      "loss": 2.3486,
+      "step": 3598
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.731651376146789e-06,
+      "loss": 2.0784,
+      "step": 3599
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.724770642201835e-06,
+      "loss": 2.2305,
+      "step": 3600
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.717889908256882e-06,
+      "loss": 2.1722,
+      "step": 3601
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.711009174311926e-06,
+      "loss": 2.2234,
+      "step": 3602
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.704128440366973e-06,
+      "loss": 2.1683,
+      "step": 3603
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.69724770642202e-06,
+      "loss": 2.1483,
+      "step": 3604
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.690366972477066e-06,
+      "loss": 2.2606,
+      "step": 3605
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.683486238532109e-06,
+      "loss": 2.3378,
+      "step": 3606
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.676605504587155e-06,
+      "loss": 2.2287,
+      "step": 3607
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.669724770642202e-06,
+      "loss": 2.1455,
+      "step": 3608
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.662844036697248e-06,
+      "loss": 2.1633,
+      "step": 3609
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.655963302752293e-06,
+      "loss": 2.1805,
+      "step": 3610
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.64908256880734e-06,
+      "loss": 2.1844,
+      "step": 3611
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.642201834862386e-06,
+      "loss": 2.0896,
+      "step": 3612
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.635321100917432e-06,
+      "loss": 2.178,
+      "step": 3613
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.628440366972477e-06,
+      "loss": 2.1183,
+      "step": 3614
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.621559633027522e-06,
+      "loss": 2.0073,
+      "step": 3615
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.614678899082568e-06,
+      "loss": 1.9078,
+      "step": 3616
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.607798165137615e-06,
+      "loss": 2.0907,
+      "step": 3617
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.600917431192661e-06,
+      "loss": 2.0563,
+      "step": 3618
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.594036697247706e-06,
+      "loss": 2.0055,
+      "step": 3619
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.587155963302753e-06,
+      "loss": 1.9792,
+      "step": 3620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.580275229357799e-06,
+      "loss": 1.9473,
+      "step": 3621
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.573394495412844e-06,
+      "loss": 1.8126,
+      "step": 3622
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.56651376146789e-06,
+      "loss": 1.9682,
+      "step": 3623
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.559633027522935e-06,
+      "loss": 1.9012,
+      "step": 3624
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.552752293577982e-06,
+      "loss": 1.9515,
+      "step": 3625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.545871559633028e-06,
+      "loss": 1.8848,
+      "step": 3626
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.538990825688074e-06,
+      "loss": 1.9107,
+      "step": 3627
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.53211009174312e-06,
+      "loss": 1.8734,
+      "step": 3628
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.525229357798166e-06,
+      "loss": 1.7819,
+      "step": 3629
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.51834862385321e-06,
+      "loss": 1.7143,
+      "step": 3630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.511467889908257e-06,
+      "loss": 1.6852,
+      "step": 3631
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.504587155963303e-06,
+      "loss": 1.6521,
+      "step": 3632
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.497706422018348e-06,
+      "loss": 1.8228,
+      "step": 3633
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.490825688073395e-06,
+      "loss": 1.5615,
+      "step": 3634
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.483944954128441e-06,
+      "loss": 1.4458,
+      "step": 3635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.477064220183488e-06,
+      "loss": 1.6003,
+      "step": 3636
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.470183486238532e-06,
+      "loss": 1.3845,
+      "step": 3637
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.463302752293577e-06,
+      "loss": 1.3757,
+      "step": 3638
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.456422018348624e-06,
+      "loss": 1.4225,
+      "step": 3639
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.44954128440367e-06,
+      "loss": 1.3632,
+      "step": 3640
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.442660550458715e-06,
+      "loss": 2.9144,
+      "step": 3641
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.435779816513761e-06,
+      "loss": 2.6226,
+      "step": 3642
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.428899082568808e-06,
+      "loss": 2.4828,
+      "step": 3643
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.422018348623854e-06,
+      "loss": 2.4752,
+      "step": 3644
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.4151376146789e-06,
+      "loss": 2.2431,
+      "step": 3645
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.408256880733944e-06,
+      "loss": 2.3194,
+      "step": 3646
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.40137614678899e-06,
+      "loss": 2.3584,
+      "step": 3647
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.394495412844037e-06,
+      "loss": 2.3616,
+      "step": 3648
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.387614678899083e-06,
+      "loss": 2.1005,
+      "step": 3649
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.380733944954128e-06,
+      "loss": 2.4735,
+      "step": 3650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.373853211009174e-06,
+      "loss": 2.1343,
+      "step": 3651
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.366972477064221e-06,
+      "loss": 2.1861,
+      "step": 3652
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.360091743119267e-06,
+      "loss": 2.154,
+      "step": 3653
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.353211009174312e-06,
+      "loss": 2.1652,
+      "step": 3654
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.346330275229357e-06,
+      "loss": 2.3425,
+      "step": 3655
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.339449541284403e-06,
+      "loss": 2.0508,
+      "step": 3656
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.33256880733945e-06,
+      "loss": 2.272,
+      "step": 3657
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.325688073394496e-06,
+      "loss": 2.3063,
+      "step": 3658
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.318807339449541e-06,
+      "loss": 2.132,
+      "step": 3659
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.311926605504588e-06,
+      "loss": 1.9807,
+      "step": 3660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.305045871559634e-06,
+      "loss": 2.1579,
+      "step": 3661
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.298165137614679e-06,
+      "loss": 2.1641,
+      "step": 3662
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.291284403669725e-06,
+      "loss": 2.0924,
+      "step": 3663
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.28440366972477e-06,
+      "loss": 2.1503,
+      "step": 3664
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.277522935779816e-06,
+      "loss": 1.9851,
+      "step": 3665
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.270642201834863e-06,
+      "loss": 2.0278,
+      "step": 3666
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.26376146788991e-06,
+      "loss": 2.0233,
+      "step": 3667
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.256880733944954e-06,
+      "loss": 2.1542,
+      "step": 3668
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.25e-06,
+      "loss": 1.836,
+      "step": 3669
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.243119266055045e-06,
+      "loss": 1.9802,
+      "step": 3670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.236238532110092e-06,
+      "loss": 2.0231,
+      "step": 3671
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.229357798165137e-06,
+      "loss": 2.1048,
+      "step": 3672
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.222477064220183e-06,
+      "loss": 1.8966,
+      "step": 3673
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.21559633027523e-06,
+      "loss": 1.8149,
+      "step": 3674
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.208715596330276e-06,
+      "loss": 1.9043,
+      "step": 3675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.201834862385323e-06,
+      "loss": 1.7416,
+      "step": 3676
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.194954128440367e-06,
+      "loss": 1.8704,
+      "step": 3677
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.188073394495412e-06,
+      "loss": 1.7063,
+      "step": 3678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.181192660550459e-06,
+      "loss": 1.6708,
+      "step": 3679
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.174311926605505e-06,
+      "loss": 1.7678,
+      "step": 3680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.16743119266055e-06,
+      "loss": 1.7652,
+      "step": 3681
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.160550458715596e-06,
+      "loss": 1.6042,
+      "step": 3682
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.153669724770643e-06,
+      "loss": 1.6797,
+      "step": 3683
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.146788990825689e-06,
+      "loss": 1.4896,
+      "step": 3684
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.139908256880736e-06,
+      "loss": 1.588,
+      "step": 3685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.133027522935779e-06,
+      "loss": 1.4055,
+      "step": 3686
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.126146788990825e-06,
+      "loss": 1.2752,
+      "step": 3687
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.119266055045872e-06,
+      "loss": 1.5201,
+      "step": 3688
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.112385321100918e-06,
+      "loss": 1.3068,
+      "step": 3689
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.105504587155963e-06,
+      "loss": 1.1919,
+      "step": 3690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.09862385321101e-06,
+      "loss": 2.8713,
+      "step": 3691
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.091743119266056e-06,
+      "loss": 2.5591,
+      "step": 3692
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.084862385321102e-06,
+      "loss": 2.587,
+      "step": 3693
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.077981651376147e-06,
+      "loss": 2.4601,
+      "step": 3694
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.071100917431192e-06,
+      "loss": 2.4512,
+      "step": 3695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.064220183486238e-06,
+      "loss": 2.3416,
+      "step": 3696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.057339449541285e-06,
+      "loss": 2.1501,
+      "step": 3697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.050458715596331e-06,
+      "loss": 2.2421,
+      "step": 3698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.043577981651376e-06,
+      "loss": 2.2622,
+      "step": 3699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.036697247706422e-06,
+      "loss": 2.306,
+      "step": 3700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.029816513761469e-06,
+      "loss": 2.0953,
+      "step": 3701
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.022935779816514e-06,
+      "loss": 2.346,
+      "step": 3702
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.016055045871558e-06,
+      "loss": 2.0941,
+      "step": 3703
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.009174311926605e-06,
+      "loss": 2.1578,
+      "step": 3704
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.002293577981651e-06,
+      "loss": 2.3471,
+      "step": 3705
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.995412844036698e-06,
+      "loss": 2.3415,
+      "step": 3706
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.988532110091744e-06,
+      "loss": 2.0697,
+      "step": 3707
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.981651376146789e-06,
+      "loss": 2.1927,
+      "step": 3708
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.974770642201836e-06,
+      "loss": 2.1028,
+      "step": 3709
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.96788990825688e-06,
+      "loss": 2.1686,
+      "step": 3710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.961009174311927e-06,
+      "loss": 2.1264,
+      "step": 3711
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.954128440366972e-06,
+      "loss": 2.1968,
+      "step": 3712
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.947247706422018e-06,
+      "loss": 2.1979,
+      "step": 3713
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.940366972477064e-06,
+      "loss": 1.9731,
+      "step": 3714
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.933486238532111e-06,
+      "loss": 2.1323,
+      "step": 3715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.926605504587157e-06,
+      "loss": 2.01,
+      "step": 3716
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.919724770642202e-06,
+      "loss": 2.0812,
+      "step": 3717
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.912844036697247e-06,
+      "loss": 2.01,
+      "step": 3718
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.905963302752293e-06,
+      "loss": 2.012,
+      "step": 3719
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.89908256880734e-06,
+      "loss": 2.0207,
+      "step": 3720
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.892201834862385e-06,
+      "loss": 1.949,
+      "step": 3721
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.885321100917431e-06,
+      "loss": 1.9932,
+      "step": 3722
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.878440366972478e-06,
+      "loss": 1.8815,
+      "step": 3723
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.871559633027524e-06,
+      "loss": 1.8371,
+      "step": 3724
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.864678899082569e-06,
+      "loss": 1.9757,
+      "step": 3725
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.857798165137614e-06,
+      "loss": 1.7592,
+      "step": 3726
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.85091743119266e-06,
+      "loss": 1.7411,
+      "step": 3727
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.844036697247707e-06,
+      "loss": 1.7527,
+      "step": 3728
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.837155963302753e-06,
+      "loss": 1.6641,
+      "step": 3729
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.830275229357798e-06,
+      "loss": 1.7677,
+      "step": 3730
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.823394495412844e-06,
+      "loss": 1.8242,
+      "step": 3731
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.81651376146789e-06,
+      "loss": 1.6428,
+      "step": 3732
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.809633027522937e-06,
+      "loss": 1.6315,
+      "step": 3733
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.80275229357798e-06,
+      "loss": 1.6432,
+      "step": 3734
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.795871559633027e-06,
+      "loss": 1.5415,
+      "step": 3735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.788990825688073e-06,
+      "loss": 1.3869,
+      "step": 3736
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.78211009174312e-06,
+      "loss": 1.4381,
+      "step": 3737
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.775229357798166e-06,
+      "loss": 1.311,
+      "step": 3738
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.768348623853211e-06,
+      "loss": 1.179,
+      "step": 3739
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.761467889908257e-06,
+      "loss": 1.2272,
+      "step": 3740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.754587155963304e-06,
+      "loss": 2.8074,
+      "step": 3741
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.747706422018349e-06,
+      "loss": 2.5189,
+      "step": 3742
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.740825688073393e-06,
+      "loss": 2.5464,
+      "step": 3743
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.73394495412844e-06,
+      "loss": 2.4462,
+      "step": 3744
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.727064220183486e-06,
+      "loss": 2.2254,
+      "step": 3745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.720183486238533e-06,
+      "loss": 2.3226,
+      "step": 3746
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.71330275229358e-06,
+      "loss": 2.1765,
+      "step": 3747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.706422018348624e-06,
+      "loss": 2.2323,
+      "step": 3748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.69954128440367e-06,
+      "loss": 2.1869,
+      "step": 3749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.692660550458715e-06,
+      "loss": 2.1678,
+      "step": 3750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.685779816513762e-06,
+      "loss": 2.307,
+      "step": 3751
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.678899082568806e-06,
+      "loss": 2.4427,
+      "step": 3752
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.672018348623853e-06,
+      "loss": 2.1765,
+      "step": 3753
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.6651376146789e-06,
+      "loss": 2.2032,
+      "step": 3754
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.658256880733946e-06,
+      "loss": 2.0771,
+      "step": 3755
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.65137614678899e-06,
+      "loss": 2.0628,
+      "step": 3756
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.644495412844037e-06,
+      "loss": 2.1664,
+      "step": 3757
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.637614678899082e-06,
+      "loss": 2.2458,
+      "step": 3758
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.630733944954128e-06,
+      "loss": 2.2511,
+      "step": 3759
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.623853211009174e-06,
+      "loss": 2.193,
+      "step": 3760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.6169724770642204e-06,
+      "loss": 2.2641,
+      "step": 3761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.610091743119266e-06,
+      "loss": 2.0358,
+      "step": 3762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.6032110091743125e-06,
+      "loss": 2.1541,
+      "step": 3763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.596330275229358e-06,
+      "loss": 1.9956,
+      "step": 3764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.589449541284405e-06,
+      "loss": 2.1064,
+      "step": 3765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.5825688073394485e-06,
+      "loss": 1.9263,
+      "step": 3766
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.575688073394495e-06,
+      "loss": 2.0448,
+      "step": 3767
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.5688073394495415e-06,
+      "loss": 2.0066,
+      "step": 3768
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.561926605504587e-06,
+      "loss": 1.9481,
+      "step": 3769
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.5550458715596335e-06,
+      "loss": 1.9254,
+      "step": 3770
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.548165137614679e-06,
+      "loss": 2.0125,
+      "step": 3771
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.541284403669726e-06,
+      "loss": 1.9148,
+      "step": 3772
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.534403669724771e-06,
+      "loss": 1.9485,
+      "step": 3773
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.527522935779816e-06,
+      "loss": 1.8848,
+      "step": 3774
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.520642201834862e-06,
+      "loss": 1.9398,
+      "step": 3775
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.513761467889908e-06,
+      "loss": 1.8657,
+      "step": 3776
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.506880733944954e-06,
+      "loss": 1.8876,
+      "step": 3777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.5e-06,
+      "loss": 1.7877,
+      "step": 3778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.493119266055047e-06,
+      "loss": 1.6354,
+      "step": 3779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.486238532110092e-06,
+      "loss": 1.744,
+      "step": 3780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.479357798165138e-06,
+      "loss": 1.6022,
+      "step": 3781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.4724770642201835e-06,
+      "loss": 1.6107,
+      "step": 3782
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.46559633027523e-06,
+      "loss": 1.6886,
+      "step": 3783
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.458715596330276e-06,
+      "loss": 1.572,
+      "step": 3784
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.451834862385321e-06,
+      "loss": 1.4552,
+      "step": 3785
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.444954128440367e-06,
+      "loss": 1.2548,
+      "step": 3786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.438073394495413e-06,
+      "loss": 1.4557,
+      "step": 3787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.431192660550459e-06,
+      "loss": 1.2726,
+      "step": 3788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.4243119266055045e-06,
+      "loss": 1.2974,
+      "step": 3789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.417431192660551e-06,
+      "loss": 1.2768,
+      "step": 3790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.410550458715597e-06,
+      "loss": 2.9198,
+      "step": 3791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.403669724770643e-06,
+      "loss": 2.442,
+      "step": 3792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.396788990825688e-06,
+      "loss": 2.666,
+      "step": 3793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.389908256880734e-06,
+      "loss": 2.2806,
+      "step": 3794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.38302752293578e-06,
+      "loss": 2.2952,
+      "step": 3795
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.376146788990826e-06,
+      "loss": 2.2336,
+      "step": 3796
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.369266055045871e-06,
+      "loss": 2.288,
+      "step": 3797
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.362385321100918e-06,
+      "loss": 2.4156,
+      "step": 3798
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.355504587155963e-06,
+      "loss": 2.2104,
+      "step": 3799
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.34862385321101e-06,
+      "loss": 2.1696,
+      "step": 3800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.341743119266055e-06,
+      "loss": 2.1827,
+      "step": 3801
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.334862385321101e-06,
+      "loss": 2.2099,
+      "step": 3802
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.3279816513761474e-06,
+      "loss": 2.2658,
+      "step": 3803
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.321100917431193e-06,
+      "loss": 2.0898,
+      "step": 3804
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.314220183486239e-06,
+      "loss": 2.1276,
+      "step": 3805
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.307339449541284e-06,
+      "loss": 2.0739,
+      "step": 3806
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.300458715596331e-06,
+      "loss": 2.2434,
+      "step": 3807
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.293577981651376e-06,
+      "loss": 2.1442,
+      "step": 3808
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.286697247706422e-06,
+      "loss": 1.9666,
+      "step": 3809
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.2798165137614685e-06,
+      "loss": 2.1358,
+      "step": 3810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.272935779816514e-06,
+      "loss": 2.0794,
+      "step": 3811
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.2660550458715605e-06,
+      "loss": 2.0447,
+      "step": 3812
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.259174311926605e-06,
+      "loss": 2.1667,
+      "step": 3813
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.252293577981652e-06,
+      "loss": 2.0882,
+      "step": 3814
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.245412844036697e-06,
+      "loss": 1.9554,
+      "step": 3815
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.238532110091744e-06,
+      "loss": 2.1559,
+      "step": 3816
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.231651376146789e-06,
+      "loss": 1.9321,
+      "step": 3817
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.224770642201835e-06,
+      "loss": 1.9445,
+      "step": 3818
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.217889908256881e-06,
+      "loss": 2.0805,
+      "step": 3819
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.211009174311927e-06,
+      "loss": 1.9869,
+      "step": 3820
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.204128440366973e-06,
+      "loss": 1.9609,
+      "step": 3821
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.1972477064220184e-06,
+      "loss": 1.9205,
+      "step": 3822
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.190366972477065e-06,
+      "loss": 1.9793,
+      "step": 3823
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.1834862385321105e-06,
+      "loss": 1.7938,
+      "step": 3824
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.176605504587156e-06,
+      "loss": 1.8212,
+      "step": 3825
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.169724770642202e-06,
+      "loss": 1.9119,
+      "step": 3826
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.162844036697248e-06,
+      "loss": 1.7672,
+      "step": 3827
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.155963302752294e-06,
+      "loss": 1.6615,
+      "step": 3828
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.1490825688073395e-06,
+      "loss": 1.6634,
+      "step": 3829
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.142201834862385e-06,
+      "loss": 1.7511,
+      "step": 3830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.1353211009174315e-06,
+      "loss": 1.6702,
+      "step": 3831
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.128440366972478e-06,
+      "loss": 1.7702,
+      "step": 3832
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.121559633027523e-06,
+      "loss": 1.5426,
+      "step": 3833
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.114678899082569e-06,
+      "loss": 1.5898,
+      "step": 3834
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.107798165137615e-06,
+      "loss": 1.66,
+      "step": 3835
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.100917431192661e-06,
+      "loss": 1.5749,
+      "step": 3836
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.094036697247706e-06,
+      "loss": 1.4471,
+      "step": 3837
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.0871559633027526e-06,
+      "loss": 1.3152,
+      "step": 3838
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.080275229357798e-06,
+      "loss": 1.2866,
+      "step": 3839
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.073394495412845e-06,
+      "loss": 1.2378,
+      "step": 3840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.06651376146789e-06,
+      "loss": 2.6311,
+      "step": 3841
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.059633027522936e-06,
+      "loss": 2.483,
+      "step": 3842
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.052752293577982e-06,
+      "loss": 2.4859,
+      "step": 3843
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.045871559633028e-06,
+      "loss": 2.399,
+      "step": 3844
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.038990825688074e-06,
+      "loss": 2.2831,
+      "step": 3845
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.032110091743119e-06,
+      "loss": 2.2895,
+      "step": 3846
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.025229357798166e-06,
+      "loss": 2.2699,
+      "step": 3847
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.018348623853211e-06,
+      "loss": 2.2481,
+      "step": 3848
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.011467889908257e-06,
+      "loss": 2.2911,
+      "step": 3849
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.0045871559633025e-06,
+      "loss": 2.18,
+      "step": 3850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.997706422018349e-06,
+      "loss": 2.2267,
+      "step": 3851
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.9908256880733955e-06,
+      "loss": 2.3523,
+      "step": 3852
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.98394495412844e-06,
+      "loss": 2.2131,
+      "step": 3853
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.977064220183487e-06,
+      "loss": 2.1146,
+      "step": 3854
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.970183486238532e-06,
+      "loss": 2.1931,
+      "step": 3855
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.963302752293579e-06,
+      "loss": 2.2101,
+      "step": 3856
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.9564220183486236e-06,
+      "loss": 2.1625,
+      "step": 3857
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.94954128440367e-06,
+      "loss": 2.116,
+      "step": 3858
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.942660550458716e-06,
+      "loss": 2.2926,
+      "step": 3859
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.935779816513762e-06,
+      "loss": 2.1174,
+      "step": 3860
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.928899082568807e-06,
+      "loss": 2.1267,
+      "step": 3861
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.922018348623853e-06,
+      "loss": 1.9563,
+      "step": 3862
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.9151376146789e-06,
+      "loss": 2.0419,
+      "step": 3863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.9082568807339454e-06,
+      "loss": 1.929,
+      "step": 3864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.901376146788991e-06,
+      "loss": 2.1802,
+      "step": 3865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.894495412844037e-06,
+      "loss": 2.1509,
+      "step": 3866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.887614678899083e-06,
+      "loss": 2.1969,
+      "step": 3867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.880733944954129e-06,
+      "loss": 1.984,
+      "step": 3868
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.873853211009174e-06,
+      "loss": 1.9001,
+      "step": 3869
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.86697247706422e-06,
+      "loss": 2.0063,
+      "step": 3870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.8600917431192665e-06,
+      "loss": 1.8078,
+      "step": 3871
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.853211009174312e-06,
+      "loss": 1.9253,
+      "step": 3872
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.846330275229358e-06,
+      "loss": 2.0094,
+      "step": 3873
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.839449541284404e-06,
+      "loss": 1.9587,
+      "step": 3874
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.83256880733945e-06,
+      "loss": 1.7557,
+      "step": 3875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.825688073394496e-06,
+      "loss": 1.9473,
+      "step": 3876
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.818807339449541e-06,
+      "loss": 1.8544,
+      "step": 3877
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.8119266055045875e-06,
+      "loss": 1.8073,
+      "step": 3878
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.805045871559633e-06,
+      "loss": 1.7376,
+      "step": 3879
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.7981651376146796e-06,
+      "loss": 1.7727,
+      "step": 3880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.791284403669724e-06,
+      "loss": 1.5524,
+      "step": 3881
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.784403669724771e-06,
+      "loss": 1.6898,
+      "step": 3882
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.777522935779817e-06,
+      "loss": 1.6956,
+      "step": 3883
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.770642201834863e-06,
+      "loss": 1.5897,
+      "step": 3884
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.7637614678899085e-06,
+      "loss": 1.5383,
+      "step": 3885
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.756880733944954e-06,
+      "loss": 1.5265,
+      "step": 3886
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 1.3752,
+      "step": 3887
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.743119266055046e-06,
+      "loss": 1.312,
+      "step": 3888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.736238532110092e-06,
+      "loss": 1.1958,
+      "step": 3889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.7293577981651375e-06,
+      "loss": 1.2636,
+      "step": 3890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.722477064220184e-06,
+      "loss": 2.7405,
+      "step": 3891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.7155963302752295e-06,
+      "loss": 2.3877,
+      "step": 3892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.708715596330275e-06,
+      "loss": 2.41,
+      "step": 3893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.701834862385322e-06,
+      "loss": 2.3731,
+      "step": 3894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.694954128440367e-06,
+      "loss": 2.2152,
+      "step": 3895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.688073394495414e-06,
+      "loss": 2.1126,
+      "step": 3896
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.6811926605504585e-06,
+      "loss": 2.2331,
+      "step": 3897
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.674311926605505e-06,
+      "loss": 2.3321,
+      "step": 3898
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.6674311926605506e-06,
+      "loss": 2.3232,
+      "step": 3899
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.660550458715597e-06,
+      "loss": 2.2773,
+      "step": 3900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.653669724770642e-06,
+      "loss": 2.2719,
+      "step": 3901
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.646788990825688e-06,
+      "loss": 2.2564,
+      "step": 3902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.639908256880734e-06,
+      "loss": 2.2337,
+      "step": 3903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.63302752293578e-06,
+      "loss": 2.1086,
+      "step": 3904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.626146788990826e-06,
+      "loss": 2.2393,
+      "step": 3905
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.619266055045872e-06,
+      "loss": 2.0902,
+      "step": 3906
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.612385321100918e-06,
+      "loss": 2.2694,
+      "step": 3907
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.605504587155964e-06,
+      "loss": 2.1957,
+      "step": 3908
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.598623853211009e-06,
+      "loss": 2.1071,
+      "step": 3909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.591743119266055e-06,
+      "loss": 2.2382,
+      "step": 3910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.584862385321101e-06,
+      "loss": 2.0901,
+      "step": 3911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.577981651376147e-06,
+      "loss": 2.0432,
+      "step": 3912
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.571100917431193e-06,
+      "loss": 2.0227,
+      "step": 3913
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.564220183486238e-06,
+      "loss": 1.9976,
+      "step": 3914
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.557339449541285e-06,
+      "loss": 2.1283,
+      "step": 3915
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.550458715596331e-06,
+      "loss": 1.881,
+      "step": 3916
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.543577981651376e-06,
+      "loss": 1.9765,
+      "step": 3917
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.536697247706422e-06,
+      "loss": 2.0278,
+      "step": 3918
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.529816513761468e-06,
+      "loss": 1.933,
+      "step": 3919
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.5229357798165145e-06,
+      "loss": 1.9329,
+      "step": 3920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.516055045871559e-06,
+      "loss": 1.9291,
+      "step": 3921
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.509174311926606e-06,
+      "loss": 1.7604,
+      "step": 3922
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.502293577981651e-06,
+      "loss": 1.853,
+      "step": 3923
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.495412844036698e-06,
+      "loss": 1.7873,
+      "step": 3924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.4885321100917434e-06,
+      "loss": 1.7002,
+      "step": 3925
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.481651376146789e-06,
+      "loss": 1.7557,
+      "step": 3926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.4747706422018355e-06,
+      "loss": 1.7271,
+      "step": 3927
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.467889908256881e-06,
+      "loss": 1.7145,
+      "step": 3928
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.461009174311927e-06,
+      "loss": 1.7617,
+      "step": 3929
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.454128440366972e-06,
+      "loss": 1.759,
+      "step": 3930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.447247706422019e-06,
+      "loss": 1.6262,
+      "step": 3931
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.4403669724770645e-06,
+      "loss": 1.5102,
+      "step": 3932
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.43348623853211e-06,
+      "loss": 1.6071,
+      "step": 3933
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.426605504587156e-06,
+      "loss": 1.5599,
+      "step": 3934
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.419724770642202e-06,
+      "loss": 1.4417,
+      "step": 3935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.412844036697249e-06,
+      "loss": 1.5164,
+      "step": 3936
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.405963302752293e-06,
+      "loss": 1.2601,
+      "step": 3937
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.39908256880734e-06,
+      "loss": 1.284,
+      "step": 3938
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.3922018348623855e-06,
+      "loss": 1.1724,
+      "step": 3939
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.385321100917432e-06,
+      "loss": 1.0948,
+      "step": 3940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.378440366972477e-06,
+      "loss": 2.8175,
+      "step": 3941
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.371559633027523e-06,
+      "loss": 2.6538,
+      "step": 3942
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.364678899082569e-06,
+      "loss": 2.5052,
+      "step": 3943
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.357798165137615e-06,
+      "loss": 2.5289,
+      "step": 3944
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.35091743119266e-06,
+      "loss": 2.365,
+      "step": 3945
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.3440366972477065e-06,
+      "loss": 2.2959,
+      "step": 3946
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.337155963302753e-06,
+      "loss": 2.2856,
+      "step": 3947
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.330275229357799e-06,
+      "loss": 2.1804,
+      "step": 3948
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.323394495412844e-06,
+      "loss": 2.4674,
+      "step": 3949
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.31651376146789e-06,
+      "loss": 2.3102,
+      "step": 3950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.309633027522936e-06,
+      "loss": 2.1561,
+      "step": 3951
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.302752293577982e-06,
+      "loss": 2.1673,
+      "step": 3952
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.2958715596330275e-06,
+      "loss": 2.2269,
+      "step": 3953
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.288990825688073e-06,
+      "loss": 2.1471,
+      "step": 3954
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.28211009174312e-06,
+      "loss": 2.2879,
+      "step": 3955
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.275229357798165e-06,
+      "loss": 2.067,
+      "step": 3956
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.268348623853211e-06,
+      "loss": 2.2658,
+      "step": 3957
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.261467889908257e-06,
+      "loss": 2.0603,
+      "step": 3958
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.254587155963303e-06,
+      "loss": 2.0898,
+      "step": 3959
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.247706422018349e-06,
+      "loss": 2.1609,
+      "step": 3960
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.240825688073394e-06,
+      "loss": 2.098,
+      "step": 3961
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.233944954128441e-06,
+      "loss": 2.0237,
+      "step": 3962
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.227064220183486e-06,
+      "loss": 2.0424,
+      "step": 3963
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.220183486238533e-06,
+      "loss": 2.0897,
+      "step": 3964
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.2133027522935775e-06,
+      "loss": 2.2128,
+      "step": 3965
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.206422018348624e-06,
+      "loss": 2.0339,
+      "step": 3966
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.1995412844036704e-06,
+      "loss": 2.1246,
+      "step": 3967
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.192660550458716e-06,
+      "loss": 2.0993,
+      "step": 3968
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.185779816513762e-06,
+      "loss": 2.0679,
+      "step": 3969
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.178899082568807e-06,
+      "loss": 2.041,
+      "step": 3970
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.172018348623854e-06,
+      "loss": 1.9872,
+      "step": 3971
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.165137614678899e-06,
+      "loss": 1.9205,
+      "step": 3972
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.158256880733945e-06,
+      "loss": 1.88,
+      "step": 3973
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.151376146788991e-06,
+      "loss": 1.8007,
+      "step": 3974
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.144495412844037e-06,
+      "loss": 1.7197,
+      "step": 3975
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.137614678899083e-06,
+      "loss": 1.7142,
+      "step": 3976
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.130733944954128e-06,
+      "loss": 1.9192,
+      "step": 3977
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.123853211009175e-06,
+      "loss": 1.7082,
+      "step": 3978
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.11697247706422e-06,
+      "loss": 1.7646,
+      "step": 3979
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.110091743119267e-06,
+      "loss": 1.7354,
+      "step": 3980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.103211009174312e-06,
+      "loss": 1.6924,
+      "step": 3981
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.096330275229358e-06,
+      "loss": 1.6106,
+      "step": 3982
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.089449541284404e-06,
+      "loss": 1.5497,
+      "step": 3983
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.08256880733945e-06,
+      "loss": 1.5193,
+      "step": 3984
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.075688073394495e-06,
+      "loss": 1.4658,
+      "step": 3985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.068807339449541e-06,
+      "loss": 1.3346,
+      "step": 3986
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.061926605504587e-06,
+      "loss": 1.3688,
+      "step": 3987
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.0550458715596335e-06,
+      "loss": 1.271,
+      "step": 3988
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.048165137614679e-06,
+      "loss": 1.2037,
+      "step": 3989
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.041284403669725e-06,
+      "loss": 1.3393,
+      "step": 3990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.034403669724771e-06,
+      "loss": 2.8051,
+      "step": 3991
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.027522935779817e-06,
+      "loss": 2.6131,
+      "step": 3992
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.0206422018348624e-06,
+      "loss": 2.3912,
+      "step": 3993
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.013761467889908e-06,
+      "loss": 2.3402,
+      "step": 3994
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.0068807339449545e-06,
+      "loss": 2.4341,
+      "step": 3995
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6e-06,
+      "loss": 2.1927,
+      "step": 3996
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.993119266055046e-06,
+      "loss": 2.2144,
+      "step": 3997
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.986238532110091e-06,
+      "loss": 2.2777,
+      "step": 3998
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.979357798165138e-06,
+      "loss": 2.3425,
+      "step": 3999
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.972477064220184e-06,
+      "loss": 2.2063,
+      "step": 4000
+    },
+    {
+      "epoch": 2.47,
+      "eval_bleu": 2.3052062590891975e-13,
+      "eval_loss": 2.2451658248901367,
+      "eval_runtime": 2533.9805,
+      "eval_samples_per_second": 5.825,
+      "eval_steps_per_second": 0.728,
+      "step": 4000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.965596330275229e-06,
+      "loss": 2.265,
+      "step": 4001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9587155963302756e-06,
+      "loss": 2.1912,
+      "step": 4002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.951834862385321e-06,
+      "loss": 2.3031,
+      "step": 4003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.944954128440368e-06,
+      "loss": 2.1859,
+      "step": 4004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.938073394495412e-06,
+      "loss": 2.242,
+      "step": 4005
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.931192660550459e-06,
+      "loss": 2.1581,
+      "step": 4006
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.9243119266055045e-06,
+      "loss": 2.0825,
+      "step": 4007
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.917431192660551e-06,
+      "loss": 2.1744,
+      "step": 4008
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.910550458715597e-06,
+      "loss": 2.1077,
+      "step": 4009
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.903669724770642e-06,
+      "loss": 2.216,
+      "step": 4010
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.896788990825689e-06,
+      "loss": 2.0901,
+      "step": 4011
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.889908256880734e-06,
+      "loss": 2.0351,
+      "step": 4012
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.88302752293578e-06,
+      "loss": 2.0512,
+      "step": 4013
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.8761467889908255e-06,
+      "loss": 2.0389,
+      "step": 4014
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.869266055045872e-06,
+      "loss": 2.1185,
+      "step": 4015
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.862385321100918e-06,
+      "loss": 2.0654,
+      "step": 4016
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.855504587155963e-06,
+      "loss": 1.9399,
+      "step": 4017
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.848623853211009e-06,
+      "loss": 2.0274,
+      "step": 4018
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.841743119266055e-06,
+      "loss": 1.9662,
+      "step": 4019
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.834862385321102e-06,
+      "loss": 1.9229,
+      "step": 4020
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.8279816513761466e-06,
+      "loss": 1.8094,
+      "step": 4021
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.821100917431193e-06,
+      "loss": 1.9064,
+      "step": 4022
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.814220183486239e-06,
+      "loss": 1.9551,
+      "step": 4023
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.807339449541285e-06,
+      "loss": 1.822,
+      "step": 4024
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.80045871559633e-06,
+      "loss": 1.7912,
+      "step": 4025
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.793577981651376e-06,
+      "loss": 1.8435,
+      "step": 4026
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.786697247706422e-06,
+      "loss": 1.9694,
+      "step": 4027
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7798165137614684e-06,
+      "loss": 1.6722,
+      "step": 4028
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.772935779816513e-06,
+      "loss": 1.714,
+      "step": 4029
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.76605504587156e-06,
+      "loss": 1.7326,
+      "step": 4030
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.759174311926606e-06,
+      "loss": 1.5729,
+      "step": 4031
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.752293577981652e-06,
+      "loss": 1.6166,
+      "step": 4032
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.745412844036697e-06,
+      "loss": 1.7122,
+      "step": 4033
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.738532110091743e-06,
+      "loss": 1.4837,
+      "step": 4034
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7316513761467895e-06,
+      "loss": 1.537,
+      "step": 4035
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.724770642201835e-06,
+      "loss": 1.3994,
+      "step": 4036
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.717889908256881e-06,
+      "loss": 1.6152,
+      "step": 4037
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.711009174311926e-06,
+      "loss": 1.2712,
+      "step": 4038
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.704128440366973e-06,
+      "loss": 1.3337,
+      "step": 4039
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.697247706422018e-06,
+      "loss": 1.0329,
+      "step": 4040
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.690366972477064e-06,
+      "loss": 2.823,
+      "step": 4041
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6834862385321105e-06,
+      "loss": 2.4797,
+      "step": 4042
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.676605504587156e-06,
+      "loss": 2.1179,
+      "step": 4043
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6697247706422026e-06,
+      "loss": 2.3292,
+      "step": 4044
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.662844036697247e-06,
+      "loss": 2.2589,
+      "step": 4045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.655963302752294e-06,
+      "loss": 2.2338,
+      "step": 4046
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.649082568807339e-06,
+      "loss": 2.2306,
+      "step": 4047
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.642201834862386e-06,
+      "loss": 2.125,
+      "step": 4048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.635321100917431e-06,
+      "loss": 2.2185,
+      "step": 4049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.628440366972477e-06,
+      "loss": 2.3527,
+      "step": 4050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.621559633027524e-06,
+      "loss": 2.1877,
+      "step": 4051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.614678899082569e-06,
+      "loss": 2.0892,
+      "step": 4052
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.607798165137615e-06,
+      "loss": 2.1914,
+      "step": 4053
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6009174311926604e-06,
+      "loss": 2.0974,
+      "step": 4054
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.594036697247707e-06,
+      "loss": 2.225,
+      "step": 4055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.5871559633027525e-06,
+      "loss": 2.2216,
+      "step": 4056
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.580275229357798e-06,
+      "loss": 2.2065,
+      "step": 4057
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.573394495412844e-06,
+      "loss": 2.134,
+      "step": 4058
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.56651376146789e-06,
+      "loss": 2.0502,
+      "step": 4059
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.559633027522936e-06,
+      "loss": 2.1295,
+      "step": 4060
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.5527522935779815e-06,
+      "loss": 2.0408,
+      "step": 4061
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.545871559633028e-06,
+      "loss": 2.2106,
+      "step": 4062
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.5389908256880736e-06,
+      "loss": 2.134,
+      "step": 4063
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.53211009174312e-06,
+      "loss": 2.0812,
+      "step": 4064
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.525229357798165e-06,
+      "loss": 1.882,
+      "step": 4065
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.518348623853211e-06,
+      "loss": 2.014,
+      "step": 4066
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.511467889908257e-06,
+      "loss": 2.0468,
+      "step": 4067
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.504587155963303e-06,
+      "loss": 1.8671,
+      "step": 4068
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.497706422018348e-06,
+      "loss": 2.0053,
+      "step": 4069
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.490825688073395e-06,
+      "loss": 1.9435,
+      "step": 4070
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.48394495412844e-06,
+      "loss": 2.0221,
+      "step": 4071
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.477064220183487e-06,
+      "loss": 1.8406,
+      "step": 4072
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.470183486238532e-06,
+      "loss": 1.8137,
+      "step": 4073
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.463302752293578e-06,
+      "loss": 1.8664,
+      "step": 4074
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.456422018348624e-06,
+      "loss": 1.9112,
+      "step": 4075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.44954128440367e-06,
+      "loss": 1.7841,
+      "step": 4076
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.442660550458716e-06,
+      "loss": 1.8564,
+      "step": 4077
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.435779816513761e-06,
+      "loss": 1.6841,
+      "step": 4078
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.428899082568808e-06,
+      "loss": 1.7562,
+      "step": 4079
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.422018348623853e-06,
+      "loss": 1.6039,
+      "step": 4080
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.415137614678899e-06,
+      "loss": 1.5508,
+      "step": 4081
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.408256880733945e-06,
+      "loss": 1.6522,
+      "step": 4082
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.401376146788991e-06,
+      "loss": 1.7065,
+      "step": 4083
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3944954128440375e-06,
+      "loss": 1.4342,
+      "step": 4084
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.387614678899082e-06,
+      "loss": 1.3542,
+      "step": 4085
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.380733944954129e-06,
+      "loss": 1.377,
+      "step": 4086
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.373853211009174e-06,
+      "loss": 1.4001,
+      "step": 4087
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.366972477064221e-06,
+      "loss": 1.2956,
+      "step": 4088
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.3600917431192656e-06,
+      "loss": 1.0924,
+      "step": 4089
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.353211009174312e-06,
+      "loss": 1.1975,
+      "step": 4090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.346330275229358e-06,
+      "loss": 2.7519,
+      "step": 4091
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.339449541284404e-06,
+      "loss": 2.4629,
+      "step": 4092
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.33256880733945e-06,
+      "loss": 2.3556,
+      "step": 4093
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.325688073394495e-06,
+      "loss": 2.3688,
+      "step": 4094
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.318807339449542e-06,
+      "loss": 2.3597,
+      "step": 4095
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.3119266055045874e-06,
+      "loss": 2.2906,
+      "step": 4096
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.305045871559633e-06,
+      "loss": 2.2029,
+      "step": 4097
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.298165137614679e-06,
+      "loss": 2.2312,
+      "step": 4098
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.291284403669725e-06,
+      "loss": 2.3773,
+      "step": 4099
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.284403669724771e-06,
+      "loss": 2.395,
+      "step": 4100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.277522935779816e-06,
+      "loss": 2.3824,
+      "step": 4101
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.270642201834862e-06,
+      "loss": 2.166,
+      "step": 4102
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.2637614678899085e-06,
+      "loss": 2.1849,
+      "step": 4103
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.256880733944955e-06,
+      "loss": 1.9308,
+      "step": 4104
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.25e-06,
+      "loss": 2.2481,
+      "step": 4105
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.243119266055046e-06,
+      "loss": 2.2545,
+      "step": 4106
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.236238532110092e-06,
+      "loss": 2.0815,
+      "step": 4107
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.229357798165138e-06,
+      "loss": 2.0915,
+      "step": 4108
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.222477064220183e-06,
+      "loss": 2.0573,
+      "step": 4109
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.2155963302752295e-06,
+      "loss": 2.1547,
+      "step": 4110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.208715596330275e-06,
+      "loss": 2.0727,
+      "step": 4111
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.201834862385322e-06,
+      "loss": 1.947,
+      "step": 4112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.194954128440366e-06,
+      "loss": 2.0358,
+      "step": 4113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.188073394495413e-06,
+      "loss": 1.9545,
+      "step": 4114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.181192660550459e-06,
+      "loss": 2.0401,
+      "step": 4115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.174311926605505e-06,
+      "loss": 1.8888,
+      "step": 4116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1674311926605505e-06,
+      "loss": 1.9817,
+      "step": 4117
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.160550458715596e-06,
+      "loss": 2.0451,
+      "step": 4118
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.153669724770643e-06,
+      "loss": 1.8734,
+      "step": 4119
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.146788990825688e-06,
+      "loss": 1.993,
+      "step": 4120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.139908256880734e-06,
+      "loss": 1.8829,
+      "step": 4121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1330275229357795e-06,
+      "loss": 1.861,
+      "step": 4122
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.126146788990826e-06,
+      "loss": 1.8573,
+      "step": 4123
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.119266055045872e-06,
+      "loss": 1.8032,
+      "step": 4124
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.112385321100917e-06,
+      "loss": 1.7708,
+      "step": 4125
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.105504587155964e-06,
+      "loss": 1.8962,
+      "step": 4126
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.098623853211009e-06,
+      "loss": 1.8115,
+      "step": 4127
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.091743119266056e-06,
+      "loss": 1.839,
+      "step": 4128
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0848623853211005e-06,
+      "loss": 1.6852,
+      "step": 4129
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.077981651376147e-06,
+      "loss": 1.4876,
+      "step": 4130
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0711009174311926e-06,
+      "loss": 1.7772,
+      "step": 4131
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.064220183486239e-06,
+      "loss": 1.578,
+      "step": 4132
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.057339449541284e-06,
+      "loss": 1.6444,
+      "step": 4133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.05045871559633e-06,
+      "loss": 1.5213,
+      "step": 4134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.043577981651377e-06,
+      "loss": 1.5585,
+      "step": 4135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.036697247706422e-06,
+      "loss": 1.3622,
+      "step": 4136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.029816513761468e-06,
+      "loss": 1.4147,
+      "step": 4137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.022935779816514e-06,
+      "loss": 1.3823,
+      "step": 4138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.01605504587156e-06,
+      "loss": 1.1102,
+      "step": 4139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.009174311926606e-06,
+      "loss": 1.2829,
+      "step": 4140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.002293577981651e-06,
+      "loss": 2.7352,
+      "step": 4141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.995412844036697e-06,
+      "loss": 2.5007,
+      "step": 4142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.988532110091743e-06,
+      "loss": 2.4606,
+      "step": 4143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.981651376146789e-06,
+      "loss": 2.3025,
+      "step": 4144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.974770642201835e-06,
+      "loss": 2.3543,
+      "step": 4145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.967889908256881e-06,
+      "loss": 2.2205,
+      "step": 4146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.961009174311927e-06,
+      "loss": 2.1545,
+      "step": 4147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.954128440366973e-06,
+      "loss": 2.1744,
+      "step": 4148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.947247706422018e-06,
+      "loss": 2.1971,
+      "step": 4149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.940366972477064e-06,
+      "loss": 2.2201,
+      "step": 4150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.93348623853211e-06,
+      "loss": 2.181,
+      "step": 4151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.9266055045871565e-06,
+      "loss": 2.1874,
+      "step": 4152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.919724770642201e-06,
+      "loss": 2.2029,
+      "step": 4153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.912844036697248e-06,
+      "loss": 2.0889,
+      "step": 4154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.905963302752293e-06,
+      "loss": 2.1723,
+      "step": 4155
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.89908256880734e-06,
+      "loss": 2.2582,
+      "step": 4156
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8922018348623854e-06,
+      "loss": 2.0659,
+      "step": 4157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.885321100917431e-06,
+      "loss": 2.2409,
+      "step": 4158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8784403669724775e-06,
+      "loss": 2.103,
+      "step": 4159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.871559633027523e-06,
+      "loss": 2.0702,
+      "step": 4160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.864678899082569e-06,
+      "loss": 2.0346,
+      "step": 4161
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.857798165137614e-06,
+      "loss": 2.1019,
+      "step": 4162
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.850917431192661e-06,
+      "loss": 2.0764,
+      "step": 4163
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8440366972477065e-06,
+      "loss": 2.0501,
+      "step": 4164
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.837155963302752e-06,
+      "loss": 2.0422,
+      "step": 4165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.8302752293577986e-06,
+      "loss": 2.023,
+      "step": 4166
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.823394495412844e-06,
+      "loss": 2.0833,
+      "step": 4167
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.816513761467891e-06,
+      "loss": 2.0905,
+      "step": 4168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.809633027522935e-06,
+      "loss": 1.9893,
+      "step": 4169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.802752293577982e-06,
+      "loss": 1.8898,
+      "step": 4170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.7958715596330275e-06,
+      "loss": 1.9489,
+      "step": 4171
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.788990825688074e-06,
+      "loss": 1.9328,
+      "step": 4172
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.782110091743119e-06,
+      "loss": 1.8578,
+      "step": 4173
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.775229357798165e-06,
+      "loss": 1.8338,
+      "step": 4174
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.768348623853211e-06,
+      "loss": 1.828,
+      "step": 4175
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.761467889908257e-06,
+      "loss": 1.9067,
+      "step": 4176
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.754587155963303e-06,
+      "loss": 1.662,
+      "step": 4177
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.7477064220183485e-06,
+      "loss": 1.8116,
+      "step": 4178
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.740825688073395e-06,
+      "loss": 1.7277,
+      "step": 4179
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.733944954128441e-06,
+      "loss": 1.7047,
+      "step": 4180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.727064220183486e-06,
+      "loss": 1.7225,
+      "step": 4181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.720183486238532e-06,
+      "loss": 1.6849,
+      "step": 4182
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.713302752293578e-06,
+      "loss": 1.6158,
+      "step": 4183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.706422018348624e-06,
+      "loss": 1.4925,
+      "step": 4184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.6995412844036695e-06,
+      "loss": 1.3776,
+      "step": 4185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.692660550458715e-06,
+      "loss": 1.5233,
+      "step": 4186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.685779816513762e-06,
+      "loss": 1.4125,
+      "step": 4187
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.678899082568808e-06,
+      "loss": 1.3259,
+      "step": 4188
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.672018348623853e-06,
+      "loss": 1.1057,
+      "step": 4189
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.665137614678899e-06,
+      "loss": 1.2777,
+      "step": 4190
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.658256880733945e-06,
+      "loss": 2.623,
+      "step": 4191
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.651376146788991e-06,
+      "loss": 2.3705,
+      "step": 4192
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.644495412844036e-06,
+      "loss": 2.3048,
+      "step": 4193
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.637614678899083e-06,
+      "loss": 2.383,
+      "step": 4194
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.630733944954128e-06,
+      "loss": 2.3173,
+      "step": 4195
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.623853211009175e-06,
+      "loss": 2.2476,
+      "step": 4196
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6169724770642195e-06,
+      "loss": 2.354,
+      "step": 4197
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.610091743119266e-06,
+      "loss": 2.2268,
+      "step": 4198
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.6032110091743124e-06,
+      "loss": 2.1668,
+      "step": 4199
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.596330275229358e-06,
+      "loss": 2.132,
+      "step": 4200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.589449541284404e-06,
+      "loss": 2.1894,
+      "step": 4201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.582568807339449e-06,
+      "loss": 2.1454,
+      "step": 4202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.575688073394496e-06,
+      "loss": 2.1143,
+      "step": 4203
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.568807339449541e-06,
+      "loss": 2.0436,
+      "step": 4204
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.561926605504587e-06,
+      "loss": 2.1584,
+      "step": 4205
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.555045871559633e-06,
+      "loss": 2.1767,
+      "step": 4206
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.548165137614679e-06,
+      "loss": 1.9963,
+      "step": 4207
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5412844036697256e-06,
+      "loss": 2.1058,
+      "step": 4208
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.53440366972477e-06,
+      "loss": 2.0763,
+      "step": 4209
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.527522935779817e-06,
+      "loss": 1.9073,
+      "step": 4210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.520642201834862e-06,
+      "loss": 2.1543,
+      "step": 4211
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.513761467889909e-06,
+      "loss": 2.1339,
+      "step": 4212
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.506880733944954e-06,
+      "loss": 2.1665,
+      "step": 4213
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.5e-06,
+      "loss": 2.0758,
+      "step": 4214
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.493119266055046e-06,
+      "loss": 1.8045,
+      "step": 4215
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.486238532110092e-06,
+      "loss": 1.9071,
+      "step": 4216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.479357798165137e-06,
+      "loss": 2.0468,
+      "step": 4217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4724770642201834e-06,
+      "loss": 2.1895,
+      "step": 4218
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.46559633027523e-06,
+      "loss": 1.8468,
+      "step": 4219
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.4587155963302755e-06,
+      "loss": 2.0438,
+      "step": 4220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.451834862385321e-06,
+      "loss": 1.7339,
+      "step": 4221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.444954128440367e-06,
+      "loss": 1.8835,
+      "step": 4222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.438073394495413e-06,
+      "loss": 1.7959,
+      "step": 4223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.431192660550459e-06,
+      "loss": 1.7704,
+      "step": 4224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.4243119266055045e-06,
+      "loss": 1.8735,
+      "step": 4225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.41743119266055e-06,
+      "loss": 1.7917,
+      "step": 4226
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.4105504587155965e-06,
+      "loss": 1.7406,
+      "step": 4227
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.403669724770642e-06,
+      "loss": 1.6564,
+      "step": 4228
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.396788990825688e-06,
+      "loss": 1.7454,
+      "step": 4229
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.389908256880734e-06,
+      "loss": 1.7173,
+      "step": 4230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.38302752293578e-06,
+      "loss": 1.5854,
+      "step": 4231
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.376146788990826e-06,
+      "loss": 1.6481,
+      "step": 4232
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.369266055045871e-06,
+      "loss": 1.5806,
+      "step": 4233
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.3623853211009176e-06,
+      "loss": 1.4992,
+      "step": 4234
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.355504587155963e-06,
+      "loss": 1.4579,
+      "step": 4235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.34862385321101e-06,
+      "loss": 1.4356,
+      "step": 4236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3417431192660544e-06,
+      "loss": 1.3206,
+      "step": 4237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.334862385321101e-06,
+      "loss": 1.2671,
+      "step": 4238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.3279816513761465e-06,
+      "loss": 1.1885,
+      "step": 4239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.321100917431193e-06,
+      "loss": 1.1647,
+      "step": 4240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.314220183486239e-06,
+      "loss": 2.613,
+      "step": 4241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.307339449541284e-06,
+      "loss": 2.494,
+      "step": 4242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.300458715596331e-06,
+      "loss": 2.3585,
+      "step": 4243
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.293577981651376e-06,
+      "loss": 2.2828,
+      "step": 4244
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.286697247706422e-06,
+      "loss": 2.3766,
+      "step": 4245
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.2798165137614675e-06,
+      "loss": 2.3101,
+      "step": 4246
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.272935779816514e-06,
+      "loss": 2.1308,
+      "step": 4247
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.26605504587156e-06,
+      "loss": 2.0898,
+      "step": 4248
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.259174311926605e-06,
+      "loss": 2.149,
+      "step": 4249
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.252293577981652e-06,
+      "loss": 2.2782,
+      "step": 4250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.245412844036697e-06,
+      "loss": 2.1831,
+      "step": 4251
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.238532110091744e-06,
+      "loss": 2.1947,
+      "step": 4252
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.2316513761467886e-06,
+      "loss": 2.1111,
+      "step": 4253
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.224770642201835e-06,
+      "loss": 2.2748,
+      "step": 4254
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.217889908256881e-06,
+      "loss": 2.1529,
+      "step": 4255
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.211009174311927e-06,
+      "loss": 2.0317,
+      "step": 4256
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.204128440366972e-06,
+      "loss": 2.1784,
+      "step": 4257
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.197247706422018e-06,
+      "loss": 2.12,
+      "step": 4258
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.190366972477064e-06,
+      "loss": 1.9068,
+      "step": 4259
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1834862385321104e-06,
+      "loss": 2.1827,
+      "step": 4260
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.176605504587156e-06,
+      "loss": 2.0539,
+      "step": 4261
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.169724770642202e-06,
+      "loss": 2.0638,
+      "step": 4262
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.162844036697248e-06,
+      "loss": 2.0349,
+      "step": 4263
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.155963302752294e-06,
+      "loss": 2.1085,
+      "step": 4264
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.149082568807339e-06,
+      "loss": 2.1633,
+      "step": 4265
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.142201834862385e-06,
+      "loss": 1.862,
+      "step": 4266
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.1353211009174315e-06,
+      "loss": 1.9389,
+      "step": 4267
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.128440366972477e-06,
+      "loss": 1.9789,
+      "step": 4268
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.121559633027523e-06,
+      "loss": 2.0647,
+      "step": 4269
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.114678899082568e-06,
+      "loss": 1.9626,
+      "step": 4270
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.107798165137615e-06,
+      "loss": 1.8267,
+      "step": 4271
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.100917431192661e-06,
+      "loss": 1.8445,
+      "step": 4272
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.094036697247706e-06,
+      "loss": 1.844,
+      "step": 4273
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0871559633027525e-06,
+      "loss": 1.7624,
+      "step": 4274
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.080275229357798e-06,
+      "loss": 1.8934,
+      "step": 4275
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0733944954128446e-06,
+      "loss": 1.8293,
+      "step": 4276
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.066513761467889e-06,
+      "loss": 1.8567,
+      "step": 4277
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.059633027522936e-06,
+      "loss": 1.7199,
+      "step": 4278
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0527522935779814e-06,
+      "loss": 1.6612,
+      "step": 4279
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.045871559633028e-06,
+      "loss": 1.7591,
+      "step": 4280
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.0389908256880735e-06,
+      "loss": 1.639,
+      "step": 4281
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.032110091743119e-06,
+      "loss": 1.6854,
+      "step": 4282
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.025229357798166e-06,
+      "loss": 1.5491,
+      "step": 4283
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.018348623853211e-06,
+      "loss": 1.6528,
+      "step": 4284
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.011467889908257e-06,
+      "loss": 1.6524,
+      "step": 4285
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.0045871559633025e-06,
+      "loss": 1.4362,
+      "step": 4286
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.997706422018349e-06,
+      "loss": 1.3147,
+      "step": 4287
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9908256880733945e-06,
+      "loss": 1.2175,
+      "step": 4288
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.98394495412844e-06,
+      "loss": 1.0309,
+      "step": 4289
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.977064220183486e-06,
+      "loss": 1.1977,
+      "step": 4290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.970183486238532e-06,
+      "loss": 2.616,
+      "step": 4291
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.963302752293579e-06,
+      "loss": 2.4718,
+      "step": 4292
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9564220183486235e-06,
+      "loss": 2.3478,
+      "step": 4293
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.94954128440367e-06,
+      "loss": 2.2748,
+      "step": 4294
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.9426605504587156e-06,
+      "loss": 2.1768,
+      "step": 4295
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.935779816513762e-06,
+      "loss": 2.145,
+      "step": 4296
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.928899082568807e-06,
+      "loss": 2.2242,
+      "step": 4297
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.922018348623853e-06,
+      "loss": 2.2247,
+      "step": 4298
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.915137614678899e-06,
+      "loss": 2.154,
+      "step": 4299
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.908256880733945e-06,
+      "loss": 2.2411,
+      "step": 4300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.90137614678899e-06,
+      "loss": 2.2094,
+      "step": 4301
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.894495412844037e-06,
+      "loss": 2.1722,
+      "step": 4302
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.887614678899083e-06,
+      "loss": 2.1862,
+      "step": 4303
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.880733944954129e-06,
+      "loss": 2.0163,
+      "step": 4304
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.873853211009174e-06,
+      "loss": 2.2539,
+      "step": 4305
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.86697247706422e-06,
+      "loss": 2.1196,
+      "step": 4306
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.860091743119266e-06,
+      "loss": 2.0819,
+      "step": 4307
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.853211009174312e-06,
+      "loss": 2.1954,
+      "step": 4308
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.846330275229358e-06,
+      "loss": 2.0469,
+      "step": 4309
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.839449541284403e-06,
+      "loss": 2.0428,
+      "step": 4310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.83256880733945e-06,
+      "loss": 2.0056,
+      "step": 4311
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.825688073394495e-06,
+      "loss": 2.0858,
+      "step": 4312
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.818807339449541e-06,
+      "loss": 2.1275,
+      "step": 4313
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.811926605504587e-06,
+      "loss": 1.9781,
+      "step": 4314
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.805045871559633e-06,
+      "loss": 2.0291,
+      "step": 4315
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.798165137614679e-06,
+      "loss": 2.1032,
+      "step": 4316
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.7912844036697243e-06,
+      "loss": 1.946,
+      "step": 4317
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7844036697247707e-06,
+      "loss": 1.9996,
+      "step": 4318
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7775229357798168e-06,
+      "loss": 1.8577,
+      "step": 4319
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.770642201834863e-06,
+      "loss": 1.9763,
+      "step": 4320
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.763761467889908e-06,
+      "loss": 1.9121,
+      "step": 4321
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.756880733944954e-06,
+      "loss": 1.9352,
+      "step": 4322
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.75e-06,
+      "loss": 1.7315,
+      "step": 4323
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.743119266055046e-06,
+      "loss": 1.7162,
+      "step": 4324
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7362385321100918e-06,
+      "loss": 1.8008,
+      "step": 4325
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.729357798165138e-06,
+      "loss": 1.7511,
+      "step": 4326
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7224770642201834e-06,
+      "loss": 1.8311,
+      "step": 4327
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7155963302752295e-06,
+      "loss": 1.6827,
+      "step": 4328
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7087155963302755e-06,
+      "loss": 1.7025,
+      "step": 4329
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.7018348623853215e-06,
+      "loss": 1.7122,
+      "step": 4330
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.694954128440367e-06,
+      "loss": 1.637,
+      "step": 4331
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.688073394495413e-06,
+      "loss": 1.6587,
+      "step": 4332
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.681192660550459e-06,
+      "loss": 1.3938,
+      "step": 4333
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.674311926605505e-06,
+      "loss": 1.5781,
+      "step": 4334
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6674311926605505e-06,
+      "loss": 1.4377,
+      "step": 4335
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6605504587155965e-06,
+      "loss": 1.3574,
+      "step": 4336
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.653669724770642e-06,
+      "loss": 1.3775,
+      "step": 4337
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.646788990825688e-06,
+      "loss": 1.3136,
+      "step": 4338
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6399082568807342e-06,
+      "loss": 1.2532,
+      "step": 4339
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6330275229357803e-06,
+      "loss": 1.2203,
+      "step": 4340
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.626146788990826e-06,
+      "loss": 2.5656,
+      "step": 4341
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.619266055045872e-06,
+      "loss": 2.4689,
+      "step": 4342
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6123853211009176e-06,
+      "loss": 2.5337,
+      "step": 4343
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.6055045871559636e-06,
+      "loss": 2.4042,
+      "step": 4344
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5986238532110092e-06,
+      "loss": 2.0985,
+      "step": 4345
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5917431192660553e-06,
+      "loss": 2.2023,
+      "step": 4346
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.584862385321101e-06,
+      "loss": 2.0257,
+      "step": 4347
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.577981651376147e-06,
+      "loss": 2.131,
+      "step": 4348
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.5711009174311925e-06,
+      "loss": 2.2663,
+      "step": 4349
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.564220183486239e-06,
+      "loss": 2.1827,
+      "step": 4350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5573394495412846e-06,
+      "loss": 2.0192,
+      "step": 4351
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5504587155963307e-06,
+      "loss": 2.2417,
+      "step": 4352
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5435779816513763e-06,
+      "loss": 2.2602,
+      "step": 4353
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5366972477064223e-06,
+      "loss": 2.075,
+      "step": 4354
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.529816513761468e-06,
+      "loss": 2.1052,
+      "step": 4355
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.522935779816514e-06,
+      "loss": 2.051,
+      "step": 4356
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5160550458715596e-06,
+      "loss": 2.0696,
+      "step": 4357
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5091743119266056e-06,
+      "loss": 2.0667,
+      "step": 4358
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.5022935779816513e-06,
+      "loss": 2.1205,
+      "step": 4359
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4954128440366977e-06,
+      "loss": 2.0288,
+      "step": 4360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4885321100917434e-06,
+      "loss": 2.0973,
+      "step": 4361
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4816513761467894e-06,
+      "loss": 2.0951,
+      "step": 4362
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.474770642201835e-06,
+      "loss": 2.0677,
+      "step": 4363
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.467889908256881e-06,
+      "loss": 2.1335,
+      "step": 4364
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4610091743119267e-06,
+      "loss": 2.0187,
+      "step": 4365
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4541284403669727e-06,
+      "loss": 2.0283,
+      "step": 4366
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4472477064220183e-06,
+      "loss": 2.1001,
+      "step": 4367
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4403669724770644e-06,
+      "loss": 1.8184,
+      "step": 4368
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.43348623853211e-06,
+      "loss": 1.8973,
+      "step": 4369
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.426605504587156e-06,
+      "loss": 1.9186,
+      "step": 4370
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.419724770642202e-06,
+      "loss": 1.7952,
+      "step": 4371
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.412844036697248e-06,
+      "loss": 1.8999,
+      "step": 4372
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.4059633027522937e-06,
+      "loss": 2.0021,
+      "step": 4373
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3990825688073398e-06,
+      "loss": 1.9107,
+      "step": 4374
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3922018348623854e-06,
+      "loss": 1.7599,
+      "step": 4375
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3853211009174314e-06,
+      "loss": 1.6901,
+      "step": 4376
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.378440366972477e-06,
+      "loss": 1.6683,
+      "step": 4377
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.371559633027523e-06,
+      "loss": 1.6626,
+      "step": 4378
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3646788990825687e-06,
+      "loss": 1.6041,
+      "step": 4379
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.3577981651376148e-06,
+      "loss": 1.7613,
+      "step": 4380
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.350917431192661e-06,
+      "loss": 1.6017,
+      "step": 4381
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.344036697247707e-06,
+      "loss": 1.5124,
+      "step": 4382
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3371559633027525e-06,
+      "loss": 1.606,
+      "step": 4383
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.3302752293577985e-06,
+      "loss": 1.5327,
+      "step": 4384
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.323394495412844e-06,
+      "loss": 1.6028,
+      "step": 4385
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.31651376146789e-06,
+      "loss": 1.4629,
+      "step": 4386
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.309633027522936e-06,
+      "loss": 1.2163,
+      "step": 4387
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.302752293577982e-06,
+      "loss": 1.3676,
+      "step": 4388
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2958715596330275e-06,
+      "loss": 1.3458,
+      "step": 4389
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2889908256880735e-06,
+      "loss": 1.2863,
+      "step": 4390
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.282110091743119e-06,
+      "loss": 2.7415,
+      "step": 4391
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2752293577981656e-06,
+      "loss": 2.5099,
+      "step": 4392
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.268348623853211e-06,
+      "loss": 2.3943,
+      "step": 4393
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2614678899082572e-06,
+      "loss": 2.1602,
+      "step": 4394
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.254587155963303e-06,
+      "loss": 2.2061,
+      "step": 4395
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.247706422018349e-06,
+      "loss": 2.1599,
+      "step": 4396
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2408256880733945e-06,
+      "loss": 2.0952,
+      "step": 4397
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.2339449541284406e-06,
+      "loss": 2.1942,
+      "step": 4398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.227064220183486e-06,
+      "loss": 2.1437,
+      "step": 4399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2201834862385322e-06,
+      "loss": 2.2261,
+      "step": 4400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.213302752293578e-06,
+      "loss": 2.1839,
+      "step": 4401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.2064220183486243e-06,
+      "loss": 2.1314,
+      "step": 4402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.19954128440367e-06,
+      "loss": 1.9636,
+      "step": 4403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.192660550458716e-06,
+      "loss": 1.982,
+      "step": 4404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1857798165137616e-06,
+      "loss": 2.1982,
+      "step": 4405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1788990825688076e-06,
+      "loss": 2.1514,
+      "step": 4406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1720183486238533e-06,
+      "loss": 1.9799,
+      "step": 4407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1651376146788993e-06,
+      "loss": 1.9757,
+      "step": 4408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.158256880733945e-06,
+      "loss": 2.1071,
+      "step": 4409
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.151376146788991e-06,
+      "loss": 1.9922,
+      "step": 4410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1444954128440366e-06,
+      "loss": 1.9544,
+      "step": 4411
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1376146788990826e-06,
+      "loss": 1.97,
+      "step": 4412
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1307339449541287e-06,
+      "loss": 2.0637,
+      "step": 4413
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.1238532110091747e-06,
+      "loss": 1.9657,
+      "step": 4414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.1169724770642203e-06,
+      "loss": 1.8899,
+      "step": 4415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.1100917431192664e-06,
+      "loss": 2.0713,
+      "step": 4416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.103211009174312e-06,
+      "loss": 1.9186,
+      "step": 4417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.096330275229358e-06,
+      "loss": 1.9285,
+      "step": 4418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0894495412844036e-06,
+      "loss": 1.8838,
+      "step": 4419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0825688073394497e-06,
+      "loss": 1.9058,
+      "step": 4420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0756880733944953e-06,
+      "loss": 1.8975,
+      "step": 4421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0688073394495413e-06,
+      "loss": 2.0665,
+      "step": 4422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0619266055045874e-06,
+      "loss": 1.8544,
+      "step": 4423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0550458715596334e-06,
+      "loss": 1.8844,
+      "step": 4424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.048165137614679e-06,
+      "loss": 1.8009,
+      "step": 4425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.041284403669725e-06,
+      "loss": 1.9799,
+      "step": 4426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0344036697247707e-06,
+      "loss": 1.6686,
+      "step": 4427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0275229357798168e-06,
+      "loss": 1.6892,
+      "step": 4428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0206422018348624e-06,
+      "loss": 1.7,
+      "step": 4429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.0137614678899084e-06,
+      "loss": 1.6361,
+      "step": 4430
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.006880733944954e-06,
+      "loss": 1.5573,
+      "step": 4431
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3e-06,
+      "loss": 1.6128,
+      "step": 4432
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9931192660550457e-06,
+      "loss": 1.456,
+      "step": 4433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.986238532110092e-06,
+      "loss": 1.5058,
+      "step": 4434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9793577981651378e-06,
+      "loss": 1.4301,
+      "step": 4435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.972477064220184e-06,
+      "loss": 1.4457,
+      "step": 4436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9655963302752294e-06,
+      "loss": 1.2346,
+      "step": 4437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9587155963302755e-06,
+      "loss": 1.1568,
+      "step": 4438
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.951834862385321e-06,
+      "loss": 1.1611,
+      "step": 4439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.944954128440367e-06,
+      "loss": 1.2495,
+      "step": 4440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9380733944954128e-06,
+      "loss": 2.6661,
+      "step": 4441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.931192660550459e-06,
+      "loss": 2.4565,
+      "step": 4442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9243119266055044e-06,
+      "loss": 2.2914,
+      "step": 4443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.917431192660551e-06,
+      "loss": 2.2403,
+      "step": 4444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9105504587155965e-06,
+      "loss": 2.4559,
+      "step": 4445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.9036697247706426e-06,
+      "loss": 2.1723,
+      "step": 4446
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.896788990825688e-06,
+      "loss": 2.1987,
+      "step": 4447
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8899082568807342e-06,
+      "loss": 2.0769,
+      "step": 4448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.88302752293578e-06,
+      "loss": 2.0708,
+      "step": 4449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.876146788990826e-06,
+      "loss": 2.1007,
+      "step": 4450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8692660550458715e-06,
+      "loss": 2.1918,
+      "step": 4451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8623853211009175e-06,
+      "loss": 2.1238,
+      "step": 4452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.855504587155963e-06,
+      "loss": 2.1553,
+      "step": 4453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.848623853211009e-06,
+      "loss": 2.2297,
+      "step": 4454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8417431192660552e-06,
+      "loss": 2.0803,
+      "step": 4455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8348623853211013e-06,
+      "loss": 2.2475,
+      "step": 4456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.827981651376147e-06,
+      "loss": 1.9885,
+      "step": 4457
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.821100917431193e-06,
+      "loss": 2.2021,
+      "step": 4458
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8142201834862386e-06,
+      "loss": 2.0487,
+      "step": 4459
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8073394495412846e-06,
+      "loss": 2.0091,
+      "step": 4460
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.8004587155963302e-06,
+      "loss": 2.0675,
+      "step": 4461
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.7935779816513763e-06,
+      "loss": 1.9726,
+      "step": 4462
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.786697247706422e-06,
+      "loss": 1.8671,
+      "step": 4463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.779816513761468e-06,
+      "loss": 1.9628,
+      "step": 4464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.772935779816514e-06,
+      "loss": 1.9898,
+      "step": 4465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.76605504587156e-06,
+      "loss": 1.9593,
+      "step": 4466
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7591743119266056e-06,
+      "loss": 2.0171,
+      "step": 4467
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7522935779816517e-06,
+      "loss": 1.928,
+      "step": 4468
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7454128440366973e-06,
+      "loss": 2.0607,
+      "step": 4469
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7385321100917433e-06,
+      "loss": 1.9178,
+      "step": 4470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.731651376146789e-06,
+      "loss": 1.9014,
+      "step": 4471
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.724770642201835e-06,
+      "loss": 1.8738,
+      "step": 4472
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7178899082568806e-06,
+      "loss": 1.9992,
+      "step": 4473
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7110091743119267e-06,
+      "loss": 1.7439,
+      "step": 4474
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.7041284403669727e-06,
+      "loss": 1.9522,
+      "step": 4475
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6972477064220187e-06,
+      "loss": 1.943,
+      "step": 4476
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6903669724770644e-06,
+      "loss": 1.6706,
+      "step": 4477
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.6834862385321104e-06,
+      "loss": 1.7905,
+      "step": 4478
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.676605504587156e-06,
+      "loss": 1.6462,
+      "step": 4479
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.669724770642202e-06,
+      "loss": 1.6941,
+      "step": 4480
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6628440366972477e-06,
+      "loss": 1.6463,
+      "step": 4481
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6559633027522937e-06,
+      "loss": 1.5561,
+      "step": 4482
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6490825688073393e-06,
+      "loss": 1.6258,
+      "step": 4483
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6422018348623854e-06,
+      "loss": 1.4053,
+      "step": 4484
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.635321100917431e-06,
+      "loss": 1.4429,
+      "step": 4485
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6284403669724775e-06,
+      "loss": 1.3782,
+      "step": 4486
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.621559633027523e-06,
+      "loss": 1.1821,
+      "step": 4487
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.614678899082569e-06,
+      "loss": 1.2983,
+      "step": 4488
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.6077981651376147e-06,
+      "loss": 1.0852,
+      "step": 4489
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.600917431192661e-06,
+      "loss": 1.2689,
+      "step": 4490
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5940366972477064e-06,
+      "loss": 2.6388,
+      "step": 4491
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5871559633027525e-06,
+      "loss": 2.4295,
+      "step": 4492
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.580275229357798e-06,
+      "loss": 2.4013,
+      "step": 4493
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.573394495412844e-06,
+      "loss": 2.2664,
+      "step": 4494
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.5665137614678897e-06,
+      "loss": 2.1764,
+      "step": 4495
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.559633027522936e-06,
+      "loss": 2.1951,
+      "step": 4496
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.552752293577982e-06,
+      "loss": 2.2472,
+      "step": 4497
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.545871559633028e-06,
+      "loss": 2.1474,
+      "step": 4498
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5389908256880735e-06,
+      "loss": 2.0925,
+      "step": 4499
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.5321100917431195e-06,
+      "loss": 2.1087,
+      "step": 4500
+    },
+    {
+      "epoch": 2.78,
+      "eval_bleu": 1.0140224330923163e-13,
+      "eval_loss": 2.135566473007202,
+      "eval_runtime": 2535.9584,
+      "eval_samples_per_second": 5.82,
+      "eval_steps_per_second": 0.728,
+      "step": 4500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.525229357798165e-06,
+      "loss": 2.2231,
+      "step": 4501
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.518348623853211e-06,
+      "loss": 2.2482,
+      "step": 4502
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.511467889908257e-06,
+      "loss": 2.1241,
+      "step": 4503
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.504587155963303e-06,
+      "loss": 2.1448,
+      "step": 4504
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4977064220183485e-06,
+      "loss": 2.2204,
+      "step": 4505
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4908256880733945e-06,
+      "loss": 2.2021,
+      "step": 4506
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4839449541284405e-06,
+      "loss": 2.0509,
+      "step": 4507
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4770642201834866e-06,
+      "loss": 2.1017,
+      "step": 4508
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.470183486238532e-06,
+      "loss": 2.1455,
+      "step": 4509
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.4633027522935782e-06,
+      "loss": 2.0809,
+      "step": 4510
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.456422018348624e-06,
+      "loss": 1.8768,
+      "step": 4511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.44954128440367e-06,
+      "loss": 1.9641,
+      "step": 4512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4426605504587155e-06,
+      "loss": 2.0716,
+      "step": 4513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4357798165137616e-06,
+      "loss": 1.8871,
+      "step": 4514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.428899082568807e-06,
+      "loss": 1.9398,
+      "step": 4515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4220183486238532e-06,
+      "loss": 2.0203,
+      "step": 4516
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4151376146788993e-06,
+      "loss": 1.9631,
+      "step": 4517
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4082568807339453e-06,
+      "loss": 1.8563,
+      "step": 4518
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.401376146788991e-06,
+      "loss": 2.0022,
+      "step": 4519
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.394495412844037e-06,
+      "loss": 1.8043,
+      "step": 4520
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3876146788990826e-06,
+      "loss": 1.9046,
+      "step": 4521
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3807339449541286e-06,
+      "loss": 1.881,
+      "step": 4522
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3738532110091743e-06,
+      "loss": 1.9277,
+      "step": 4523
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3669724770642203e-06,
+      "loss": 1.8105,
+      "step": 4524
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.360091743119266e-06,
+      "loss": 1.7087,
+      "step": 4525
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.353211009174312e-06,
+      "loss": 1.7741,
+      "step": 4526
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3463302752293576e-06,
+      "loss": 1.6716,
+      "step": 4527
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.339449541284404e-06,
+      "loss": 1.598,
+      "step": 4528
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3325688073394497e-06,
+      "loss": 1.5719,
+      "step": 4529
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3256880733944957e-06,
+      "loss": 1.6272,
+      "step": 4530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3188073394495413e-06,
+      "loss": 1.5688,
+      "step": 4531
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3119266055045874e-06,
+      "loss": 1.4274,
+      "step": 4532
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.305045871559633e-06,
+      "loss": 1.58,
+      "step": 4533
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.298165137614679e-06,
+      "loss": 1.5637,
+      "step": 4534
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2912844036697246e-06,
+      "loss": 1.4032,
+      "step": 4535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2844036697247707e-06,
+      "loss": 1.3886,
+      "step": 4536
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2775229357798163e-06,
+      "loss": 1.2573,
+      "step": 4537
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2706422018348628e-06,
+      "loss": 1.2881,
+      "step": 4538
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2637614678899084e-06,
+      "loss": 1.1234,
+      "step": 4539
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2568807339449544e-06,
+      "loss": 1.3377,
+      "step": 4540
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.25e-06,
+      "loss": 2.7676,
+      "step": 4541
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.243119266055046e-06,
+      "loss": 2.3947,
+      "step": 4542
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2362385321100917e-06,
+      "loss": 2.207,
+      "step": 4543
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2293577981651378e-06,
+      "loss": 2.1999,
+      "step": 4544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2224770642201834e-06,
+      "loss": 2.0953,
+      "step": 4545
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2155963302752294e-06,
+      "loss": 2.2176,
+      "step": 4546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.208715596330275e-06,
+      "loss": 2.1633,
+      "step": 4547
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.201834862385321e-06,
+      "loss": 2.1374,
+      "step": 4548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.194954128440367e-06,
+      "loss": 2.1352,
+      "step": 4549
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.188073394495413e-06,
+      "loss": 2.0635,
+      "step": 4550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1811926605504588e-06,
+      "loss": 2.1164,
+      "step": 4551
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.174311926605505e-06,
+      "loss": 2.2743,
+      "step": 4552
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1674311926605504e-06,
+      "loss": 1.9629,
+      "step": 4553
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1605504587155965e-06,
+      "loss": 2.1459,
+      "step": 4554
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.153669724770642e-06,
+      "loss": 2.2676,
+      "step": 4555
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.146788990825688e-06,
+      "loss": 2.1793,
+      "step": 4556
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1399082568807338e-06,
+      "loss": 2.0742,
+      "step": 4557
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.13302752293578e-06,
+      "loss": 2.1064,
+      "step": 4558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.126146788990826e-06,
+      "loss": 2.0527,
+      "step": 4559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.119266055045872e-06,
+      "loss": 1.8197,
+      "step": 4560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1123853211009175e-06,
+      "loss": 2.0624,
+      "step": 4561
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1055045871559636e-06,
+      "loss": 2.0266,
+      "step": 4562
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.098623853211009e-06,
+      "loss": 1.9652,
+      "step": 4563
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0917431192660552e-06,
+      "loss": 2.0454,
+      "step": 4564
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.084862385321101e-06,
+      "loss": 2.0518,
+      "step": 4565
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.077981651376147e-06,
+      "loss": 2.1109,
+      "step": 4566
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0711009174311925e-06,
+      "loss": 2.0077,
+      "step": 4567
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0642201834862385e-06,
+      "loss": 1.9185,
+      "step": 4568
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.057339449541284e-06,
+      "loss": 2.058,
+      "step": 4569
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0504587155963306e-06,
+      "loss": 1.8024,
+      "step": 4570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0435779816513762e-06,
+      "loss": 2.0106,
+      "step": 4571
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0366972477064223e-06,
+      "loss": 1.9652,
+      "step": 4572
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.029816513761468e-06,
+      "loss": 1.6412,
+      "step": 4573
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.022935779816514e-06,
+      "loss": 1.881,
+      "step": 4574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0160550458715596e-06,
+      "loss": 1.8917,
+      "step": 4575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0091743119266056e-06,
+      "loss": 1.7382,
+      "step": 4576
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0022935779816512e-06,
+      "loss": 1.696,
+      "step": 4577
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9954128440366973e-06,
+      "loss": 1.8032,
+      "step": 4578
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.988532110091743e-06,
+      "loss": 1.5545,
+      "step": 4579
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9816513761467894e-06,
+      "loss": 1.6808,
+      "step": 4580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.974770642201835e-06,
+      "loss": 1.7285,
+      "step": 4581
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.967889908256881e-06,
+      "loss": 1.5878,
+      "step": 4582
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9610091743119266e-06,
+      "loss": 1.4901,
+      "step": 4583
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9541284403669727e-06,
+      "loss": 1.5007,
+      "step": 4584
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9472477064220183e-06,
+      "loss": 1.4378,
+      "step": 4585
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9403669724770643e-06,
+      "loss": 1.3985,
+      "step": 4586
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.93348623853211e-06,
+      "loss": 1.3698,
+      "step": 4587
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.926605504587156e-06,
+      "loss": 1.1733,
+      "step": 4588
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9197247706422016e-06,
+      "loss": 1.3073,
+      "step": 4589
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9128440366972477e-06,
+      "loss": 1.1552,
+      "step": 4590
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.9059633027522935e-06,
+      "loss": 2.7751,
+      "step": 4591
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8990825688073395e-06,
+      "loss": 2.3882,
+      "step": 4592
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8922018348623854e-06,
+      "loss": 2.357,
+      "step": 4593
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8853211009174314e-06,
+      "loss": 2.2244,
+      "step": 4594
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.878440366972477e-06,
+      "loss": 2.1895,
+      "step": 4595
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.871559633027523e-06,
+      "loss": 2.2617,
+      "step": 4596
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.864678899082569e-06,
+      "loss": 2.2397,
+      "step": 4597
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8577981651376147e-06,
+      "loss": 2.046,
+      "step": 4598
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8509174311926608e-06,
+      "loss": 2.3836,
+      "step": 4599
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8440366972477066e-06,
+      "loss": 2.0547,
+      "step": 4600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8371559633027524e-06,
+      "loss": 2.2102,
+      "step": 4601
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8302752293577983e-06,
+      "loss": 2.1054,
+      "step": 4602
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.823394495412844e-06,
+      "loss": 2.0847,
+      "step": 4603
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8165137614678901e-06,
+      "loss": 2.1168,
+      "step": 4604
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.809633027522936e-06,
+      "loss": 2.1375,
+      "step": 4605
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8027522935779818e-06,
+      "loss": 2.0271,
+      "step": 4606
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7958715596330276e-06,
+      "loss": 2.0371,
+      "step": 4607
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7889908256880735e-06,
+      "loss": 1.9572,
+      "step": 4608
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7821100917431195e-06,
+      "loss": 1.9469,
+      "step": 4609
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7752293577981653e-06,
+      "loss": 2.2403,
+      "step": 4610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7683486238532112e-06,
+      "loss": 1.9193,
+      "step": 4611
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.761467889908257e-06,
+      "loss": 1.9205,
+      "step": 4612
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7545871559633028e-06,
+      "loss": 1.9462,
+      "step": 4613
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7477064220183489e-06,
+      "loss": 2.1121,
+      "step": 4614
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7408256880733947e-06,
+      "loss": 1.8259,
+      "step": 4615
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7339449541284405e-06,
+      "loss": 1.9596,
+      "step": 4616
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7270642201834864e-06,
+      "loss": 1.8657,
+      "step": 4617
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.7201834862385322e-06,
+      "loss": 1.899,
+      "step": 4618
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.713302752293578e-06,
+      "loss": 1.9392,
+      "step": 4619
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.706422018348624e-06,
+      "loss": 1.971,
+      "step": 4620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6995412844036699e-06,
+      "loss": 1.9506,
+      "step": 4621
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6926605504587157e-06,
+      "loss": 1.9314,
+      "step": 4622
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6857798165137616e-06,
+      "loss": 1.9242,
+      "step": 4623
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6788990825688074e-06,
+      "loss": 1.8262,
+      "step": 4624
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6720183486238534e-06,
+      "loss": 1.8676,
+      "step": 4625
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6651376146788993e-06,
+      "loss": 1.6508,
+      "step": 4626
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.658256880733945e-06,
+      "loss": 1.7122,
+      "step": 4627
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.651376146788991e-06,
+      "loss": 1.5661,
+      "step": 4628
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6444954128440367e-06,
+      "loss": 1.7842,
+      "step": 4629
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6376146788990828e-06,
+      "loss": 1.5723,
+      "step": 4630
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6307339449541286e-06,
+      "loss": 1.6815,
+      "step": 4631
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6238532110091745e-06,
+      "loss": 1.5471,
+      "step": 4632
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6169724770642203e-06,
+      "loss": 1.5999,
+      "step": 4633
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6100917431192661e-06,
+      "loss": 1.4725,
+      "step": 4634
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6032110091743122e-06,
+      "loss": 1.4273,
+      "step": 4635
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.596330275229358e-06,
+      "loss": 1.32,
+      "step": 4636
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5894495412844038e-06,
+      "loss": 1.2658,
+      "step": 4637
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5825688073394496e-06,
+      "loss": 1.2394,
+      "step": 4638
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5756880733944955e-06,
+      "loss": 1.28,
+      "step": 4639
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5688073394495413e-06,
+      "loss": 1.2132,
+      "step": 4640
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.5619266055045874e-06,
+      "loss": 2.5671,
+      "step": 4641
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5550458715596332e-06,
+      "loss": 2.535,
+      "step": 4642
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.548165137614679e-06,
+      "loss": 2.3981,
+      "step": 4643
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5412844036697248e-06,
+      "loss": 2.2769,
+      "step": 4644
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5344036697247707e-06,
+      "loss": 2.1835,
+      "step": 4645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5275229357798167e-06,
+      "loss": 2.1997,
+      "step": 4646
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5206422018348625e-06,
+      "loss": 2.1492,
+      "step": 4647
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5137614678899084e-06,
+      "loss": 2.1786,
+      "step": 4648
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5068807339449542e-06,
+      "loss": 2.2895,
+      "step": 4649
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.5e-06,
+      "loss": 2.1751,
+      "step": 4650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.493119266055046e-06,
+      "loss": 2.0646,
+      "step": 4651
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.486238532110092e-06,
+      "loss": 2.1212,
+      "step": 4652
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4793577981651377e-06,
+      "loss": 2.1289,
+      "step": 4653
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4724770642201836e-06,
+      "loss": 2.1273,
+      "step": 4654
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4655963302752294e-06,
+      "loss": 2.0849,
+      "step": 4655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4587155963302754e-06,
+      "loss": 2.0631,
+      "step": 4656
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.4518348623853213e-06,
+      "loss": 2.2046,
+      "step": 4657
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4449541284403671e-06,
+      "loss": 2.1248,
+      "step": 4658
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.438073394495413e-06,
+      "loss": 2.0093,
+      "step": 4659
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4311926605504588e-06,
+      "loss": 2.11,
+      "step": 4660
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4243119266055046e-06,
+      "loss": 2.0498,
+      "step": 4661
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4174311926605506e-06,
+      "loss": 2.1494,
+      "step": 4662
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4105504587155965e-06,
+      "loss": 2.1399,
+      "step": 4663
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.4036697247706423e-06,
+      "loss": 2.0157,
+      "step": 4664
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3967889908256881e-06,
+      "loss": 1.9857,
+      "step": 4665
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.389908256880734e-06,
+      "loss": 2.0235,
+      "step": 4666
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.38302752293578e-06,
+      "loss": 2.0312,
+      "step": 4667
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3761467889908258e-06,
+      "loss": 1.9189,
+      "step": 4668
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3692660550458717e-06,
+      "loss": 1.9059,
+      "step": 4669
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3623853211009175e-06,
+      "loss": 1.8077,
+      "step": 4670
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3555045871559633e-06,
+      "loss": 1.8419,
+      "step": 4671
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3486238532110094e-06,
+      "loss": 1.8378,
+      "step": 4672
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.3417431192660552e-06,
+      "loss": 1.7033,
+      "step": 4673
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.334862385321101e-06,
+      "loss": 1.8382,
+      "step": 4674
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3279816513761469e-06,
+      "loss": 1.7892,
+      "step": 4675
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3211009174311927e-06,
+      "loss": 1.8945,
+      "step": 4676
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3142201834862387e-06,
+      "loss": 1.8737,
+      "step": 4677
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3073394495412846e-06,
+      "loss": 1.6845,
+      "step": 4678
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.3004587155963304e-06,
+      "loss": 1.6384,
+      "step": 4679
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2935779816513762e-06,
+      "loss": 1.6202,
+      "step": 4680
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.286697247706422e-06,
+      "loss": 1.6956,
+      "step": 4681
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.279816513761468e-06,
+      "loss": 1.5889,
+      "step": 4682
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.272935779816514e-06,
+      "loss": 1.4948,
+      "step": 4683
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2660550458715598e-06,
+      "loss": 1.3622,
+      "step": 4684
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2591743119266056e-06,
+      "loss": 1.4763,
+      "step": 4685
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2522935779816514e-06,
+      "loss": 1.3566,
+      "step": 4686
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2454128440366973e-06,
+      "loss": 1.2996,
+      "step": 4687
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2385321100917433e-06,
+      "loss": 1.3646,
+      "step": 4688
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.2316513761467891e-06,
+      "loss": 1.1318,
+      "step": 4689
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.224770642201835e-06,
+      "loss": 1.215,
+      "step": 4690
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2178899082568808e-06,
+      "loss": 2.7271,
+      "step": 4691
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2110091743119266e-06,
+      "loss": 2.3795,
+      "step": 4692
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.2041284403669727e-06,
+      "loss": 2.3648,
+      "step": 4693
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1972477064220185e-06,
+      "loss": 2.3723,
+      "step": 4694
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1903669724770643e-06,
+      "loss": 2.2895,
+      "step": 4695
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1834862385321102e-06,
+      "loss": 2.2202,
+      "step": 4696
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.176605504587156e-06,
+      "loss": 2.2981,
+      "step": 4697
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.169724770642202e-06,
+      "loss": 2.0552,
+      "step": 4698
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1628440366972479e-06,
+      "loss": 2.1438,
+      "step": 4699
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1559633027522937e-06,
+      "loss": 1.9216,
+      "step": 4700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1490825688073395e-06,
+      "loss": 2.1883,
+      "step": 4701
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1422018348623853e-06,
+      "loss": 2.0252,
+      "step": 4702
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1353211009174314e-06,
+      "loss": 2.122,
+      "step": 4703
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1284403669724772e-06,
+      "loss": 2.0475,
+      "step": 4704
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.121559633027523e-06,
+      "loss": 2.0035,
+      "step": 4705
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.1146788990825689e-06,
+      "loss": 2.1689,
+      "step": 4706
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1077981651376147e-06,
+      "loss": 2.1032,
+      "step": 4707
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.1009174311926605e-06,
+      "loss": 2.1527,
+      "step": 4708
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0940366972477066e-06,
+      "loss": 2.1392,
+      "step": 4709
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0871559633027524e-06,
+      "loss": 1.968,
+      "step": 4710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0802752293577982e-06,
+      "loss": 2.1225,
+      "step": 4711
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.073394495412844e-06,
+      "loss": 2.0523,
+      "step": 4712
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.06651376146789e-06,
+      "loss": 2.0247,
+      "step": 4713
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.059633027522936e-06,
+      "loss": 2.0592,
+      "step": 4714
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0527522935779818e-06,
+      "loss": 1.9755,
+      "step": 4715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0458715596330276e-06,
+      "loss": 1.9527,
+      "step": 4716
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0389908256880734e-06,
+      "loss": 1.8312,
+      "step": 4717
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0321100917431193e-06,
+      "loss": 1.9744,
+      "step": 4718
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0252293577981653e-06,
+      "loss": 1.9069,
+      "step": 4719
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0183486238532111e-06,
+      "loss": 1.9633,
+      "step": 4720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.011467889908257e-06,
+      "loss": 2.1377,
+      "step": 4721
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.0045871559633028e-06,
+      "loss": 1.867,
+      "step": 4722
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.977064220183486e-07,
+      "loss": 1.8918,
+      "step": 4723
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.908256880733947e-07,
+      "loss": 1.7963,
+      "step": 4724
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.839449541284405e-07,
+      "loss": 1.7204,
+      "step": 4725
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.770642201834863e-07,
+      "loss": 1.7415,
+      "step": 4726
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.701834862385322e-07,
+      "loss": 1.7839,
+      "step": 4727
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.63302752293578e-07,
+      "loss": 1.7711,
+      "step": 4728
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.564220183486238e-07,
+      "loss": 1.5701,
+      "step": 4729
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.495412844036698e-07,
+      "loss": 1.708,
+      "step": 4730
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.426605504587157e-07,
+      "loss": 1.6652,
+      "step": 4731
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.357798165137615e-07,
+      "loss": 1.43,
+      "step": 4732
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.288990825688074e-07,
+      "loss": 1.7157,
+      "step": 4733
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.220183486238533e-07,
+      "loss": 1.4685,
+      "step": 4734
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.151376146788991e-07,
+      "loss": 1.3968,
+      "step": 4735
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.082568807339451e-07,
+      "loss": 1.3868,
+      "step": 4736
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.013761467889909e-07,
+      "loss": 1.3261,
+      "step": 4737
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.944954128440367e-07,
+      "loss": 1.3978,
+      "step": 4738
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.876146788990827e-07,
+      "loss": 1.2001,
+      "step": 4739
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.807339449541285e-07,
+      "loss": 1.114,
+      "step": 4740
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.738532110091744e-07,
+      "loss": 2.5616,
+      "step": 4741
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.669724770642203e-07,
+      "loss": 2.6872,
+      "step": 4742
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.600917431192661e-07,
+      "loss": 2.2574,
+      "step": 4743
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.53211009174312e-07,
+      "loss": 2.23,
+      "step": 4744
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.463302752293579e-07,
+      "loss": 2.167,
+      "step": 4745
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.394495412844037e-07,
+      "loss": 2.0943,
+      "step": 4746
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.325688073394496e-07,
+      "loss": 2.2791,
+      "step": 4747
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.256880733944955e-07,
+      "loss": 2.2147,
+      "step": 4748
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.188073394495414e-07,
+      "loss": 2.0082,
+      "step": 4749
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.119266055045872e-07,
+      "loss": 2.2273,
+      "step": 4750
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.050458715596331e-07,
+      "loss": 2.1166,
+      "step": 4751
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.98165137614679e-07,
+      "loss": 2.1926,
+      "step": 4752
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.912844036697248e-07,
+      "loss": 1.9687,
+      "step": 4753
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.844036697247707e-07,
+      "loss": 2.1042,
+      "step": 4754
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.775229357798166e-07,
+      "loss": 1.9864,
+      "step": 4755
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.706422018348624e-07,
+      "loss": 1.8738,
+      "step": 4756
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.637614678899084e-07,
+      "loss": 1.9464,
+      "step": 4757
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.568807339449542e-07,
+      "loss": 1.939,
+      "step": 4758
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.5e-07,
+      "loss": 2.0584,
+      "step": 4759
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.43119266055046e-07,
+      "loss": 1.9236,
+      "step": 4760
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.362385321100918e-07,
+      "loss": 1.9346,
+      "step": 4761
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.293577981651377e-07,
+      "loss": 2.0143,
+      "step": 4762
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.224770642201836e-07,
+      "loss": 2.0015,
+      "step": 4763
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.155963302752294e-07,
+      "loss": 1.885,
+      "step": 4764
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.087155963302753e-07,
+      "loss": 1.9783,
+      "step": 4765
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.018348623853212e-07,
+      "loss": 1.928,
+      "step": 4766
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.94954128440367e-07,
+      "loss": 2.0102,
+      "step": 4767
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.880733944954129e-07,
+      "loss": 1.857,
+      "step": 4768
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.811926605504587e-07,
+      "loss": 1.8128,
+      "step": 4769
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.743119266055047e-07,
+      "loss": 1.9225,
+      "step": 4770
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.674311926605505e-07,
+      "loss": 1.8057,
+      "step": 4771
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.605504587155963e-07,
+      "loss": 1.9379,
+      "step": 4772
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.536697247706423e-07,
+      "loss": 1.7693,
+      "step": 4773
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.467889908256881e-07,
+      "loss": 1.8097,
+      "step": 4774
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.39908256880734e-07,
+      "loss": 1.8684,
+      "step": 4775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.330275229357799e-07,
+      "loss": 1.7887,
+      "step": 4776
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.261467889908257e-07,
+      "loss": 1.7344,
+      "step": 4777
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.192660550458716e-07,
+      "loss": 1.7416,
+      "step": 4778
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.123853211009175e-07,
+      "loss": 1.7276,
+      "step": 4779
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.055045871559633e-07,
+      "loss": 1.5946,
+      "step": 4780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.986238532110092e-07,
+      "loss": 1.4952,
+      "step": 4781
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.917431192660551e-07,
+      "loss": 1.6927,
+      "step": 4782
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.84862385321101e-07,
+      "loss": 1.4342,
+      "step": 4783
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.779816513761468e-07,
+      "loss": 1.4477,
+      "step": 4784
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.711009174311927e-07,
+      "loss": 1.2983,
+      "step": 4785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.642201834862386e-07,
+      "loss": 1.4106,
+      "step": 4786
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.573394495412844e-07,
+      "loss": 1.4201,
+      "step": 4787
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.504587155963303e-07,
+      "loss": 1.3049,
+      "step": 4788
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.435779816513762e-07,
+      "loss": 1.0723,
+      "step": 4789
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.36697247706422e-07,
+      "loss": 1.2267,
+      "step": 4790
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.29816513761468e-07,
+      "loss": 2.7615,
+      "step": 4791
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.229357798165138e-07,
+      "loss": 2.4917,
+      "step": 4792
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.160550458715596e-07,
+      "loss": 2.281,
+      "step": 4793
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.091743119266056e-07,
+      "loss": 2.3855,
+      "step": 4794
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.022935779816514e-07,
+      "loss": 2.2936,
+      "step": 4795
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.954128440366973e-07,
+      "loss": 2.1441,
+      "step": 4796
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.885321100917432e-07,
+      "loss": 2.1928,
+      "step": 4797
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.81651376146789e-07,
+      "loss": 2.2242,
+      "step": 4798
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.747706422018349e-07,
+      "loss": 2.0921,
+      "step": 4799
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.6788990825688077e-07,
+      "loss": 2.1559,
+      "step": 4800
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.6100917431192665e-07,
+      "loss": 2.0905,
+      "step": 4801
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.5412844036697253e-07,
+      "loss": 2.0648,
+      "step": 4802
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.4724770642201836e-07,
+      "loss": 1.9818,
+      "step": 4803
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.4036697247706425e-07,
+      "loss": 2.1928,
+      "step": 4804
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3348623853211013e-07,
+      "loss": 2.0196,
+      "step": 4805
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.26605504587156e-07,
+      "loss": 2.1267,
+      "step": 4806
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1972477064220185e-07,
+      "loss": 2.026,
+      "step": 4807
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1284403669724773e-07,
+      "loss": 2.0072,
+      "step": 4808
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.059633027522936e-07,
+      "loss": 2.0409,
+      "step": 4809
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.990825688073395e-07,
+      "loss": 2.0851,
+      "step": 4810
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9220183486238533e-07,
+      "loss": 1.9425,
+      "step": 4811
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.853211009174312e-07,
+      "loss": 2.0059,
+      "step": 4812
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.784403669724771e-07,
+      "loss": 1.9822,
+      "step": 4813
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.71559633027523e-07,
+      "loss": 1.8247,
+      "step": 4814
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6467889908256886e-07,
+      "loss": 2.0127,
+      "step": 4815
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.577981651376147e-07,
+      "loss": 1.8805,
+      "step": 4816
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.509174311926606e-07,
+      "loss": 1.9393,
+      "step": 4817
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4403669724770646e-07,
+      "loss": 1.9636,
+      "step": 4818
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3715596330275234e-07,
+      "loss": 1.9148,
+      "step": 4819
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3027522935779817e-07,
+      "loss": 1.9247,
+      "step": 4820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2339449541284406e-07,
+      "loss": 1.9841,
+      "step": 4821
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1651376146788994e-07,
+      "loss": 1.8273,
+      "step": 4822
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.096330275229358e-07,
+      "loss": 1.8113,
+      "step": 4823
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0275229357798165e-07,
+      "loss": 1.7906,
+      "step": 4824
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9587155963302754e-07,
+      "loss": 1.828,
+      "step": 4825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.889908256880734e-07,
+      "loss": 1.7527,
+      "step": 4826
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.821100917431193e-07,
+      "loss": 1.7711,
+      "step": 4827
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7522935779816514e-07,
+      "loss": 1.6235,
+      "step": 4828
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.68348623853211e-07,
+      "loss": 1.6509,
+      "step": 4829
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.614678899082569e-07,
+      "loss": 1.6061,
+      "step": 4830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.545871559633028e-07,
+      "loss": 1.7435,
+      "step": 4831
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4770642201834867e-07,
+      "loss": 1.6905,
+      "step": 4832
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.408256880733945e-07,
+      "loss": 1.4257,
+      "step": 4833
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3394495412844038e-07,
+      "loss": 1.5441,
+      "step": 4834
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2706422018348627e-07,
+      "loss": 1.5175,
+      "step": 4835
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2018348623853212e-07,
+      "loss": 1.3075,
+      "step": 4836
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.13302752293578e-07,
+      "loss": 1.358,
+      "step": 4837
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0642201834862386e-07,
+      "loss": 1.2172,
+      "step": 4838
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9954128440366975e-07,
+      "loss": 1.2035,
+      "step": 4839
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.926605504587156e-07,
+      "loss": 1.1218,
+      "step": 4840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.857798165137615e-07,
+      "loss": 2.5489,
+      "step": 4841
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7889908256880735e-07,
+      "loss": 2.3888,
+      "step": 4842
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7201834862385323e-07,
+      "loss": 2.0508,
+      "step": 4843
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6513761467889909e-07,
+      "loss": 2.0899,
+      "step": 4844
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5825688073394497e-07,
+      "loss": 2.1755,
+      "step": 4845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5137614678899083e-07,
+      "loss": 2.1044,
+      "step": 4846
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.444954128440367e-07,
+      "loss": 2.002,
+      "step": 4847
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3761467889908257e-07,
+      "loss": 1.9641,
+      "step": 4848
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3073394495412845e-07,
+      "loss": 2.089,
+      "step": 4849
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2385321100917433e-07,
+      "loss": 1.9465,
+      "step": 4850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1697247706422019e-07,
+      "loss": 2.0243,
+      "step": 4851
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1009174311926606e-07,
+      "loss": 1.9055,
+      "step": 4852
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0321100917431193e-07,
+      "loss": 1.9511,
+      "step": 4853
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.63302752293578e-08,
+      "loss": 1.8217,
+      "step": 4854
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.944954128440367e-08,
+      "loss": 1.6875,
+      "step": 4855
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.256880733944954e-08,
+      "loss": 1.5827,
+      "step": 4856
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.568807339449541e-08,
+      "loss": 1.5101,
+      "step": 4857
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.880733944954128e-08,
+      "loss": 1.3348,
+      "step": 4858
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.192660550458717e-08,
+      "loss": 1.2372,
+      "step": 4859
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.504587155963303e-08,
+      "loss": 1.1552,
+      "step": 4860
+    },
+    {
+      "epoch": 3.0,
+      "step": 4860,
+      "total_flos": 0.0,
+      "train_loss": 2.8622734340865916,
+      "train_runtime": 68135.2194,
+      "train_samples_per_second": 9.131,
+      "train_steps_per_second": 0.071
+    }
+  ],
+  "max_steps": 4860,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}