diff --git "a/VQA_lora_PMC_LLaMA_PMCCLIP/choice/checkpoint-4000/trainer_state.json" "b/VQA_lora_PMC_LLaMA_PMCCLIP/choice/checkpoint-4000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/VQA_lora_PMC_LLaMA_PMCCLIP/choice/checkpoint-4000/trainer_state.json"
@@ -0,0 +1,24080 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.8933092224231465,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 1.0298,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 0.9204,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 0.9751,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 0.9744,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 0.9722,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 0.8623,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 0.9453,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 0.9832,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 0.9021,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 0.9519,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 0.98,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.9255,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 0.9575,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 0.9324,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.8699,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.56e-06,
+      "loss": 0.8057,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 0.8328,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-06,
+      "loss": 0.7427,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.04e-06,
+      "loss": 0.8071,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 0.8086,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 0.7517,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 0.6747,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 0.6746,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 0.6533,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.6287,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 0.5751,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.32e-06,
+      "loss": 0.5809,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.48e-06,
+      "loss": 0.5381,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 0.4198,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.3943,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 0.401,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.12e-06,
+      "loss": 0.3701,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.28e-06,
+      "loss": 0.364,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 0.3278,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 0.303,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 0.2937,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.92e-06,
+      "loss": 0.2938,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.08e-06,
+      "loss": 0.2686,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.24e-06,
+      "loss": 0.2626,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 0.2544,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 0.2406,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 0.2302,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.88e-06,
+      "loss": 0.2073,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.04e-06,
+      "loss": 0.2256,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 0.2046,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 0.1965,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 0.1911,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 0.2046,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 0.17,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.2005,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.16e-06,
+      "loss": 0.1688,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.32e-06,
+      "loss": 0.1754,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.48e-06,
+      "loss": 0.1828,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.64e-06,
+      "loss": 0.1679,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 0.1667,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.96e-06,
+      "loss": 0.1599,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.12e-06,
+      "loss": 0.207,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 0.1746,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 0.1826,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 0.1906,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 0.1665,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.1638,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.008e-05,
+      "loss": 0.1688,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.024e-05,
+      "loss": 0.1682,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.04e-05,
+      "loss": 0.1606,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.056e-05,
+      "loss": 0.1596,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.072e-05,
+      "loss": 0.1528,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0880000000000001e-05,
+      "loss": 0.1633,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 0.1566,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.1474,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.136e-05,
+      "loss": 0.1806,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.152e-05,
+      "loss": 0.1834,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.168e-05,
+      "loss": 0.1748,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.184e-05,
+      "loss": 0.1515,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.1592,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.216e-05,
+      "loss": 0.1494,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.232e-05,
+      "loss": 0.1744,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.248e-05,
+      "loss": 0.1516,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2640000000000001e-05,
+      "loss": 0.157,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.1559,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2960000000000001e-05,
+      "loss": 0.1463,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.1542,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 0.1552,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3440000000000002e-05,
+      "loss": 0.1584,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.1644,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.376e-05,
+      "loss": 0.144,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.392e-05,
+      "loss": 0.1672,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.408e-05,
+      "loss": 0.1444,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 0.1522,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4400000000000001e-05,
+      "loss": 0.1572,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 0.1475,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4720000000000001e-05,
+      "loss": 0.1471,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4880000000000002e-05,
+      "loss": 0.1599,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.1643,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5200000000000002e-05,
+      "loss": 0.1558,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 0.1447,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.552e-05,
+      "loss": 0.1592,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5680000000000002e-05,
+      "loss": 0.1457,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.584e-05,
+      "loss": 0.1504,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.1599,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.616e-05,
+      "loss": 0.1507,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.632e-05,
+      "loss": 0.1455,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.648e-05,
+      "loss": 0.1476,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.664e-05,
+      "loss": 0.1463,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.152,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.696e-05,
+      "loss": 0.1447,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7120000000000002e-05,
+      "loss": 0.151,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.728e-05,
+      "loss": 0.1436,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 0.1431,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.76e-05,
+      "loss": 0.1568,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 0.1468,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.792e-05,
+      "loss": 0.1602,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8080000000000003e-05,
+      "loss": 0.1486,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.824e-05,
+      "loss": 0.1358,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8400000000000003e-05,
+      "loss": 0.1535,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8560000000000002e-05,
+      "loss": 0.151,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8720000000000004e-05,
+      "loss": 0.1479,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8880000000000002e-05,
+      "loss": 0.1508,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904e-05,
+      "loss": 0.1454,
+      "step": 119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.1486,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.936e-05,
+      "loss": 0.1463,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 0.1381,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.968e-05,
+      "loss": 0.1397,
+      "step": 123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840000000000003e-05,
+      "loss": 0.1544,
+      "step": 124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.1391,
+      "step": 125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999996947880165e-05,
+      "loss": 0.1518,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998779152252e-05,
+      "loss": 0.1593,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999972530932657e-05,
+      "loss": 0.1462,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999951166119886e-05,
+      "loss": 0.1394,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999923697097254e-05,
+      "loss": 0.1452,
+      "step": 130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999890123881527e-05,
+      "loss": 0.1561,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999850446493196e-05,
+      "loss": 0.1424,
+      "step": 132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999980466495649e-05,
+      "loss": 0.1604,
+      "step": 133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999975277929934e-05,
+      "loss": 0.1469,
+      "step": 134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999694789553435e-05,
+      "loss": 0.1378,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999630695754163e-05,
+      "loss": 0.1548,
+      "step": 136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999956049794065e-05,
+      "loss": 0.1415,
+      "step": 137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999484196155748e-05,
+      "loss": 0.1346,
+      "step": 138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999401790446033e-05,
+      "loss": 0.1414,
+      "step": 139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999313280861806e-05,
+      "loss": 0.1449,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999218667457096e-05,
+      "loss": 0.1497,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999911795028966e-05,
+      "loss": 0.1439,
+      "step": 142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999011129420976e-05,
+      "loss": 0.1366,
+      "step": 143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999889820491625e-05,
+      "loss": 0.1476,
+      "step": 144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998779176844414e-05,
+      "loss": 0.1363,
+      "step": 145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998654045278126e-05,
+      "loss": 0.145,
+      "step": 146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998522810293766e-05,
+      "loss": 0.1465,
+      "step": 147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998385471971448e-05,
+      "loss": 0.1445,
+      "step": 148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998242030395004e-05,
+      "loss": 0.1526,
+      "step": 149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998092485651994e-05,
+      "loss": 0.1418,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997936837833704e-05,
+      "loss": 0.153,
+      "step": 151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997775087035148e-05,
+      "loss": 0.1432,
+      "step": 152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997607233355055e-05,
+      "loss": 0.1419,
+      "step": 153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997433276895897e-05,
+      "loss": 0.1501,
+      "step": 154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997253217763853e-05,
+      "loss": 0.1396,
+      "step": 155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999706705606884e-05,
+      "loss": 0.1376,
+      "step": 156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996874791924493e-05,
+      "loss": 0.1429,
+      "step": 157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996676425448175e-05,
+      "loss": 0.1421,
+      "step": 158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996471956760978e-05,
+      "loss": 0.1301,
+      "step": 159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999626138598771e-05,
+      "loss": 0.1325,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996044713256907e-05,
+      "loss": 0.1332,
+      "step": 161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995821938700835e-05,
+      "loss": 0.1406,
+      "step": 162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999559306245548e-05,
+      "loss": 0.1241,
+      "step": 163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999535808466055e-05,
+      "loss": 0.1405,
+      "step": 164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995117005459484e-05,
+      "loss": 0.141,
+      "step": 165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994869824999448e-05,
+      "loss": 0.1424,
+      "step": 166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994616543431315e-05,
+      "loss": 0.1431,
+      "step": 167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999435716090971e-05,
+      "loss": 0.1393,
+      "step": 168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994091677592945e-05,
+      "loss": 0.1507,
+      "step": 169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993820093643097e-05,
+      "loss": 0.135,
+      "step": 170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999354240922594e-05,
+      "loss": 0.1478,
+      "step": 171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993258624510977e-05,
+      "loss": 0.1472,
+      "step": 172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999296873967144e-05,
+      "loss": 0.1352,
+      "step": 173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992672754884276e-05,
+      "loss": 0.142,
+      "step": 174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999237067033017e-05,
+      "loss": 0.137,
+      "step": 175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992062486193517e-05,
+      "loss": 0.136,
+      "step": 176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999174820266244e-05,
+      "loss": 0.1342,
+      "step": 177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991427819928784e-05,
+      "loss": 0.133,
+      "step": 178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991101338188125e-05,
+      "loss": 0.1361,
+      "step": 179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990768757639747e-05,
+      "loss": 0.1307,
+      "step": 180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990430078486665e-05,
+      "loss": 0.1429,
+      "step": 181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990085300935627e-05,
+      "loss": 0.1432,
+      "step": 182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989734425197084e-05,
+      "loss": 0.1443,
+      "step": 183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998937745148522e-05,
+      "loss": 0.1457,
+      "step": 184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989014380017943e-05,
+      "loss": 0.1449,
+      "step": 185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998864521101688e-05,
+      "loss": 0.1274,
+      "step": 186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988269944707377e-05,
+      "loss": 0.1499,
+      "step": 187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987888581318515e-05,
+      "loss": 0.1405,
+      "step": 188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987501121083076e-05,
+      "loss": 0.1424,
+      "step": 189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998710756423758e-05,
+      "loss": 0.1389,
+      "step": 190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986707911022263e-05,
+      "loss": 0.1347,
+      "step": 191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986302161681087e-05,
+      "loss": 0.1371,
+      "step": 192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985890316461728e-05,
+      "loss": 0.1423,
+      "step": 193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985472375615582e-05,
+      "loss": 0.1377,
+      "step": 194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985048339397775e-05,
+      "loss": 0.1472,
+      "step": 195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998461820806715e-05,
+      "loss": 0.1361,
+      "step": 196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984181981886266e-05,
+      "loss": 0.1396,
+      "step": 197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983739661121406e-05,
+      "loss": 0.1544,
+      "step": 198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983291246042574e-05,
+      "loss": 0.139,
+      "step": 199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982836736923495e-05,
+      "loss": 0.1371,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998237613404161e-05,
+      "loss": 0.1432,
+      "step": 201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981909437678082e-05,
+      "loss": 0.1384,
+      "step": 202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981436648117797e-05,
+      "loss": 0.145,
+      "step": 203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980957765649354e-05,
+      "loss": 0.1466,
+      "step": 204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980472790565075e-05,
+      "loss": 0.1358,
+      "step": 205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979981723161e-05,
+      "loss": 0.1363,
+      "step": 206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997948456373689e-05,
+      "loss": 0.1383,
+      "step": 207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997898131259622e-05,
+      "loss": 0.1466,
+      "step": 208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978471970046186e-05,
+      "loss": 0.1229,
+      "step": 209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997795653639771e-05,
+      "loss": 0.1377,
+      "step": 210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997743501196542e-05,
+      "loss": 0.1372,
+      "step": 211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976907397067664e-05,
+      "loss": 0.1281,
+      "step": 212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976373692026518e-05,
+      "loss": 0.1339,
+      "step": 213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997583389716776e-05,
+      "loss": 0.1501,
+      "step": 214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975288012820903e-05,
+      "loss": 0.1431,
+      "step": 215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974736039319156e-05,
+      "loss": 0.1452,
+      "step": 216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997417797699947e-05,
+      "loss": 0.1444,
+      "step": 217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973613826202493e-05,
+      "loss": 0.1337,
+      "step": 218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973043587272592e-05,
+      "loss": 0.146,
+      "step": 219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972467260557864e-05,
+      "loss": 0.1259,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971884846410104e-05,
+      "loss": 0.1317,
+      "step": 221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997129634518484e-05,
+      "loss": 0.1436,
+      "step": 222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99707017572413e-05,
+      "loss": 0.1375,
+      "step": 223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997010108294244e-05,
+      "loss": 0.1595,
+      "step": 224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996949432265492e-05,
+      "loss": 0.1358,
+      "step": 225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968881476749126e-05,
+      "loss": 0.138,
+      "step": 226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968262545599152e-05,
+      "loss": 0.1472,
+      "step": 227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967637529582806e-05,
+      "loss": 0.1484,
+      "step": 228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967006429081616e-05,
+      "loss": 0.1444,
+      "step": 229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966369244480825e-05,
+      "loss": 0.1458,
+      "step": 230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965725976169374e-05,
+      "loss": 0.1374,
+      "step": 231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965076624539944e-05,
+      "loss": 0.1327,
+      "step": 232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99644211899889e-05,
+      "loss": 0.1378,
+      "step": 233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963759672916344e-05,
+      "loss": 0.1417,
+      "step": 234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963092073726083e-05,
+      "loss": 0.1329,
+      "step": 235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996241839282563e-05,
+      "loss": 0.14,
+      "step": 236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996173863062622e-05,
+      "loss": 0.1352,
+      "step": 237
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961052787542794e-05,
+      "loss": 0.1329,
+      "step": 238
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996036086399401e-05,
+      "loss": 0.1408,
+      "step": 239
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959662860402228e-05,
+      "loss": 0.1334,
+      "step": 240
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958958777193534e-05,
+      "loss": 0.1475,
+      "step": 241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9958248614797712e-05,
+      "loss": 0.1371,
+      "step": 242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957532373648263e-05,
+      "loss": 0.1412,
+      "step": 243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99568100541824e-05,
+      "loss": 0.1491,
+      "step": 244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956081656841045e-05,
+      "loss": 0.1326,
+      "step": 245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995534718206882e-05,
+      "loss": 0.1198,
+      "step": 246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954606630314082e-05,
+      "loss": 0.1376,
+      "step": 247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995386000202887e-05,
+      "loss": 0.1307,
+      "step": 248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953107297668946e-05,
+      "loss": 0.1423,
+      "step": 249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952348517693774e-05,
+      "loss": 0.1488,
+      "step": 250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951583662566543e-05,
+      "loss": 0.1397,
+      "step": 251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995081273275413e-05,
+      "loss": 0.1367,
+      "step": 252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995003572872713e-05,
+      "loss": 0.1416,
+      "step": 253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994925265095985e-05,
+      "loss": 0.1292,
+      "step": 254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948463499930297e-05,
+      "loss": 0.1487,
+      "step": 255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947668276120182e-05,
+      "loss": 0.1341,
+      "step": 256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946866980014935e-05,
+      "loss": 0.1412,
+      "step": 257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946059612103684e-05,
+      "loss": 0.1313,
+      "step": 258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945246172879268e-05,
+      "loss": 0.1367,
+      "step": 259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994442666283823e-05,
+      "loss": 0.1387,
+      "step": 260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943601082480812e-05,
+      "loss": 0.1406,
+      "step": 261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942769432310973e-05,
+      "loss": 0.1453,
+      "step": 262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941931712836373e-05,
+      "loss": 0.1405,
+      "step": 263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941087924568377e-05,
+      "loss": 0.1424,
+      "step": 264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994023806802205e-05,
+      "loss": 0.136,
+      "step": 265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939382143716167e-05,
+      "loss": 0.131,
+      "step": 266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938520152173203e-05,
+      "loss": 0.1432,
+      "step": 267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993765209391934e-05,
+      "loss": 0.126,
+      "step": 268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993677796948446e-05,
+      "loss": 0.1476,
+      "step": 269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935897779402152e-05,
+      "loss": 0.1395,
+      "step": 270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935011524209705e-05,
+      "loss": 0.141,
+      "step": 271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934119204448107e-05,
+      "loss": 0.1368,
+      "step": 272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933220820662053e-05,
+      "loss": 0.1305,
+      "step": 273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993231637339994e-05,
+      "loss": 0.1359,
+      "step": 274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993140586321386e-05,
+      "loss": 0.1334,
+      "step": 275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930489290659615e-05,
+      "loss": 0.1371,
+      "step": 276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929566656296696e-05,
+      "loss": 0.146,
+      "step": 277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992863796068831e-05,
+      "loss": 0.1263,
+      "step": 278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927703204401353e-05,
+      "loss": 0.1303,
+      "step": 279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926762388006416e-05,
+      "loss": 0.1286,
+      "step": 280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925815512077803e-05,
+      "loss": 0.124,
+      "step": 281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924862577193508e-05,
+      "loss": 0.1396,
+      "step": 282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923903583935226e-05,
+      "loss": 0.1276,
+      "step": 283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922938532888342e-05,
+      "loss": 0.1356,
+      "step": 284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921967424641957e-05,
+      "loss": 0.1239,
+      "step": 285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920990259788854e-05,
+      "loss": 0.1454,
+      "step": 286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920007038925516e-05,
+      "loss": 0.1382,
+      "step": 287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919017762652125e-05,
+      "loss": 0.1361,
+      "step": 288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991802243157256e-05,
+      "loss": 0.1402,
+      "step": 289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99170210462944e-05,
+      "loss": 0.136,
+      "step": 290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916013607428908e-05,
+      "loss": 0.1323,
+      "step": 291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991500011559105e-05,
+      "loss": 0.1374,
+      "step": 292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913980571399486e-05,
+      "loss": 0.1462,
+      "step": 293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991295497547657e-05,
+      "loss": 0.1473,
+      "step": 294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911923328448353e-05,
+      "loss": 0.1373,
+      "step": 295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910885630944575e-05,
+      "loss": 0.1389,
+      "step": 296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.990984188359867e-05,
+      "loss": 0.1286,
+      "step": 297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908792087047766e-05,
+      "loss": 0.1353,
+      "step": 298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907736241932692e-05,
+      "loss": 0.1392,
+      "step": 299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990667434889795e-05,
+      "loss": 0.1392,
+      "step": 300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990560640859175e-05,
+      "loss": 0.1387,
+      "step": 301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990453242166599e-05,
+      "loss": 0.1331,
+      "step": 302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903452388776253e-05,
+      "loss": 0.1307,
+      "step": 303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902366310581823e-05,
+      "loss": 0.1544,
+      "step": 304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901274187745663e-05,
+      "loss": 0.1432,
+      "step": 305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900176020934437e-05,
+      "loss": 0.125,
+      "step": 306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899071810818485e-05,
+      "loss": 0.1406,
+      "step": 307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897961558071843e-05,
+      "loss": 0.1508,
+      "step": 308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896845263372242e-05,
+      "loss": 0.1356,
+      "step": 309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895722927401094e-05,
+      "loss": 0.148,
+      "step": 310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894594550843494e-05,
+      "loss": 0.1277,
+      "step": 311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893460134388237e-05,
+      "loss": 0.1229,
+      "step": 312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892319678727797e-05,
+      "loss": 0.1344,
+      "step": 313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989117318455833e-05,
+      "loss": 0.1337,
+      "step": 314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989002065257969e-05,
+      "loss": 0.1319,
+      "step": 315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888862083495407e-05,
+      "loss": 0.1394,
+      "step": 316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98876974780127e-05,
+      "loss": 0.1255,
+      "step": 317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886526836842468e-05,
+      "loss": 0.1262,
+      "step": 318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885350160699305e-05,
+      "loss": 0.1372,
+      "step": 319
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988416745030148e-05,
+      "loss": 0.1415,
+      "step": 320
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988297870637095e-05,
+      "loss": 0.1318,
+      "step": 321
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881783929633344e-05,
+      "loss": 0.1377,
+      "step": 322
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880583120817992e-05,
+      "loss": 0.1372,
+      "step": 323
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987937628065789e-05,
+      "loss": 0.1312,
+      "step": 324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878163409889727e-05,
+      "loss": 0.1359,
+      "step": 325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876944509253867e-05,
+      "loss": 0.1401,
+      "step": 326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875719579494357e-05,
+      "loss": 0.1416,
+      "step": 327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987448862135892e-05,
+      "loss": 0.1401,
+      "step": 328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873251635598966e-05,
+      "loss": 0.1351,
+      "step": 329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872008622969576e-05,
+      "loss": 0.1404,
+      "step": 330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987075958422952e-05,
+      "loss": 0.1346,
+      "step": 331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986950452014124e-05,
+      "loss": 0.142,
+      "step": 332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868243431470857e-05,
+      "loss": 0.1389,
+      "step": 333
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866976318988165e-05,
+      "loss": 0.1404,
+      "step": 334
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865703183466645e-05,
+      "loss": 0.142,
+      "step": 335
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986442402568345e-05,
+      "loss": 0.1281,
+      "step": 336
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863138846419408e-05,
+      "loss": 0.1515,
+      "step": 337
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986184764645902e-05,
+      "loss": 0.1444,
+      "step": 338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860550426590465e-05,
+      "loss": 0.1414,
+      "step": 339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859247187605604e-05,
+      "loss": 0.1412,
+      "step": 340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985793793029996e-05,
+      "loss": 0.1412,
+      "step": 341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985662265547274e-05,
+      "loss": 0.1288,
+      "step": 342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855301363926807e-05,
+      "loss": 0.1305,
+      "step": 343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985397405646872e-05,
+      "loss": 0.1469,
+      "step": 344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98526407339087e-05,
+      "loss": 0.1363,
+      "step": 345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851301397060633e-05,
+      "loss": 0.1232,
+      "step": 346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984995604674208e-05,
+      "loss": 0.1393,
+      "step": 347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848604683774288e-05,
+      "loss": 0.1438,
+      "step": 348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984724730898215e-05,
+      "loss": 0.1386,
+      "step": 349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845883923194243e-05,
+      "loss": 0.1411,
+      "step": 350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984451452724281e-05,
+      "loss": 0.1331,
+      "step": 351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843139121963766e-05,
+      "loss": 0.1418,
+      "step": 352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984175770819669e-05,
+      "loss": 0.1364,
+      "step": 353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840370286784826e-05,
+      "loss": 0.1371,
+      "step": 354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838976858575095e-05,
+      "loss": 0.1411,
+      "step": 355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837577424418074e-05,
+      "loss": 0.1441,
+      "step": 356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836171985168018e-05,
+      "loss": 0.1363,
+      "step": 357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834760541682834e-05,
+      "loss": 0.1481,
+      "step": 358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98333430948241e-05,
+      "loss": 0.136,
+      "step": 359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983191964545707e-05,
+      "loss": 0.1358,
+      "step": 360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830490194450642e-05,
+      "loss": 0.1291,
+      "step": 361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829054742677387e-05,
+      "loss": 0.1516,
+      "step": 362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827613291013546e-05,
+      "loss": 0.1387,
+      "step": 363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982616584033901e-05,
+      "loss": 0.123,
+      "step": 364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824712391537337e-05,
+      "loss": 0.144,
+      "step": 365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982325294549575e-05,
+      "loss": 0.1345,
+      "step": 366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982178750310513e-05,
+      "loss": 0.1412,
+      "step": 367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820316065260017e-05,
+      "loss": 0.1336,
+      "step": 368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981883863285861e-05,
+      "loss": 0.1321,
+      "step": 369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817355206802777e-05,
+      "loss": 0.1327,
+      "step": 370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815865787998026e-05,
+      "loss": 0.1412,
+      "step": 371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981437037735354e-05,
+      "loss": 0.14,
+      "step": 372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812868975782152e-05,
+      "loss": 0.1369,
+      "step": 373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811361584200356e-05,
+      "loss": 0.1437,
+      "step": 374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809848203528298e-05,
+      "loss": 0.1284,
+      "step": 375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808328834689778e-05,
+      "loss": 0.137,
+      "step": 376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806803478612263e-05,
+      "loss": 0.1434,
+      "step": 377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805272136226857e-05,
+      "loss": 0.131,
+      "step": 378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980373480846834e-05,
+      "loss": 0.1464,
+      "step": 379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802191496275123e-05,
+      "loss": 0.1343,
+      "step": 380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980064220058929e-05,
+      "loss": 0.1432,
+      "step": 381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979908692235656e-05,
+      "loss": 0.1365,
+      "step": 382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797525662526318e-05,
+      "loss": 0.1466,
+      "step": 383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979595842205159e-05,
+      "loss": 0.1482,
+      "step": 384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979438520188906e-05,
+      "loss": 0.1427,
+      "step": 385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979280600299906e-05,
+      "loss": 0.1369,
+      "step": 386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979122082634557e-05,
+      "loss": 0.1392,
+      "step": 387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789629672896215e-05,
+      "loss": 0.134,
+      "step": 388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978803254362228e-05,
+      "loss": 0.1294,
+      "step": 389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786429439498688e-05,
+      "loss": 0.1398,
+      "step": 390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784820361504015e-05,
+      "loss": 0.1363,
+      "step": 391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783205310620474e-05,
+      "loss": 0.1408,
+      "step": 392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978158428783394e-05,
+      "loss": 0.1506,
+      "step": 393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779957294133915e-05,
+      "loss": 0.1341,
+      "step": 394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778324330513563e-05,
+      "loss": 0.1407,
+      "step": 395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776685397969678e-05,
+      "loss": 0.1358,
+      "step": 396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775040497502707e-05,
+      "loss": 0.1329,
+      "step": 397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773389630116737e-05,
+      "loss": 0.1405,
+      "step": 398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771732796819492e-05,
+      "loss": 0.1407,
+      "step": 399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977006999862235e-05,
+      "loss": 0.1332,
+      "step": 400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768401236540318e-05,
+      "loss": 0.1359,
+      "step": 401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766726511592047e-05,
+      "loss": 0.1365,
+      "step": 402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765045824799834e-05,
+      "loss": 0.1336,
+      "step": 403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763359177189612e-05,
+      "loss": 0.1373,
+      "step": 404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761666569790943e-05,
+      "loss": 0.1336,
+      "step": 405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759968003637044e-05,
+      "loss": 0.1423,
+      "step": 406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758263479764753e-05,
+      "loss": 0.1328,
+      "step": 407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756552999214554e-05,
+      "loss": 0.1344,
+      "step": 408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975483656303057e-05,
+      "loss": 0.136,
+      "step": 409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975311417226055e-05,
+      "loss": 0.1424,
+      "step": 410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751385827955886e-05,
+      "loss": 0.1264,
+      "step": 411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749651531171597e-05,
+      "loss": 0.1342,
+      "step": 412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974791128296634e-05,
+      "loss": 0.1339,
+      "step": 413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974616508440241e-05,
+      "loss": 0.1431,
+      "step": 414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744412936545722e-05,
+      "loss": 0.1334,
+      "step": 415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974265484046583e-05,
+      "loss": 0.1397,
+      "step": 416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974089079723592e-05,
+      "loss": 0.1351,
+      "step": 417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739120807932805e-05,
+      "loss": 0.1303,
+      "step": 418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973734487363693e-05,
+      "loss": 0.1358,
+      "step": 419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735562995432365e-05,
+      "loss": 0.1365,
+      "step": 420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733775174406814e-05,
+      "loss": 0.1273,
+      "step": 421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731981411651607e-05,
+      "loss": 0.1353,
+      "step": 422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730181708261694e-05,
+      "loss": 0.1436,
+      "step": 423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728376065335663e-05,
+      "loss": 0.1302,
+      "step": 424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972656448397572e-05,
+      "loss": 0.144,
+      "step": 425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724746965287695e-05,
+      "loss": 0.1558,
+      "step": 426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722923510381048e-05,
+      "loss": 0.1356,
+      "step": 427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972109412036886e-05,
+      "loss": 0.1388,
+      "step": 428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719258796367832e-05,
+      "loss": 0.1359,
+      "step": 429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717417539498295e-05,
+      "loss": 0.1255,
+      "step": 430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715570350884187e-05,
+      "loss": 0.1341,
+      "step": 431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713717231653083e-05,
+      "loss": 0.144,
+      "step": 432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971185818293617e-05,
+      "loss": 0.1313,
+      "step": 433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970999320586826e-05,
+      "loss": 0.1317,
+      "step": 434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970812230158777e-05,
+      "loss": 0.1368,
+      "step": 435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706245471236752e-05,
+      "loss": 0.1377,
+      "step": 436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970436271596087e-05,
+      "loss": 0.1389,
+      "step": 437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702474036909395e-05,
+      "loss": 0.1301,
+      "step": 438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970057943523523e-05,
+      "loss": 0.1367,
+      "step": 439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969867891209488e-05,
+      "loss": 0.1389,
+      "step": 440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696772468648474e-05,
+      "loss": 0.1384,
+      "step": 441
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694860106059744e-05,
+      "loss": 0.1396,
+      "step": 442
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692941825496048e-05,
+      "loss": 0.1356,
+      "step": 443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969101762812835e-05,
+      "loss": 0.1394,
+      "step": 444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689087515131224e-05,
+      "loss": 0.1282,
+      "step": 445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687151487682858e-05,
+      "loss": 0.1464,
+      "step": 446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685209546965048e-05,
+      "loss": 0.1339,
+      "step": 447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968326169416321e-05,
+      "loss": 0.1343,
+      "step": 448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681307930466346e-05,
+      "loss": 0.134,
+      "step": 449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967934825706709e-05,
+      "loss": 0.1222,
+      "step": 450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677382675161668e-05,
+      "loss": 0.1339,
+      "step": 451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675411185949927e-05,
+      "loss": 0.1282,
+      "step": 452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673433790635302e-05,
+      "loss": 0.1391,
+      "step": 453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671450490424842e-05,
+      "loss": 0.14,
+      "step": 454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966946128652921e-05,
+      "loss": 0.1374,
+      "step": 455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966746618016266e-05,
+      "loss": 0.1367,
+      "step": 456
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665465172543045e-05,
+      "loss": 0.1308,
+      "step": 457
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966345826489184e-05,
+      "loss": 0.1293,
+      "step": 458
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96614454584341e-05,
+      "loss": 0.134,
+      "step": 459
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659426754398495e-05,
+      "loss": 0.1393,
+      "step": 460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965740215401729e-05,
+      "loss": 0.1435,
+      "step": 461
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965537165852635e-05,
+      "loss": 0.1341,
+      "step": 462
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653335269165135e-05,
+      "loss": 0.1311,
+      "step": 463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965129298717671e-05,
+      "loss": 0.1369,
+      "step": 464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964924481380773e-05,
+      "loss": 0.1392,
+      "step": 465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647190750308452e-05,
+      "loss": 0.1358,
+      "step": 466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645130797932722e-05,
+      "loss": 0.1338,
+      "step": 467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643064957937986e-05,
+      "loss": 0.1449,
+      "step": 468
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964099323158528e-05,
+      "loss": 0.1323,
+      "step": 469
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963891562013924e-05,
+      "loss": 0.1455,
+      "step": 470
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963683212486809e-05,
+      "loss": 0.1303,
+      "step": 471
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634742747043637e-05,
+      "loss": 0.1438,
+      "step": 472
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632647487941296e-05,
+      "loss": 0.1281,
+      "step": 473
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630546348840064e-05,
+      "loss": 0.1395,
+      "step": 474
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962843933102252e-05,
+      "loss": 0.1442,
+      "step": 475
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962632643577484e-05,
+      "loss": 0.1321,
+      "step": 476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962420766438679e-05,
+      "loss": 0.1433,
+      "step": 477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622083018151713e-05,
+      "loss": 0.1288,
+      "step": 478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961995249836655e-05,
+      "loss": 0.1378,
+      "step": 479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617816106331818e-05,
+      "loss": 0.1333,
+      "step": 480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615673843351622e-05,
+      "loss": 0.1402,
+      "step": 481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961352571073365e-05,
+      "loss": 0.135,
+      "step": 482
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611371709789172e-05,
+      "loss": 0.1393,
+      "step": 483
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960921184183305e-05,
+      "loss": 0.1378,
+      "step": 484
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607046108183712e-05,
+      "loss": 0.1376,
+      "step": 485
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604874510163175e-05,
+      "loss": 0.1412,
+      "step": 486
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602697049097035e-05,
+      "loss": 0.1444,
+      "step": 487
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600513726314464e-05,
+      "loss": 0.133,
+      "step": 488
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959832454314822e-05,
+      "loss": 0.1281,
+      "step": 489
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596129500934627e-05,
+      "loss": 0.1356,
+      "step": 490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593928601013595e-05,
+      "loss": 0.1375,
+      "step": 491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591721844728606e-05,
+      "loss": 0.1311,
+      "step": 492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589509233426716e-05,
+      "loss": 0.1356,
+      "step": 493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587290768458557e-05,
+      "loss": 0.1359,
+      "step": 494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958506645117833e-05,
+      "loss": 0.133,
+      "step": 495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582836282943814e-05,
+      "loss": 0.139,
+      "step": 496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958060026511636e-05,
+      "loss": 0.134,
+      "step": 497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578358399060878e-05,
+      "loss": 0.1316,
+      "step": 498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576110686145867e-05,
+      "loss": 0.1427,
+      "step": 499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573857127743378e-05,
+      "loss": 0.1366,
+      "step": 500
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 0.13298147916793823,
+      "eval_runtime": 711.7389,
+      "eval_samples_per_second": 70.25,
+      "eval_steps_per_second": 2.196,
+      "step": 500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571597725229043e-05,
+      "loss": 0.1479,
+      "step": 501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569332479982045e-05,
+      "loss": 0.1405,
+      "step": 502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567061393385155e-05,
+      "loss": 0.1434,
+      "step": 503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564784466824698e-05,
+      "loss": 0.1327,
+      "step": 504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562501701690558e-05,
+      "loss": 0.1363,
+      "step": 505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560213099376194e-05,
+      "loss": 0.1322,
+      "step": 506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557918661278622e-05,
+      "loss": 0.1384,
+      "step": 507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955561838879842e-05,
+      "loss": 0.1344,
+      "step": 508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553312283339735e-05,
+      "loss": 0.1333,
+      "step": 509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955100034631026e-05,
+      "loss": 0.1276,
+      "step": 510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548682579121266e-05,
+      "loss": 0.1402,
+      "step": 511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546358983187576e-05,
+      "loss": 0.1228,
+      "step": 512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544029559927556e-05,
+      "loss": 0.1378,
+      "step": 513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954169431076315e-05,
+      "loss": 0.1499,
+      "step": 514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953935323711985e-05,
+      "loss": 0.1346,
+      "step": 515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95370063404267e-05,
+      "loss": 0.1431,
+      "step": 516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534653622116307e-05,
+      "loss": 0.1338,
+      "step": 517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532295083624824e-05,
+      "loss": 0.1448,
+      "step": 518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952993072639196e-05,
+      "loss": 0.1358,
+      "step": 519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527560551860972e-05,
+      "loss": 0.1368,
+      "step": 520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525184561478676e-05,
+      "loss": 0.1371,
+      "step": 521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952280275669543e-05,
+      "loss": 0.1352,
+      "step": 522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520415138965148e-05,
+      "loss": 0.1321,
+      "step": 523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518021709745284e-05,
+      "loss": 0.14,
+      "step": 524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515622470496848e-05,
+      "loss": 0.1352,
+      "step": 525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513217422684397e-05,
+      "loss": 0.1277,
+      "step": 526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510806567776023e-05,
+      "loss": 0.1339,
+      "step": 527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508389907243374e-05,
+      "loss": 0.1378,
+      "step": 528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505967442561632e-05,
+      "loss": 0.1306,
+      "step": 529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503539175209535e-05,
+      "loss": 0.1208,
+      "step": 530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501105106669352e-05,
+      "loss": 0.1357,
+      "step": 531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498665238426894e-05,
+      "loss": 0.1425,
+      "step": 532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949621957197152e-05,
+      "loss": 0.1347,
+      "step": 533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493768108796122e-05,
+      "loss": 0.1361,
+      "step": 534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949131085039713e-05,
+      "loss": 0.1393,
+      "step": 535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488847798274516e-05,
+      "loss": 0.1359,
+      "step": 536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948637895393178e-05,
+      "loss": 0.1378,
+      "step": 537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483904318875974e-05,
+      "loss": 0.1241,
+      "step": 538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481423894617668e-05,
+      "loss": 0.1357,
+      "step": 539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947893768267097e-05,
+      "loss": 0.1312,
+      "step": 540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947644568455353e-05,
+      "loss": 0.1447,
+      "step": 541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473947901786518e-05,
+      "loss": 0.1319,
+      "step": 542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471444335894644e-05,
+      "loss": 0.1391,
+      "step": 543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946893498840614e-05,
+      "loss": 0.1405,
+      "step": 544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946641986085278e-05,
+      "loss": 0.1379,
+      "step": 545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946389895476985e-05,
+      "loss": 0.1259,
+      "step": 546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461372271696172e-05,
+      "loss": 0.1427,
+      "step": 547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94588398131741e-05,
+      "loss": 0.1401,
+      "step": 548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456301580749507e-05,
+      "loss": 0.1431,
+      "step": 549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453757575971785e-05,
+      "loss": 0.146,
+      "step": 550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451207800393853e-05,
+      "loss": 0.1436,
+      "step": 551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448652255572166e-05,
+      "loss": 0.1401,
+      "step": 552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446090943066686e-05,
+      "loss": 0.1325,
+      "step": 553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443523864440895e-05,
+      "loss": 0.1338,
+      "step": 554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94409510212618e-05,
+      "loss": 0.1353,
+      "step": 555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438372415099934e-05,
+      "loss": 0.1458,
+      "step": 556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435788047529328e-05,
+      "loss": 0.1394,
+      "step": 557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943319792012755e-05,
+      "loss": 0.1368,
+      "step": 558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430602034475676e-05,
+      "loss": 0.1465,
+      "step": 559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9428000392158295e-05,
+      "loss": 0.1338,
+      "step": 560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942539299476351e-05,
+      "loss": 0.1309,
+      "step": 561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942277984388294e-05,
+      "loss": 0.1389,
+      "step": 562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420160941111716e-05,
+      "loss": 0.1398,
+      "step": 563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941753628804848e-05,
+      "loss": 0.1375,
+      "step": 564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414905886295377e-05,
+      "loss": 0.1359,
+      "step": 565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412269737458077e-05,
+      "loss": 0.1421,
+      "step": 566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409627843145742e-05,
+      "loss": 0.1375,
+      "step": 567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406980204971048e-05,
+      "loss": 0.1286,
+      "step": 568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404326824550176e-05,
+      "loss": 0.1327,
+      "step": 569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940166770350282e-05,
+      "loss": 0.1379,
+      "step": 570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939900284345216e-05,
+      "loss": 0.1349,
+      "step": 571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396332246024897e-05,
+      "loss": 0.1461,
+      "step": 572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393655912851233e-05,
+      "loss": 0.127,
+      "step": 573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390973845564853e-05,
+      "loss": 0.1312,
+      "step": 574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388286045802965e-05,
+      "loss": 0.1363,
+      "step": 575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385592515206263e-05,
+      "loss": 0.137,
+      "step": 576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382893255418948e-05,
+      "loss": 0.1434,
+      "step": 577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93801882680887e-05,
+      "loss": 0.1406,
+      "step": 578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937747755486672e-05,
+      "loss": 0.1439,
+      "step": 579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937476111740769e-05,
+      "loss": 0.1416,
+      "step": 580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372038957369784e-05,
+      "loss": 0.1436,
+      "step": 581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936931107641468e-05,
+      "loss": 0.1383,
+      "step": 582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366577476207533e-05,
+      "loss": 0.1413,
+      "step": 583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363838158417008e-05,
+      "loss": 0.1337,
+      "step": 584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361093124715245e-05,
+      "loss": 0.1282,
+      "step": 585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935834237677788e-05,
+      "loss": 0.1344,
+      "step": 586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935558591628403e-05,
+      "loss": 0.1288,
+      "step": 587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352823744916317e-05,
+      "loss": 0.1353,
+      "step": 588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350055864360825e-05,
+      "loss": 0.1336,
+      "step": 589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934728227630714e-05,
+      "loss": 0.1446,
+      "step": 590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934450298244832e-05,
+      "loss": 0.1396,
+      "step": 591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341717984480924e-05,
+      "loss": 0.1274,
+      "step": 592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933892728410497e-05,
+      "loss": 0.1305,
+      "step": 593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933613088302398e-05,
+      "loss": 0.123,
+      "step": 594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333328782944933e-05,
+      "loss": 0.1343,
+      "step": 595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330520985578303e-05,
+      "loss": 0.1301,
+      "step": 596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932770749263804e-05,
+      "loss": 0.1438,
+      "step": 597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932488830584156e-05,
+      "loss": 0.1348,
+      "step": 598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932206342690977e-05,
+      "loss": 0.1339,
+      "step": 599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319232857567036e-05,
+      "loss": 0.1495,
+      "step": 600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316396599541215e-05,
+      "loss": 0.1369,
+      "step": 601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313554654563614e-05,
+      "loss": 0.133,
+      "step": 602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931070702436904e-05,
+      "loss": 0.1392,
+      "step": 603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307853710695744e-05,
+      "loss": 0.141,
+      "step": 604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930499471528546e-05,
+      "loss": 0.1391,
+      "step": 605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302130039883385e-05,
+      "loss": 0.1316,
+      "step": 606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299259686238184e-05,
+      "loss": 0.1335,
+      "step": 607
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296383656101998e-05,
+      "loss": 0.1406,
+      "step": 608
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929350195123042e-05,
+      "loss": 0.133,
+      "step": 609
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929061457338251e-05,
+      "loss": 0.1392,
+      "step": 610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287721524320797e-05,
+      "loss": 0.1293,
+      "step": 611
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928482280581126e-05,
+      "loss": 0.1357,
+      "step": 612
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281918419623356e-05,
+      "loss": 0.1331,
+      "step": 613
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279008367529983e-05,
+      "loss": 0.1294,
+      "step": 614
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927609265130751e-05,
+      "loss": 0.1317,
+      "step": 615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273171272735764e-05,
+      "loss": 0.1368,
+      "step": 616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927024423359802e-05,
+      "loss": 0.1334,
+      "step": 617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926731153568101e-05,
+      "loss": 0.1321,
+      "step": 618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264373180774928e-05,
+      "loss": 0.1333,
+      "step": 619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926142917067342e-05,
+      "loss": 0.1318,
+      "step": 620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258479507173572e-05,
+      "loss": 0.1283,
+      "step": 621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255524192075933e-05,
+      "loss": 0.1349,
+      "step": 622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252563227184494e-05,
+      "loss": 0.1535,
+      "step": 623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924959661430671e-05,
+      "loss": 0.1313,
+      "step": 624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246624355253464e-05,
+      "loss": 0.1429,
+      "step": 625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243646451839093e-05,
+      "loss": 0.142,
+      "step": 626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240662905881384e-05,
+      "loss": 0.1305,
+      "step": 627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237673719201564e-05,
+      "loss": 0.1295,
+      "step": 628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234678893624303e-05,
+      "loss": 0.1314,
+      "step": 629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923167843097772e-05,
+      "loss": 0.1252,
+      "step": 630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922867233309336e-05,
+      "loss": 0.1313,
+      "step": 631
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225660601806225e-05,
+      "loss": 0.1348,
+      "step": 632
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222643238954746e-05,
+      "loss": 0.1272,
+      "step": 633
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219620246380793e-05,
+      "loss": 0.1353,
+      "step": 634
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921659162592967e-05,
+      "loss": 0.1214,
+      "step": 635
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213557379450124e-05,
+      "loss": 0.1334,
+      "step": 636
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921051750879433e-05,
+      "loss": 0.118,
+      "step": 637
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9207472015817902e-05,
+      "loss": 0.1241,
+      "step": 638
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204420902379872e-05,
+      "loss": 0.134,
+      "step": 639
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920136417034272e-05,
+      "loss": 0.1326,
+      "step": 640
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198301821572348e-05,
+      "loss": 0.1234,
+      "step": 641
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919523385793809e-05,
+      "loss": 0.1414,
+      "step": 642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9192160281312696e-05,
+      "loss": 0.1363,
+      "step": 643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189081093572357e-05,
+      "loss": 0.141,
+      "step": 644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185996296596682e-05,
+      "loss": 0.1265,
+      "step": 645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182905892268704e-05,
+      "loss": 0.1309,
+      "step": 646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179809882474886e-05,
+      "loss": 0.1284,
+      "step": 647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176708269105092e-05,
+      "loss": 0.1384,
+      "step": 648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173601054052637e-05,
+      "loss": 0.1382,
+      "step": 649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917048823921423e-05,
+      "loss": 0.1392,
+      "step": 650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167369826490014e-05,
+      "loss": 0.1371,
+      "step": 651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164245817783536e-05,
+      "loss": 0.1162,
+      "step": 652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916111621500177e-05,
+      "loss": 0.1281,
+      "step": 653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91579810200551e-05,
+      "loss": 0.1434,
+      "step": 654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154840234857324e-05,
+      "loss": 0.1326,
+      "step": 655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915169386132565e-05,
+      "loss": 0.1279,
+      "step": 656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914854190138071e-05,
+      "loss": 0.1379,
+      "step": 657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145384356946522e-05,
+      "loss": 0.125,
+      "step": 658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142221229950535e-05,
+      "loss": 0.1343,
+      "step": 659
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139052522323592e-05,
+      "loss": 0.135,
+      "step": 660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135878235999955e-05,
+      "loss": 0.1342,
+      "step": 661
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132698372917282e-05,
+      "loss": 0.1419,
+      "step": 662
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129512935016638e-05,
+      "loss": 0.1336,
+      "step": 663
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126321924242485e-05,
+      "loss": 0.1422,
+      "step": 664
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123125342542697e-05,
+      "loss": 0.1312,
+      "step": 665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119923191868547e-05,
+      "loss": 0.1418,
+      "step": 666
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91167154741747e-05,
+      "loss": 0.1299,
+      "step": 667
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113502191419225e-05,
+      "loss": 0.1382,
+      "step": 668
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110283345563587e-05,
+      "loss": 0.129,
+      "step": 669
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107058938572645e-05,
+      "loss": 0.142,
+      "step": 670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910382897241466e-05,
+      "loss": 0.1407,
+      "step": 671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100593449061272e-05,
+      "loss": 0.1365,
+      "step": 672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097352370487532e-05,
+      "loss": 0.1339,
+      "step": 673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094105738671864e-05,
+      "loss": 0.138,
+      "step": 674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909085355559609e-05,
+      "loss": 0.1356,
+      "step": 675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908759582324543e-05,
+      "loss": 0.1307,
+      "step": 676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908433254360847e-05,
+      "loss": 0.1342,
+      "step": 677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90810637186772e-05,
+      "loss": 0.1318,
+      "step": 678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907778935044699e-05,
+      "loss": 0.1247,
+      "step": 679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907450944091659e-05,
+      "loss": 0.1454,
+      "step": 680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9071223992088133e-05,
+      "loss": 0.1375,
+      "step": 681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906793300596714e-05,
+      "loss": 0.1313,
+      "step": 682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064636484562513e-05,
+      "loss": 0.1307,
+      "step": 683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061334429886517e-05,
+      "loss": 0.1452,
+      "step": 684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9058026843954813e-05,
+      "loss": 0.1342,
+      "step": 685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054713728786424e-05,
+      "loss": 0.137,
+      "step": 686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905139508640376e-05,
+      "loss": 0.1365,
+      "step": 687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90480709188326e-05,
+      "loss": 0.1418,
+      "step": 688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044741228102093e-05,
+      "loss": 0.1443,
+      "step": 689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904140601624476e-05,
+      "loss": 0.1328,
+      "step": 690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90380652852965e-05,
+      "loss": 0.1277,
+      "step": 691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9034719037296573e-05,
+      "loss": 0.1396,
+      "step": 692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903136727428761e-05,
+      "loss": 0.1246,
+      "step": 693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028009998315603e-05,
+      "loss": 0.1315,
+      "step": 694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024647211429916e-05,
+      "loss": 0.1376,
+      "step": 695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902127891568328e-05,
+      "loss": 0.1355,
+      "step": 696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017905113131774e-05,
+      "loss": 0.14,
+      "step": 697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014525805834854e-05,
+      "loss": 0.128,
+      "step": 698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901114099585533e-05,
+      "loss": 0.132,
+      "step": 699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900775068525937e-05,
+      "loss": 0.1266,
+      "step": 700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004354876116503e-05,
+      "loss": 0.1314,
+      "step": 701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000953570499605e-05,
+      "loss": 0.1245,
+      "step": 702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899754677048492e-05,
+      "loss": 0.1339,
+      "step": 703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994134478152045e-05,
+      "loss": 0.1391,
+      "step": 704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899071669558392e-05,
+      "loss": 0.1356,
+      "step": 705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898729342486684e-05,
+      "loss": 0.1296,
+      "step": 706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983864668090452e-05,
+      "loss": 0.1339,
+      "step": 707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898043042734775e-05,
+      "loss": 0.1449,
+      "step": 708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976990704735082e-05,
+      "loss": 0.1291,
+      "step": 709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973545502352134e-05,
+      "loss": 0.1407,
+      "step": 710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897009482230194e-05,
+      "loss": 0.133,
+      "step": 711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896663866669088e-05,
+      "loss": 0.1184,
+      "step": 712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896317703762867e-05,
+      "loss": 0.1348,
+      "step": 713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959709937228375e-05,
+      "loss": 0.1302,
+      "step": 714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956237367606395e-05,
+      "loss": 0.146,
+      "step": 715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895275933088247e-05,
+      "loss": 0.132,
+      "step": 716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949275829179675e-05,
+      "loss": 0.1353,
+      "step": 717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945786864624428e-05,
+      "loss": 0.1308,
+      "step": 718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894229243934647e-05,
+      "loss": 0.1358,
+      "step": 719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938792555478885e-05,
+      "loss": 0.1349,
+      "step": 720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935287215158087e-05,
+      "loss": 0.1409,
+      "step": 721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893177642052382e-05,
+      "loss": 0.1406,
+      "step": 722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928260173719156e-05,
+      "loss": 0.1236,
+      "step": 723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924738476890492e-05,
+      "loss": 0.1324,
+      "step": 724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921211332187566e-05,
+      "loss": 0.1339,
+      "step": 725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917678741763425e-05,
+      "loss": 0.134,
+      "step": 726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914140707774448e-05,
+      "loss": 0.1318,
+      "step": 727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891059723238033e-05,
+      "loss": 0.1351,
+      "step": 728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8907048317744107e-05,
+      "loss": 0.1341,
+      "step": 729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890349396603211e-05,
+      "loss": 0.1359,
+      "step": 730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8899934179414004e-05,
+      "loss": 0.1355,
+      "step": 731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889636896006277e-05,
+      "loss": 0.127,
+      "step": 732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892798310154698e-05,
+      "loss": 0.1268,
+      "step": 733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889222231869407e-05,
+      "loss": 0.1302,
+      "step": 734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888564072738981e-05,
+      "loss": 0.1354,
+      "step": 735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888205379890215e-05,
+      "loss": 0.1295,
+      "step": 736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878461448595976e-05,
+      "loss": 0.1284,
+      "step": 737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874863678664143e-05,
+      "loss": 0.1311,
+      "step": 738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887126049130281e-05,
+      "loss": 0.1462,
+      "step": 739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8867651888711456e-05,
+      "loss": 0.1398,
+      "step": 740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864037873092857e-05,
+      "loss": 0.1189,
+      "step": 741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886041844665309e-05,
+      "loss": 0.1283,
+      "step": 742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885679361160155e-05,
+      "loss": 0.1289,
+      "step": 743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885316337015091e-05,
+      "loss": 0.1315,
+      "step": 744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849527724517167e-05,
+      "loss": 0.1496,
+      "step": 745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88458866769196e-05,
+      "loss": 0.1313,
+      "step": 746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842240229580796e-05,
+      "loss": 0.1376,
+      "step": 747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883858838472663e-05,
+      "loss": 0.127,
+      "step": 748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883493114458628e-05,
+      "loss": 0.133,
+      "step": 749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831268511392208e-05,
+      "loss": 0.1357,
+      "step": 750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882760048738018e-05,
+      "loss": 0.1282,
+      "step": 751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882392707478924e-05,
+      "loss": 0.1325,
+      "step": 752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820248275861725e-05,
+      "loss": 0.1279,
+      "step": 753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816564092843268e-05,
+      "loss": 0.1309,
+      "step": 754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881287452798278e-05,
+      "loss": 0.1336,
+      "step": 755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880917958353246e-05,
+      "loss": 0.1412,
+      "step": 756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880547926174779e-05,
+      "loss": 0.1277,
+      "step": 757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801773564887536e-05,
+      "loss": 0.1243,
+      "step": 758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798062495213745e-05,
+      "loss": 0.1225,
+      "step": 759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879434605499174e-05,
+      "loss": 0.1259,
+      "step": 760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790624246490127e-05,
+      "loss": 0.1399,
+      "step": 761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878689707198079e-05,
+      "loss": 0.1394,
+      "step": 762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878316453373888e-05,
+      "loss": 0.1313,
+      "step": 763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779426634042833e-05,
+      "loss": 0.1509,
+      "step": 764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775683375174346e-05,
+      "loss": 0.1317,
+      "step": 765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771934759418402e-05,
+      "loss": 0.1358,
+      "step": 766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876818078906324e-05,
+      "loss": 0.1407,
+      "step": 767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8764421466400376e-05,
+      "loss": 0.1373,
+      "step": 768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876065679372459e-05,
+      "loss": 0.1387,
+      "step": 769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756886773333927e-05,
+      "loss": 0.1318,
+      "step": 770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87531114075297e-05,
+      "loss": 0.1367,
+      "step": 771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749330698616484e-05,
+      "loss": 0.1278,
+      "step": 772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874554464890211e-05,
+      "loss": 0.1326,
+      "step": 773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874175326069768e-05,
+      "loss": 0.1412,
+      "step": 774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737956536317537e-05,
+      "loss": 0.1414,
+      "step": 775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734154478079306e-05,
+      "loss": 0.1236,
+      "step": 776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730347088303843e-05,
+      "loss": 0.1277,
+      "step": 777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872653436931528e-05,
+      "loss": 0.1239,
+      "step": 778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722716323440984e-05,
+      "loss": 0.1337,
+      "step": 779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718892953011588e-05,
+      "loss": 0.1349,
+      "step": 780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715064260360965e-05,
+      "loss": 0.136,
+      "step": 781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711230247826238e-05,
+      "loss": 0.1334,
+      "step": 782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707390917747787e-05,
+      "loss": 0.1342,
+      "step": 783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870354627246923e-05,
+      "loss": 0.1312,
+      "step": 784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699696314337422e-05,
+      "loss": 0.1271,
+      "step": 785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869584104570248e-05,
+      "loss": 0.1325,
+      "step": 786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869198046891775e-05,
+      "loss": 0.1255,
+      "step": 787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8688114586339818e-05,
+      "loss": 0.1345,
+      "step": 788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868424340032851e-05,
+      "loss": 0.1336,
+      "step": 789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8680366913246894e-05,
+      "loss": 0.1372,
+      "step": 790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867648512746127e-05,
+      "loss": 0.136,
+      "step": 791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672598045341175e-05,
+      "loss": 0.1339,
+      "step": 792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668705669259377e-05,
+      "loss": 0.1295,
+      "step": 793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664808001591866e-05,
+      "loss": 0.1354,
+      "step": 794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660905044717888e-05,
+      "loss": 0.1292,
+      "step": 795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865699680101989e-05,
+      "loss": 0.1343,
+      "step": 796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653083272883556e-05,
+      "loss": 0.1422,
+      "step": 797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649164462697805e-05,
+      "loss": 0.1313,
+      "step": 798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864524037285477e-05,
+      "loss": 0.1293,
+      "step": 799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864131100574981e-05,
+      "loss": 0.1381,
+      "step": 800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637376363781504e-05,
+      "loss": 0.1256,
+      "step": 801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633436449351648e-05,
+      "loss": 0.1326,
+      "step": 802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862949126486527e-05,
+      "loss": 0.1265,
+      "step": 803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625540812730592e-05,
+      "loss": 0.1394,
+      "step": 804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621585095359075e-05,
+      "loss": 0.1345,
+      "step": 805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861762411516538e-05,
+      "loss": 0.1322,
+      "step": 806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613657874567383e-05,
+      "loss": 0.1325,
+      "step": 807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609686375986176e-05,
+      "loss": 0.1332,
+      "step": 808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605709621846055e-05,
+      "loss": 0.1298,
+      "step": 809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601727614574522e-05,
+      "loss": 0.1325,
+      "step": 810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85977403566023e-05,
+      "loss": 0.1394,
+      "step": 811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593747850363298e-05,
+      "loss": 0.1273,
+      "step": 812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589750098294636e-05,
+      "loss": 0.1381,
+      "step": 813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585747102836643e-05,
+      "loss": 0.1417,
+      "step": 814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8581738866432847e-05,
+      "loss": 0.1433,
+      "step": 815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857772539152996e-05,
+      "loss": 0.1367,
+      "step": 816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573706680577913e-05,
+      "loss": 0.1384,
+      "step": 817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569682736029822e-05,
+      "loss": 0.1252,
+      "step": 818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565653560341994e-05,
+      "loss": 0.1358,
+      "step": 819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856161915597394e-05,
+      "loss": 0.1364,
+      "step": 820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557579525388353e-05,
+      "loss": 0.1348,
+      "step": 821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553534671051126e-05,
+      "loss": 0.1438,
+      "step": 822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549484595431326e-05,
+      "loss": 0.1351,
+      "step": 823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545429301001225e-05,
+      "loss": 0.1367,
+      "step": 824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854136879023627e-05,
+      "loss": 0.1357,
+      "step": 825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853730306561509e-05,
+      "loss": 0.1367,
+      "step": 826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533232129619502e-05,
+      "loss": 0.1366,
+      "step": 827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529155984734503e-05,
+      "loss": 0.1318,
+      "step": 828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525074633448276e-05,
+      "loss": 0.1292,
+      "step": 829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520988078252168e-05,
+      "loss": 0.1335,
+      "step": 830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516896321640712e-05,
+      "loss": 0.1292,
+      "step": 831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512799366111615e-05,
+      "loss": 0.1313,
+      "step": 832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508697214165758e-05,
+      "loss": 0.1284,
+      "step": 833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850458986830719e-05,
+      "loss": 0.134,
+      "step": 834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500477331043133e-05,
+      "loss": 0.1215,
+      "step": 835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496359604883984e-05,
+      "loss": 0.1389,
+      "step": 836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84922366923433e-05,
+      "loss": 0.1489,
+      "step": 837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488108595937798e-05,
+      "loss": 0.1442,
+      "step": 838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483975318187377e-05,
+      "loss": 0.1316,
+      "step": 839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479836861615084e-05,
+      "loss": 0.1292,
+      "step": 840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847569322874713e-05,
+      "loss": 0.1372,
+      "step": 841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471544422112894e-05,
+      "loss": 0.1364,
+      "step": 842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84673904442449e-05,
+      "loss": 0.141,
+      "step": 843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8463231297678838e-05,
+      "loss": 0.145,
+      "step": 844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459066984953555e-05,
+      "loss": 0.1245,
+      "step": 845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454897508611044e-05,
+      "loss": 0.1301,
+      "step": 846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450722871196452e-05,
+      "loss": 0.1407,
+      "step": 847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844654307525808e-05,
+      "loss": 0.1348,
+      "step": 848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442358123347374e-05,
+      "loss": 0.1373,
+      "step": 849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438168018018927e-05,
+      "loss": 0.1354,
+      "step": 850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433972761830486e-05,
+      "loss": 0.1204,
+      "step": 851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842977235734293e-05,
+      "loss": 0.1319,
+      "step": 852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842556680712029e-05,
+      "loss": 0.1341,
+      "step": 853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421356113729734e-05,
+      "loss": 0.1355,
+      "step": 854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841714027974157e-05,
+      "loss": 0.1359,
+      "step": 855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841291930772924e-05,
+      "loss": 0.136,
+      "step": 856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408693200269336e-05,
+      "loss": 0.1345,
+      "step": 857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404461959941565e-05,
+      "loss": 0.131,
+      "step": 858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8400225589328783e-05,
+      "loss": 0.1288,
+      "step": 859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839598409101697e-05,
+      "loss": 0.1294,
+      "step": 860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839173746759524e-05,
+      "loss": 0.1309,
+      "step": 861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387485721655837e-05,
+      "loss": 0.125,
+      "step": 862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383228855794124e-05,
+      "loss": 0.139,
+      "step": 863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8378966872608594e-05,
+      "loss": 0.1337,
+      "step": 864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374699774700863e-05,
+      "loss": 0.1482,
+      "step": 865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837042756467567e-05,
+      "loss": 0.1337,
+      "step": 866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366150245140876e-05,
+      "loss": 0.1318,
+      "step": 867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836186781870746e-05,
+      "loss": 0.1537,
+      "step": 868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835758028798952e-05,
+      "loss": 0.1302,
+      "step": 869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835328765560426e-05,
+      "loss": 0.1315,
+      "step": 870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834898992417201e-05,
+      "loss": 0.1352,
+      "step": 871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834468709631621e-05,
+      "loss": 0.1346,
+      "step": 872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340379174663405e-05,
+      "loss": 0.1345,
+      "step": 873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833606616184326e-05,
+      "loss": 0.1314,
+      "step": 874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833174806048853e-05,
+      "loss": 0.1338,
+      "step": 875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83274248732351e-05,
+      "loss": 0.1325,
+      "step": 876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323096602721943e-05,
+      "loss": 0.1262,
+      "step": 877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318763251591137e-05,
+      "loss": 0.1348,
+      "step": 878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314424822487865e-05,
+      "loss": 0.1189,
+      "step": 879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831008131806041e-05,
+      "loss": 0.127,
+      "step": 880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305732740960147e-05,
+      "loss": 0.1206,
+      "step": 881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301379093841554e-05,
+      "loss": 0.1412,
+      "step": 882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297020379362197e-05,
+      "loss": 0.1361,
+      "step": 883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8292656600182747e-05,
+      "loss": 0.132,
+      "step": 884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8288287758966956e-05,
+      "loss": 0.1285,
+      "step": 885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283913858381673e-05,
+      "loss": 0.1321,
+      "step": 886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279534901096824e-05,
+      "loss": 0.1406,
+      "step": 887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275150889785435e-05,
+      "loss": 0.1284,
+      "step": 888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270761827123612e-05,
+      "loss": 0.1333,
+      "step": 889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266367715790543e-05,
+      "loss": 0.1346,
+      "step": 890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261968558468495e-05,
+      "loss": 0.1308,
+      "step": 891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257564357842822e-05,
+      "loss": 0.1352,
+      "step": 892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825315511660196e-05,
+      "loss": 0.1407,
+      "step": 893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248740837437405e-05,
+      "loss": 0.1254,
+      "step": 894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244321523043744e-05,
+      "loss": 0.1266,
+      "step": 895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823989717611863e-05,
+      "loss": 0.13,
+      "step": 896
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235467799362795e-05,
+      "loss": 0.1465,
+      "step": 897
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231033395480034e-05,
+      "loss": 0.1261,
+      "step": 898
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226593967177212e-05,
+      "loss": 0.1398,
+      "step": 899
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222149517164264e-05,
+      "loss": 0.127,
+      "step": 900
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217700048154192e-05,
+      "loss": 0.1328,
+      "step": 901
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821324556286305e-05,
+      "loss": 0.1295,
+      "step": 902
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820878606400997e-05,
+      "loss": 0.145,
+      "step": 903
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204321554317136e-05,
+      "loss": 0.1314,
+      "step": 904
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819985203650979e-05,
+      "loss": 0.1269,
+      "step": 905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195377513316233e-05,
+      "loss": 0.1324,
+      "step": 906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190897987467822e-05,
+      "loss": 0.1364,
+      "step": 907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186413461698965e-05,
+      "loss": 0.1437,
+      "step": 908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181923938747123e-05,
+      "loss": 0.1296,
+      "step": 909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177429421352815e-05,
+      "loss": 0.137,
+      "step": 910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81729299122596e-05,
+      "loss": 0.1347,
+      "step": 911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168425414214075e-05,
+      "loss": 0.134,
+      "step": 912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816391592996591e-05,
+      "loss": 0.1312,
+      "step": 913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159401462267794e-05,
+      "loss": 0.1299,
+      "step": 914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8154882013875466e-05,
+      "loss": 0.1284,
+      "step": 915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150357587547708e-05,
+      "loss": 0.1329,
+      "step": 916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145828186046336e-05,
+      "loss": 0.1307,
+      "step": 917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141293812136206e-05,
+      "loss": 0.1256,
+      "step": 918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813675446858521e-05,
+      "loss": 0.1252,
+      "step": 919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813221015816427e-05,
+      "loss": 0.1326,
+      "step": 920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127660883647346e-05,
+      "loss": 0.1381,
+      "step": 921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123106647811417e-05,
+      "loss": 0.1363,
+      "step": 922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118547453436504e-05,
+      "loss": 0.1344,
+      "step": 923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113983303305646e-05,
+      "loss": 0.1315,
+      "step": 924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810941420020491e-05,
+      "loss": 0.1332,
+      "step": 925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104840146923387e-05,
+      "loss": 0.1375,
+      "step": 926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8100261146253185e-05,
+      "loss": 0.1244,
+      "step": 927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809567720098944e-05,
+      "loss": 0.1259,
+      "step": 928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8091088313930303e-05,
+      "loss": 0.1307,
+      "step": 929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086494487876934e-05,
+      "loss": 0.137,
+      "step": 930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808189572563352e-05,
+      "loss": 0.1342,
+      "step": 931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807729203000726e-05,
+      "loss": 0.1203,
+      "step": 932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807268340380835e-05,
+      "loss": 0.1423,
+      "step": 933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806806984985001e-05,
+      "loss": 0.1309,
+      "step": 934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806345137094846e-05,
+      "loss": 0.1342,
+      "step": 935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805882796992294e-05,
+      "loss": 0.1254,
+      "step": 936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805419964959568e-05,
+      "loss": 0.145,
+      "step": 937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804956641279192e-05,
+      "loss": 0.1295,
+      "step": 938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804492826233989e-05,
+      "loss": 0.1311,
+      "step": 939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040285201070836e-05,
+      "loss": 0.1274,
+      "step": 940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803563723181899e-05,
+      "loss": 0.133,
+      "step": 941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030984357421592e-05,
+      "loss": 0.1284,
+      "step": 942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8026326580718855e-05,
+      "loss": 0.1408,
+      "step": 943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802166390455401e-05,
+      "loss": 0.1393,
+      "step": 944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016996331773254e-05,
+      "loss": 0.1255,
+      "step": 945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80123238652258e-05,
+      "loss": 0.1249,
+      "step": 946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800764650776382e-05,
+      "loss": 0.1363,
+      "step": 947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800296426224249e-05,
+      "loss": 0.127,
+      "step": 948
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799827713151997e-05,
+      "loss": 0.1301,
+      "step": 949
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993585118457387e-05,
+      "loss": 0.1312,
+      "step": 950
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988888225918866e-05,
+      "loss": 0.1153,
+      "step": 951
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79841864567715e-05,
+      "loss": 0.1373,
+      "step": 952
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979479813885365e-05,
+      "loss": 0.1289,
+      "step": 953
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974768300133503e-05,
+      "loss": 0.1397,
+      "step": 954
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970051918391938e-05,
+      "loss": 0.1293,
+      "step": 955
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796533067153966e-05,
+      "loss": 0.1279,
+      "step": 956
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960604562458635e-05,
+      "loss": 0.1334,
+      "step": 957
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955873594033796e-05,
+      "loss": 0.1362,
+      "step": 958
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951137769153028e-05,
+      "loss": 0.1321,
+      "step": 959
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946397090707203e-05,
+      "loss": 0.1377,
+      "step": 960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.794165156159014e-05,
+      "loss": 0.1339,
+      "step": 961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936901184698625e-05,
+      "loss": 0.1209,
+      "step": 962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932145962932398e-05,
+      "loss": 0.1302,
+      "step": 963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927385899194164e-05,
+      "loss": 0.1319,
+      "step": 964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922620996389587e-05,
+      "loss": 0.1311,
+      "step": 965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791785125742726e-05,
+      "loss": 0.1331,
+      "step": 966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913076685218765e-05,
+      "loss": 0.1417,
+      "step": 967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7908297282678602e-05,
+      "loss": 0.1318,
+      "step": 968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790351305272424e-05,
+      "loss": 0.1242,
+      "step": 969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898723998276086e-05,
+      "loss": 0.1339,
+      "step": 970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789393012225749e-05,
+      "loss": 0.1286,
+      "step": 971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889131427594754e-05,
+      "loss": 0.1292,
+      "step": 972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7884327917217113e-05,
+      "loss": 0.129,
+      "step": 973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879519594056748e-05,
+      "loss": 0.1355,
+      "step": 974
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787470646104877e-05,
+      "loss": 0.1189,
+      "step": 975
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7869888521131234e-05,
+      "loss": 0.1311,
+      "step": 976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865065777245127e-05,
+      "loss": 0.1177,
+      "step": 977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860238232334364e-05,
+      "loss": 0.1293,
+      "step": 978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7855405889345796e-05,
+      "loss": 0.1359,
+      "step": 979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78505687512292e-05,
+      "loss": 0.1373,
+      "step": 980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845726820937287e-05,
+      "loss": 0.1366,
+      "step": 981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840880101425676e-05,
+      "loss": 0.1188,
+      "step": 982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783602859565293e-05,
+      "loss": 0.1238,
+      "step": 983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831172306580522e-05,
+      "loss": 0.1335,
+      "step": 984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826311237172843e-05,
+      "loss": 0.1389,
+      "step": 985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7821445390397213e-05,
+      "loss": 0.1438,
+      "step": 986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816574769223857e-05,
+      "loss": 0.1332,
+      "step": 987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781169937662592e-05,
+      "loss": 0.1275,
+      "step": 988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780681921557946e-05,
+      "loss": 0.1374,
+      "step": 989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801934289063444e-05,
+      "loss": 0.1281,
+      "step": 990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797044600059745e-05,
+      "loss": 0.1272,
+      "step": 991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792150151553153e-05,
+      "loss": 0.1344,
+      "step": 992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787250946531348e-05,
+      "loss": 0.1425,
+      "step": 993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782346987984925e-05,
+      "loss": 0.1411,
+      "step": 994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777438278907384e-05,
+      "loss": 0.1482,
+      "step": 995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772524822295108e-05,
+      "loss": 0.1182,
+      "step": 996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77676066211474e-05,
+      "loss": 0.1359,
+      "step": 997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776268367846644e-05,
+      "loss": 0.1454,
+      "step": 998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757755997257315e-05,
+      "loss": 0.1423,
+      "step": 999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752823580527996e-05,
+      "loss": 0.1272,
+      "step": 1000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.12931811809539795,
+      "eval_runtime": 712.3744,
+      "eval_samples_per_second": 70.188,
+      "eval_steps_per_second": 2.194,
+      "step": 1000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774788643128935e-05,
+      "loss": 0.1271,
+      "step": 1001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774294455255513e-05,
+      "loss": 0.1303,
+      "step": 1002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737997947341977e-05,
+      "loss": 0.1332,
+      "step": 1003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733046618669423e-05,
+      "loss": 0.123,
+      "step": 1004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728090569559868e-05,
+      "loss": 0.1265,
+      "step": 1005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772312980303861e-05,
+      "loss": 0.1275,
+      "step": 1006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771816432213382e-05,
+      "loss": 0.1375,
+      "step": 1007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7713194129876546e-05,
+      "loss": 0.1296,
+      "step": 1008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770821922930071e-05,
+      "loss": 0.1225,
+      "step": 1009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770323962344311e-05,
+      "loss": 0.1303,
+      "step": 1010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769825531534342e-05,
+      "loss": 0.1331,
+      "step": 1011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769326630804418e-05,
+      "loss": 0.1299,
+      "step": 1012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76882726045908e-05,
+      "loss": 0.1244,
+      "step": 1013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683274208031552e-05,
+      "loss": 0.1283,
+      "step": 1014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678271121417586e-05,
+      "loss": 0.1321,
+      "step": 1015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7673263347802897e-05,
+      "loss": 0.1266,
+      "step": 1016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668250890244355e-05,
+      "loss": 0.1303,
+      "step": 1017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663233751801684e-05,
+      "loss": 0.127,
+      "step": 1018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765821193553746e-05,
+      "loss": 0.1312,
+      "step": 1019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765318544451713e-05,
+      "loss": 0.1261,
+      "step": 1020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648154281808975e-05,
+      "loss": 0.1303,
+      "step": 1021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643118450484142e-05,
+      "loss": 0.127,
+      "step": 1022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638077953616624e-05,
+      "loss": 0.1272,
+      "step": 1023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763303279428326e-05,
+      "loss": 0.1195,
+      "step": 1024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627982975563735e-05,
+      "loss": 0.1343,
+      "step": 1025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622928500540578e-05,
+      "loss": 0.1299,
+      "step": 1026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7617869372299168e-05,
+      "loss": 0.1288,
+      "step": 1027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761280559392771e-05,
+      "loss": 0.1506,
+      "step": 1028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760773716851726e-05,
+      "loss": 0.1328,
+      "step": 1029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760266409916171e-05,
+      "loss": 0.1403,
+      "step": 1030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7597586388957777e-05,
+      "loss": 0.131,
+      "step": 1031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759250404100502e-05,
+      "loss": 0.1238,
+      "step": 1032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587417058405822e-05,
+      "loss": 0.1306,
+      "step": 1033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582325444265404e-05,
+      "loss": 0.1365,
+      "step": 1034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577229201691807e-05,
+      "loss": 0.1273,
+      "step": 1035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572128333795902e-05,
+      "loss": 0.1276,
+      "step": 1036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756702284369138e-05,
+      "loss": 0.1389,
+      "step": 1037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756191273449475e-05,
+      "loss": 0.1406,
+      "step": 1038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7556798009325347e-05,
+      "loss": 0.1435,
+      "step": 1039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755167867130533e-05,
+      "loss": 0.1307,
+      "step": 1040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546554723559653e-05,
+      "loss": 0.1294,
+      "step": 1041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541426169216108e-05,
+      "loss": 0.122,
+      "step": 1042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753629301140528e-05,
+      "loss": 0.1318,
+      "step": 1043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531155253260574e-05,
+      "loss": 0.1245,
+      "step": 1044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.75260128979182e-05,
+      "loss": 0.1375,
+      "step": 1045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520865948517175e-05,
+      "loss": 0.1289,
+      "step": 1046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515714408199323e-05,
+      "loss": 0.1381,
+      "step": 1047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510558280109266e-05,
+      "loss": 0.1262,
+      "step": 1048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505397567394424e-05,
+      "loss": 0.1295,
+      "step": 1049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500232273205025e-05,
+      "loss": 0.1323,
+      "step": 1050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495062400694086e-05,
+      "loss": 0.1352,
+      "step": 1051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489887953017428e-05,
+      "loss": 0.1323,
+      "step": 1052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484708933333642e-05,
+      "loss": 0.1302,
+      "step": 1053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747952534480414e-05,
+      "loss": 0.1295,
+      "step": 1054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474337190593104e-05,
+      "loss": 0.1338,
+      "step": 1055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746914447386751e-05,
+      "loss": 0.1266,
+      "step": 1056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746394719779711e-05,
+      "loss": 0.1284,
+      "step": 1057
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7458745365554452e-05,
+      "loss": 0.1246,
+      "step": 1058
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453538980314858e-05,
+      "loss": 0.1255,
+      "step": 1059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744832804525643e-05,
+      "loss": 0.1291,
+      "step": 1060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443112563560048e-05,
+      "loss": 0.129,
+      "step": 1061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7437892538409364e-05,
+      "loss": 0.1356,
+      "step": 1062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432667972990812e-05,
+      "loss": 0.1185,
+      "step": 1063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427438870493584e-05,
+      "loss": 0.1323,
+      "step": 1064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742220523410966e-05,
+      "loss": 0.1259,
+      "step": 1065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741696706703377e-05,
+      "loss": 0.1286,
+      "step": 1066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411724372463416e-05,
+      "loss": 0.1347,
+      "step": 1067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406477153598865e-05,
+      "loss": 0.1358,
+      "step": 1068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401225413643153e-05,
+      "loss": 0.1311,
+      "step": 1069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395969155802057e-05,
+      "loss": 0.1326,
+      "step": 1070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739070838328413e-05,
+      "loss": 0.1365,
+      "step": 1071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385443099300673e-05,
+      "loss": 0.1236,
+      "step": 1072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380173307065736e-05,
+      "loss": 0.1306,
+      "step": 1073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737489900979613e-05,
+      "loss": 0.1226,
+      "step": 1074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.736962021071141e-05,
+      "loss": 0.1321,
+      "step": 1075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364336913033888e-05,
+      "loss": 0.136,
+      "step": 1076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7359049119988612e-05,
+      "loss": 0.132,
+      "step": 1077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353756834803377e-05,
+      "loss": 0.1271,
+      "step": 1078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348460060708717e-05,
+      "loss": 0.1386,
+      "step": 1079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343158800937918e-05,
+      "loss": 0.1391,
+      "step": 1080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733785305872699e-05,
+      "loss": 0.1344,
+      "step": 1081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332542837314686e-05,
+      "loss": 0.1266,
+      "step": 1082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327228139942497e-05,
+      "loss": 0.1414,
+      "step": 1083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321908969854635e-05,
+      "loss": 0.1255,
+      "step": 1084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316585330298053e-05,
+      "loss": 0.1256,
+      "step": 1085
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311257224522427e-05,
+      "loss": 0.1325,
+      "step": 1086
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730592465578016e-05,
+      "loss": 0.1204,
+      "step": 1087
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300587627326383e-05,
+      "loss": 0.1187,
+      "step": 1088
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295246142418943e-05,
+      "loss": 0.1407,
+      "step": 1089
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728990020431841e-05,
+      "loss": 0.122,
+      "step": 1090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728454981628807e-05,
+      "loss": 0.1314,
+      "step": 1091
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279194981593935e-05,
+      "loss": 0.1313,
+      "step": 1092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727383570350472e-05,
+      "loss": 0.1387,
+      "step": 1093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268471985291858e-05,
+      "loss": 0.1249,
+      "step": 1094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263103830229493e-05,
+      "loss": 0.1374,
+      "step": 1095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725773124159447e-05,
+      "loss": 0.1198,
+      "step": 1096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7252354222666356e-05,
+      "loss": 0.1501,
+      "step": 1097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72469727767274e-05,
+      "loss": 0.1375,
+      "step": 1098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724158690706257e-05,
+      "loss": 0.1244,
+      "step": 1099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723619661695954e-05,
+      "loss": 0.1347,
+      "step": 1100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230801909708653e-05,
+      "loss": 0.1248,
+      "step": 1101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7225402788602986e-05,
+      "loss": 0.126,
+      "step": 1102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721999925693828e-05,
+      "loss": 0.1322,
+      "step": 1103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214591318012985e-05,
+      "loss": 0.1245,
+      "step": 1104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209178975128237e-05,
+      "loss": 0.1334,
+      "step": 1105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203762231587856e-05,
+      "loss": 0.1269,
+      "step": 1106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198341090698357e-05,
+      "loss": 0.1262,
+      "step": 1107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719291555576893e-05,
+      "loss": 0.1276,
+      "step": 1108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187485630111455e-05,
+      "loss": 0.1297,
+      "step": 1109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182051317040487e-05,
+      "loss": 0.1292,
+      "step": 1110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717661261987326e-05,
+      "loss": 0.134,
+      "step": 1111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171169541929685e-05,
+      "loss": 0.1203,
+      "step": 1112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165722086532355e-05,
+      "loss": 0.1317,
+      "step": 1113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160270257006514e-05,
+      "loss": 0.1208,
+      "step": 1114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71548140566801e-05,
+      "loss": 0.1257,
+      "step": 1115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149353488883704e-05,
+      "loss": 0.1404,
+      "step": 1116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714388855695059e-05,
+      "loss": 0.135,
+      "step": 1117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713841926421668e-05,
+      "loss": 0.1166,
+      "step": 1118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713294561402056e-05,
+      "loss": 0.1299,
+      "step": 1119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712746760970348e-05,
+      "loss": 0.1292,
+      "step": 1120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712198525460935e-05,
+      "loss": 0.1194,
+      "step": 1121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116498552084727e-05,
+      "loss": 0.1394,
+      "step": 1122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7111007505478822e-05,
+      "loss": 0.1354,
+      "step": 1123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710551211814351e-05,
+      "loss": 0.1228,
+      "step": 1124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100012393433294e-05,
+      "loss": 0.1222,
+      "step": 1125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709450833470535e-05,
+      "loss": 0.1406,
+      "step": 1126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708899994531948e-05,
+      "loss": 0.124,
+      "step": 1127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708348722863814e-05,
+      "loss": 0.1124,
+      "step": 1128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707797018802643e-05,
+      "loss": 0.1249,
+      "step": 1129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707244882685207e-05,
+      "loss": 0.1334,
+      "step": 1130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066923148485445e-05,
+      "loss": 0.1263,
+      "step": 1131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706139315629955e-05,
+      "loss": 0.1322,
+      "step": 1132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055858853670035e-05,
+      "loss": 0.1175,
+      "step": 1133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050320243975165e-05,
+      "loss": 0.1278,
+      "step": 1134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704477733059584e-05,
+      "loss": 0.1329,
+      "step": 1135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703923011691559e-05,
+      "loss": 0.134,
+      "step": 1136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703367860632057e-05,
+      "loss": 0.1307,
+      "step": 1137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028122802199545e-05,
+      "loss": 0.1327,
+      "step": 1138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022562707943918e-05,
+      "loss": 0.1315,
+      "step": 1139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016998326947705e-05,
+      "loss": 0.1349,
+      "step": 1140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011429662607532e-05,
+      "loss": 0.1257,
+      "step": 1141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7005856718322654e-05,
+      "loss": 0.1307,
+      "step": 1142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000279497494918e-05,
+      "loss": 0.1255,
+      "step": 1143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994698003528803e-05,
+      "loss": 0.1281,
+      "step": 1144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698911223983138e-05,
+      "loss": 0.1307,
+      "step": 1145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698352220981234e-05,
+      "loss": 0.1312,
+      "step": 1146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977927916883966e-05,
+      "loss": 0.1272,
+      "step": 1147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697232936446115e-05,
+      "loss": 0.1222,
+      "step": 1148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696672655596138e-05,
+      "loss": 0.1338,
+      "step": 1149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696111949480475e-05,
+      "loss": 0.1324,
+      "step": 1150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955508184413942e-05,
+      "loss": 0.1286,
+      "step": 1151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949892628214233e-05,
+      "loss": 0.1308,
+      "step": 1152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944272829633494e-05,
+      "loss": 0.1217,
+      "step": 1153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938648792102185e-05,
+      "loss": 0.1378,
+      "step": 1154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6933020519053354e-05,
+      "loss": 0.1265,
+      "step": 1155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6927388013922634e-05,
+      "loss": 0.1375,
+      "step": 1156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921751280148236e-05,
+      "loss": 0.1247,
+      "step": 1157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6916110321170963e-05,
+      "loss": 0.1339,
+      "step": 1158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691046514043419e-05,
+      "loss": 0.1274,
+      "step": 1159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690481574138387e-05,
+      "loss": 0.1215,
+      "step": 1160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899162127468537e-05,
+      "loss": 0.1313,
+      "step": 1161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893504302139283e-05,
+      "loss": 0.1289,
+      "step": 1162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887842268849785e-05,
+      "loss": 0.1317,
+      "step": 1163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882176031056287e-05,
+      "loss": 0.1274,
+      "step": 1164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687650559221759e-05,
+      "loss": 0.14,
+      "step": 1165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687083095579507e-05,
+      "loss": 0.125,
+      "step": 1166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865152125252657e-05,
+      "loss": 0.1313,
+      "step": 1167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.685946910405685e-05,
+      "loss": 0.1318,
+      "step": 1168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6853781895676703e-05,
+      "loss": 0.1336,
+      "step": 1169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848090503583816e-05,
+      "loss": 0.135,
+      "step": 1170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842394931252356e-05,
+      "loss": 0.1367,
+      "step": 1171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683669518215904e-05,
+      "loss": 0.1236,
+      "step": 1172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683099125978313e-05,
+      "loss": 0.1408,
+      "step": 1173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825283167606427e-05,
+      "loss": 0.1234,
+      "step": 1174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68195709091133e-05,
+      "loss": 0.1405,
+      "step": 1175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813854487790644e-05,
+      "loss": 0.1308,
+      "step": 1176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808133907127898e-05,
+      "loss": 0.1348,
+      "step": 1177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802409170617045e-05,
+      "loss": 0.1339,
+      "step": 1178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67966802817526e-05,
+      "loss": 0.1356,
+      "step": 1179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790947244031612e-05,
+      "loss": 0.1302,
+      "step": 1180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785210060953664e-05,
+      "loss": 0.1367,
+      "step": 1181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6779468736020877e-05,
+      "loss": 0.1261,
+      "step": 1182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773723272737885e-05,
+      "loss": 0.1453,
+      "step": 1183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676797367461186e-05,
+      "loss": 0.1255,
+      "step": 1184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762219945152497e-05,
+      "loss": 0.1246,
+      "step": 1185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756462087872005e-05,
+      "loss": 0.127,
+      "step": 1186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750700106285124e-05,
+      "loss": 0.136,
+      "step": 1187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67449340039091e-05,
+      "loss": 0.1281,
+      "step": 1188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739163784263702e-05,
+      "loss": 0.1278,
+      "step": 1189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6733389450871214e-05,
+      "loss": 0.1227,
+      "step": 1190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727611007256422e-05,
+      "loss": 0.1328,
+      "step": 1191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672182845694663e-05,
+      "loss": 0.1363,
+      "step": 1192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6716041803471645e-05,
+      "loss": 0.1271,
+      "step": 1193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6710251050363777e-05,
+      "loss": 0.1238,
+      "step": 1194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6704456201157842e-05,
+      "loss": 0.1235,
+      "step": 1195
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698657259391157e-05,
+      "loss": 0.125,
+      "step": 1196
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669285422860353e-05,
+      "loss": 0.1329,
+      "step": 1197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687047112337274e-05,
+      "loss": 0.1191,
+      "step": 1198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668123591413719e-05,
+      "loss": 0.1404,
+      "step": 1199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675420637550574e-05,
+      "loss": 0.133,
+      "step": 1200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669601286127208e-05,
+      "loss": 0.1304,
+      "step": 1201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6663777863419367e-05,
+      "loss": 0.1288,
+      "step": 1202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6657950372981802e-05,
+      "loss": 0.1341,
+      "step": 1203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665211881837176e-05,
+      "loss": 0.1337,
+      "step": 1204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646283203148957e-05,
+      "loss": 0.1378,
+      "step": 1205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640443530875598e-05,
+      "loss": 0.1264,
+      "step": 1206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634599805116352e-05,
+      "loss": 0.1342,
+      "step": 1207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662875202943837e-05,
+      "loss": 0.1304,
+      "step": 1208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662290020741128e-05,
+      "loss": 0.1292,
+      "step": 1209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617044342607166e-05,
+      "loss": 0.1375,
+      "step": 1210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66111844386006e-05,
+      "loss": 0.1301,
+      "step": 1211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6605320498968596e-05,
+      "loss": 0.1214,
+      "step": 1212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659945252729065e-05,
+      "loss": 0.1287,
+      "step": 1213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659358052714871e-05,
+      "loss": 0.1292,
+      "step": 1214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587704502127187e-05,
+      "loss": 0.1376,
+      "step": 1215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6581824455812945e-05,
+      "loss": 0.133,
+      "step": 1216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575940391795308e-05,
+      "loss": 0.1223,
+      "step": 1217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657005231366605e-05,
+      "loss": 0.1313,
+      "step": 1218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6564160225019392e-05,
+      "loss": 0.1262,
+      "step": 1219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655826412945201e-05,
+      "loss": 0.1291,
+      "step": 1220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655236403056302e-05,
+      "loss": 0.1359,
+      "step": 1221
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546459931953983e-05,
+      "loss": 0.1249,
+      "step": 1222
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540551837228905e-05,
+      "loss": 0.1297,
+      "step": 1223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534639749994223e-05,
+      "loss": 0.1216,
+      "step": 1224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528723673858822e-05,
+      "loss": 0.1364,
+      "step": 1225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6522803612434015e-05,
+      "loss": 0.1372,
+      "step": 1226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651687956933355e-05,
+      "loss": 0.1376,
+      "step": 1227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510951548173605e-05,
+      "loss": 0.1327,
+      "step": 1228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505019552572786e-05,
+      "loss": 0.1268,
+      "step": 1229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499083586152122e-05,
+      "loss": 0.1242,
+      "step": 1230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493143652535074e-05,
+      "loss": 0.1273,
+      "step": 1231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6487199755347517e-05,
+      "loss": 0.1262,
+      "step": 1232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6481251898217746e-05,
+      "loss": 0.1391,
+      "step": 1233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475300084776486e-05,
+      "loss": 0.1272,
+      "step": 1234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469344318656856e-05,
+      "loss": 0.1318,
+      "step": 1235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463384603494402e-05,
+      "loss": 0.1317,
+      "step": 1236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457420942927073e-05,
+      "loss": 0.1267,
+      "step": 1237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645145334059524e-05,
+      "loss": 0.126,
+      "step": 1238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644548180014166e-05,
+      "loss": 0.134,
+      "step": 1239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643950632521151e-05,
+      "loss": 0.1236,
+      "step": 1240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433526919452362e-05,
+      "loss": 0.1158,
+      "step": 1241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642754358651419e-05,
+      "loss": 0.125,
+      "step": 1242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642155633004936e-05,
+      "loss": 0.1216,
+      "step": 1243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641556515371264e-05,
+      "loss": 0.1263,
+      "step": 1244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640957006116119e-05,
+      "loss": 0.1285,
+      "step": 1245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403571056054554e-05,
+      "loss": 0.1365,
+      "step": 1246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6397568142054665e-05,
+      "loss": 0.1262,
+      "step": 1247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391561322825853e-05,
+      "loss": 0.1325,
+      "step": 1248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638555060203482e-05,
+      "loss": 0.1346,
+      "step": 1249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637953598335066e-05,
+      "loss": 0.1313,
+      "step": 1250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637351747044483e-05,
+      "loss": 0.1347,
+      "step": 1251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367495066991187e-05,
+      "loss": 0.1439,
+      "step": 1252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361468776665944e-05,
+      "loss": 0.1314,
+      "step": 1253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355438603147686e-05,
+      "loss": 0.1216,
+      "step": 1254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634940455011739e-05,
+      "loss": 0.123,
+      "step": 1255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343366621258376e-05,
+      "loss": 0.1343,
+      "step": 1256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633732482025634e-05,
+      "loss": 0.1136,
+      "step": 1257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633127915079935e-05,
+      "loss": 0.1293,
+      "step": 1258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632522961657782e-05,
+      "loss": 0.1226,
+      "step": 1259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6319176221284537e-05,
+      "loss": 0.1253,
+      "step": 1260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631311896861463e-05,
+      "loss": 0.1248,
+      "step": 1261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307057862265602e-05,
+      "loss": 0.1299,
+      "step": 1262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300992905937292e-05,
+      "loss": 0.1215,
+      "step": 1263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629492410333189e-05,
+      "loss": 0.1432,
+      "step": 1264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628885145815395e-05,
+      "loss": 0.1302,
+      "step": 1265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282774974110347e-05,
+      "loss": 0.1177,
+      "step": 1266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276694654910326e-05,
+      "loss": 0.1199,
+      "step": 1267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6270610504265453e-05,
+      "loss": 0.1315,
+      "step": 1268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626452252588963e-05,
+      "loss": 0.1208,
+      "step": 1269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258430723499123e-05,
+      "loss": 0.1295,
+      "step": 1270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6252335100812502e-05,
+      "loss": 0.1338,
+      "step": 1271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246235661550685e-05,
+      "loss": 0.1258,
+      "step": 1272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6240132409436914e-05,
+      "loss": 0.127,
+      "step": 1273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623402534819676e-05,
+      "loss": 0.1349,
+      "step": 1274
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227914481558125e-05,
+      "loss": 0.118,
+      "step": 1275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6221799813251222e-05,
+      "loss": 0.1183,
+      "step": 1276
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215681347008595e-05,
+      "loss": 0.1261,
+      "step": 1277
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62095590865651e-05,
+      "loss": 0.1333,
+      "step": 1278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203433035657914e-05,
+      "loss": 0.1294,
+      "step": 1279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619730319802652e-05,
+      "loss": 0.1188,
+      "step": 1280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6191169577412728e-05,
+      "loss": 0.1342,
+      "step": 1281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618503217756064e-05,
+      "loss": 0.1231,
+      "step": 1282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617889100221667e-05,
+      "loss": 0.1234,
+      "step": 1283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172746055129544e-05,
+      "loss": 0.1393,
+      "step": 1284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166597340050278e-05,
+      "loss": 0.1272,
+      "step": 1285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160444860732203e-05,
+      "loss": 0.1215,
+      "step": 1286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154288620930935e-05,
+      "loss": 0.1297,
+      "step": 1287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6148128624404393e-05,
+      "loss": 0.1282,
+      "step": 1288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6141964874912783e-05,
+      "loss": 0.1307,
+      "step": 1289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613579737621861e-05,
+      "loss": 0.1239,
+      "step": 1290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129626132086657e-05,
+      "loss": 0.1248,
+      "step": 1291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6123451146284005e-05,
+      "loss": 0.1286,
+      "step": 1292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611727242258001e-05,
+      "loss": 0.1315,
+      "step": 1293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6111089964746318e-05,
+      "loss": 0.1286,
+      "step": 1294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610490377655684e-05,
+      "loss": 0.1297,
+      "step": 1295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609871386178778e-05,
+      "loss": 0.1226,
+      "step": 1296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609252022421761e-05,
+      "loss": 0.1339,
+      "step": 1297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608632286762708e-05,
+      "loss": 0.1239,
+      "step": 1298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6080121795799197e-05,
+      "loss": 0.1235,
+      "step": 1299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6073917012519244e-05,
+      "loss": 0.1319,
+      "step": 1300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606770852157477e-05,
+      "loss": 0.1149,
+      "step": 1301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606149632675559e-05,
+      "loss": 0.1281,
+      "step": 1302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6055280431853775e-05,
+      "loss": 0.1334,
+      "step": 1303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604906084066366e-05,
+      "loss": 0.1261,
+      "step": 1304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604283755698182e-05,
+      "loss": 0.1265,
+      "step": 1305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036610584607113e-05,
+      "loss": 0.1394,
+      "step": 1306
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603037992734062e-05,
+      "loss": 0.1239,
+      "step": 1307
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602414558898569e-05,
+      "loss": 0.1322,
+      "step": 1308
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017907573347912e-05,
+      "loss": 0.1274,
+      "step": 1309
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6011665884235114e-05,
+      "loss": 0.1305,
+      "step": 1310
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600542052545738e-05,
+      "loss": 0.126,
+      "step": 1311
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999171500827026e-05,
+      "loss": 0.1238,
+      "step": 1312
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59929188141586e-05,
+      "loss": 0.1249,
+      "step": 1313
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5986662469268897e-05,
+      "loss": 0.1244,
+      "step": 1314
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5980402469976945e-05,
+      "loss": 0.1234,
+      "step": 1315
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5974138820103985e-05,
+      "loss": 0.1198,
+      "step": 1316
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967871523473508e-05,
+      "loss": 0.128,
+      "step": 1317
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961600583911222e-05,
+      "loss": 0.1292,
+      "step": 1318
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955326005245052e-05,
+      "loss": 0.1331,
+      "step": 1319
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594904779130516e-05,
+      "loss": 0.1257,
+      "step": 1320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942765945923912e-05,
+      "loss": 0.1238,
+      "step": 1321
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59364804729359e-05,
+      "loss": 0.1181,
+      "step": 1322
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5930191376177924e-05,
+      "loss": 0.1275,
+      "step": 1323
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923898659489004e-05,
+      "loss": 0.1341,
+      "step": 1324
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917602326710363e-05,
+      "loss": 0.1207,
+      "step": 1325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911302381685433e-05,
+      "loss": 0.1397,
+      "step": 1326
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590499882825985e-05,
+      "loss": 0.1284,
+      "step": 1327
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589869167028146e-05,
+      "loss": 0.134,
+      "step": 1328
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892380911600293e-05,
+      "loss": 0.1249,
+      "step": 1329
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5886066556068596e-05,
+      "loss": 0.1354,
+      "step": 1330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5879748607540803e-05,
+      "loss": 0.1254,
+      "step": 1331
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873427069873537e-05,
+      "loss": 0.1318,
+      "step": 1332
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5867101946925618e-05,
+      "loss": 0.1233,
+      "step": 1333
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586077324255805e-05,
+      "loss": 0.1394,
+      "step": 1334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5854440960634033e-05,
+      "loss": 0.1288,
+      "step": 1335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848105105018935e-05,
+      "loss": 0.1202,
+      "step": 1336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584176567958032e-05,
+      "loss": 0.1295,
+      "step": 1337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583542268818792e-05,
+      "loss": 0.1333,
+      "step": 1338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829076134713656e-05,
+      "loss": 0.1253,
+      "step": 1339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822726023031613e-05,
+      "loss": 0.1273,
+      "step": 1340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581637235701805e-05,
+      "loss": 0.1336,
+      "step": 1341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810015140551396e-05,
+      "loss": 0.1283,
+      "step": 1342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580365437751225e-05,
+      "loss": 0.1329,
+      "step": 1343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579729007178338e-05,
+      "loss": 0.1194,
+      "step": 1344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790922227249702e-05,
+      "loss": 0.1235,
+      "step": 1345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784550847798304e-05,
+      "loss": 0.1252,
+      "step": 1346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577817593731843e-05,
+      "loss": 0.129,
+      "step": 1347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5771797499701472e-05,
+      "loss": 0.1336,
+      "step": 1348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5765415538840993e-05,
+      "loss": 0.1238,
+      "step": 1349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5759030058632683e-05,
+      "loss": 0.1467,
+      "step": 1350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57526410629744e-05,
+      "loss": 0.1308,
+      "step": 1351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746248555766136e-05,
+      "loss": 0.1335,
+      "step": 1352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573985254091003e-05,
+      "loss": 0.1353,
+      "step": 1353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573345302231037e-05,
+      "loss": 0.1359,
+      "step": 1354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727050003873567e-05,
+      "loss": 0.1335,
+      "step": 1355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5720643489508178e-05,
+      "loss": 0.1445,
+      "step": 1356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57142334831249e-05,
+      "loss": 0.1409,
+      "step": 1357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570781998863655e-05,
+      "loss": 0.1312,
+      "step": 1358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701403009958075e-05,
+      "loss": 0.1345,
+      "step": 1359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5694982551006556e-05,
+      "loss": 0.1263,
+      "step": 1360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688558615701195e-05,
+      "loss": 0.1313,
+      "step": 1361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5682131207963317e-05,
+      "loss": 0.1373,
+      "step": 1362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567570033171636e-05,
+      "loss": 0.1315,
+      "step": 1363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669265990885895e-05,
+      "loss": 0.1302,
+      "step": 1364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566282818939959e-05,
+      "loss": 0.1291,
+      "step": 1365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565638693118724e-05,
+      "loss": 0.1343,
+      "step": 1366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5649942220180735e-05,
+      "loss": 0.1288,
+      "step": 1367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564349406031409e-05,
+      "loss": 0.1286,
+      "step": 1368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5637042455523407e-05,
+      "loss": 0.123,
+      "step": 1369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5630587409746906e-05,
+      "loss": 0.116,
+      "step": 1370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624128926924903e-05,
+      "loss": 0.1373,
+      "step": 1371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617667010999805e-05,
+      "loss": 0.1205,
+      "step": 1372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611201665916124e-05,
+      "loss": 0.1236,
+      "step": 1373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604732895620465e-05,
+      "loss": 0.1265,
+      "step": 1374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5598260704061514e-05,
+      "loss": 0.142,
+      "step": 1375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591785095190055e-05,
+      "loss": 0.1306,
+      "step": 1376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585306072958954e-05,
+      "loss": 0.1341,
+      "step": 1377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5578823641323163e-05,
+      "loss": 0.1344,
+      "step": 1378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557233780423971e-05,
+      "loss": 0.1216,
+      "step": 1379
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565848565667705e-05,
+      "loss": 0.126,
+      "step": 1380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555935592956834e-05,
+      "loss": 0.1357,
+      "step": 1381
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555285989990488e-05,
+      "loss": 0.1308,
+      "step": 1382
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554636048064264e-05,
+      "loss": 0.1384,
+      "step": 1383
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553985767574904e-05,
+      "loss": 0.1294,
+      "step": 1384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533351489193537e-05,
+      "loss": 0.1218,
+      "step": 1385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5526841924947663e-05,
+      "loss": 0.1216,
+      "step": 1386
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552032898698502e-05,
+      "loss": 0.1262,
+      "step": 1387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5513812679281253e-05,
+      "loss": 0.1217,
+      "step": 1388
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507293005814075e-05,
+      "loss": 0.1222,
+      "step": 1389
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5500769970563257e-05,
+      "loss": 0.1431,
+      "step": 1390
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5494243577510607e-05,
+      "loss": 0.1325,
+      "step": 1391
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487713830639996e-05,
+      "loss": 0.1238,
+      "step": 1392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481180733937334e-05,
+      "loss": 0.1306,
+      "step": 1393
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547464429139059e-05,
+      "loss": 0.1279,
+      "step": 1394
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468104506989754e-05,
+      "loss": 0.1236,
+      "step": 1395
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5461561384726872e-05,
+      "loss": 0.1203,
+      "step": 1396
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5455014928596025e-05,
+      "loss": 0.1389,
+      "step": 1397
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448465142593318e-05,
+      "loss": 0.1371,
+      "step": 1398
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544191203071691e-05,
+      "loss": 0.1238,
+      "step": 1399
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543535559696697e-05,
+      "loss": 0.1397,
+      "step": 1400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54287958453457e-05,
+      "loss": 0.1339,
+      "step": 1401
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422232779857333e-05,
+      "loss": 0.1276,
+      "step": 1402
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415666404508125e-05,
+      "loss": 0.1238,
+      "step": 1403
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540909672330634e-05,
+      "loss": 0.1302,
+      "step": 1404
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540252374026228e-05,
+      "loss": 0.1246,
+      "step": 1405
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395947459388242e-05,
+      "loss": 0.1333,
+      "step": 1406
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538936788469855e-05,
+      "loss": 0.137,
+      "step": 1407
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538278502020953e-05,
+      "loss": 0.1351,
+      "step": 1408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5376198869939525e-05,
+      "loss": 0.1238,
+      "step": 1409
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536960943790888e-05,
+      "loss": 0.1253,
+      "step": 1410
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363016728139936e-05,
+      "loss": 0.1221,
+      "step": 1411
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5356420744657044e-05,
+      "loss": 0.1309,
+      "step": 1412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349821491486555e-05,
+      "loss": 0.1321,
+      "step": 1413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5343218972656805e-05,
+      "loss": 0.1246,
+      "step": 1414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336613192198132e-05,
+      "loss": 0.1302,
+      "step": 1415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533000415414286e-05,
+      "loss": 0.1323,
+      "step": 1416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532339186252531e-05,
+      "loss": 0.1162,
+      "step": 1417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531677632138178e-05,
+      "loss": 0.1365,
+      "step": 1418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5310157534750555e-05,
+      "loss": 0.1361,
+      "step": 1419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303535506671903e-05,
+      "loss": 0.1243,
+      "step": 1420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296910241188064e-05,
+      "loss": 0.1315,
+      "step": 1421
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290281742343263e-05,
+      "loss": 0.1204,
+      "step": 1422
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528365001418369e-05,
+      "loss": 0.1212,
+      "step": 1423
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277015060757515e-05,
+      "loss": 0.1242,
+      "step": 1424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5270376886114874e-05,
+      "loss": 0.1428,
+      "step": 1425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5263735494307862e-05,
+      "loss": 0.1441,
+      "step": 1426
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257090889390554e-05,
+      "loss": 0.1251,
+      "step": 1427
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250443075418963e-05,
+      "loss": 0.129,
+      "step": 1428
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243792056451081e-05,
+      "loss": 0.1418,
+      "step": 1429
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237137836546852e-05,
+      "loss": 0.1204,
+      "step": 1430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230480419768163e-05,
+      "loss": 0.1257,
+      "step": 1431
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223819810178867e-05,
+      "loss": 0.1323,
+      "step": 1432
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521715601184476e-05,
+      "loss": 0.1326,
+      "step": 1433
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210489028833584e-05,
+      "loss": 0.1279,
+      "step": 1434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.520381886521502e-05,
+      "loss": 0.1322,
+      "step": 1435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5197145525060703e-05,
+      "loss": 0.1321,
+      "step": 1436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190469012444196e-05,
+      "loss": 0.1268,
+      "step": 1437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183789331441e-05,
+      "loss": 0.134,
+      "step": 1438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177106486128557e-05,
+      "loss": 0.1211,
+      "step": 1439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170420480586233e-05,
+      "loss": 0.1245,
+      "step": 1440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163731318895324e-05,
+      "loss": 0.1245,
+      "step": 1441
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157039005139065e-05,
+      "loss": 0.1156,
+      "step": 1442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5150343543402592e-05,
+      "loss": 0.1331,
+      "step": 1443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143644937772984e-05,
+      "loss": 0.1281,
+      "step": 1444
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136943192339224e-05,
+      "loss": 0.1429,
+      "step": 1445
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130238311192222e-05,
+      "loss": 0.1315,
+      "step": 1446
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5123530298424799e-05,
+      "loss": 0.1216,
+      "step": 1447
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511681915813168e-05,
+      "loss": 0.1383,
+      "step": 1448
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110104894409516e-05,
+      "loss": 0.1424,
+      "step": 1449
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510338751135685e-05,
+      "loss": 0.1349,
+      "step": 1450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5096667013074128e-05,
+      "loss": 0.1371,
+      "step": 1451
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089943403663714e-05,
+      "loss": 0.129,
+      "step": 1452
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5083216687229852e-05,
+      "loss": 0.1147,
+      "step": 1453
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507648686787869e-05,
+      "loss": 0.1296,
+      "step": 1454
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506975394971828e-05,
+      "loss": 0.1367,
+      "step": 1455
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063017936858551e-05,
+      "loss": 0.1163,
+      "step": 1456
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056278833411324e-05,
+      "loss": 0.1329,
+      "step": 1457
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049536643490312e-05,
+      "loss": 0.1234,
+      "step": 1458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5042791371211111e-05,
+      "loss": 0.1455,
+      "step": 1459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5036043020691196e-05,
+      "loss": 0.1204,
+      "step": 1460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502929159604992e-05,
+      "loss": 0.1298,
+      "step": 1461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022537101408516e-05,
+      "loss": 0.1292,
+      "step": 1462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501577954089009e-05,
+      "loss": 0.1237,
+      "step": 1463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5009018918619616e-05,
+      "loss": 0.136,
+      "step": 1464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5002255238723945e-05,
+      "loss": 0.131,
+      "step": 1465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995488505331782e-05,
+      "loss": 0.1274,
+      "step": 1466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988718722573707e-05,
+      "loss": 0.137,
+      "step": 1467
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498194589458216e-05,
+      "loss": 0.1293,
+      "step": 1468
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975170025491434e-05,
+      "loss": 0.1261,
+      "step": 1469
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968391119437686e-05,
+      "loss": 0.12,
+      "step": 1470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496160918055892e-05,
+      "loss": 0.122,
+      "step": 1471
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4954824212994993e-05,
+      "loss": 0.1257,
+      "step": 1472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948036220887614e-05,
+      "loss": 0.1406,
+      "step": 1473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4941245208380335e-05,
+      "loss": 0.1284,
+      "step": 1474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934451179618553e-05,
+      "loss": 0.1339,
+      "step": 1475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4927654138749503e-05,
+      "loss": 0.1306,
+      "step": 1476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920854089922267e-05,
+      "loss": 0.1236,
+      "step": 1477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914051037287755e-05,
+      "loss": 0.1251,
+      "step": 1478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4907244984998714e-05,
+      "loss": 0.1214,
+      "step": 1479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900435937209723e-05,
+      "loss": 0.1373,
+      "step": 1480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893623898077181e-05,
+      "loss": 0.1322,
+      "step": 1481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886808871759328e-05,
+      "loss": 0.1212,
+      "step": 1482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879990862416218e-05,
+      "loss": 0.1254,
+      "step": 1483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873169874209724e-05,
+      "loss": 0.1345,
+      "step": 1484
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486634591130354e-05,
+      "loss": 0.1254,
+      "step": 1485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4859518977863181e-05,
+      "loss": 0.1327,
+      "step": 1486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.485268907805597e-05,
+      "loss": 0.1274,
+      "step": 1487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484585621605104e-05,
+      "loss": 0.1315,
+      "step": 1488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839020396019334e-05,
+      "loss": 0.12,
+      "step": 1489
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832181622133599e-05,
+      "loss": 0.1294,
+      "step": 1490
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4825339898568389e-05,
+      "loss": 0.1247,
+      "step": 1491
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818495229500057e-05,
+      "loss": 0.1324,
+      "step": 1492
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811647619106748e-05,
+      "loss": 0.1272,
+      "step": 1493
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480479707156841e-05,
+      "loss": 0.1249,
+      "step": 1494
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797943591066784e-05,
+      "loss": 0.1264,
+      "step": 1495
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791087181785396e-05,
+      "loss": 0.1315,
+      "step": 1496
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784227847909562e-05,
+      "loss": 0.1297,
+      "step": 1497
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4777365593626384e-05,
+      "loss": 0.1256,
+      "step": 1498
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477050042312475e-05,
+      "loss": 0.1231,
+      "step": 1499
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4763632340595319e-05,
+      "loss": 0.134,
+      "step": 1500
+    },
+    {
+      "epoch": 1.08,
+      "eval_loss": 0.1258772611618042,
+      "eval_runtime": 712.5917,
+      "eval_samples_per_second": 70.166,
+      "eval_steps_per_second": 2.193,
+      "step": 1500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756761350230542e-05,
+      "loss": 0.1382,
+      "step": 1501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4749887456224626e-05,
+      "loss": 0.1352,
+      "step": 1502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743010662773567e-05,
+      "loss": 0.1251,
+      "step": 1503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736130974075123e-05,
+      "loss": 0.1305,
+      "step": 1504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729248394328817e-05,
+      "loss": 0.1331,
+      "step": 1505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722362927735947e-05,
+      "loss": 0.1302,
+      "step": 1506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715474578499565e-05,
+      "loss": 0.1227,
+      "step": 1507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470858335082448e-05,
+      "loss": 0.1363,
+      "step": 1508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701689248917267e-05,
+      "loss": 0.1269,
+      "step": 1509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4694792276986252e-05,
+      "loss": 0.137,
+      "step": 1510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687892439241508e-05,
+      "loss": 0.1172,
+      "step": 1511
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680989739894864e-05,
+      "loss": 0.1296,
+      "step": 1512
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4674084183159894e-05,
+      "loss": 0.1283,
+      "step": 1513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4667175773251911e-05,
+      "loss": 0.127,
+      "step": 1514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660264514387978e-05,
+      "loss": 0.1173,
+      "step": 1515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653350410786892e-05,
+      "loss": 0.1328,
+      "step": 1516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464643346666919e-05,
+      "loss": 0.1378,
+      "step": 1517
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639513686257134e-05,
+      "loss": 0.1268,
+      "step": 1518
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463259107377473e-05,
+      "loss": 0.1349,
+      "step": 1519
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625665633447702e-05,
+      "loss": 0.1273,
+      "step": 1520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618737369503506e-05,
+      "loss": 0.1388,
+      "step": 1521
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611806286171324e-05,
+      "loss": 0.1327,
+      "step": 1522
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4604872387682052e-05,
+      "loss": 0.1303,
+      "step": 1523
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459793567826831e-05,
+      "loss": 0.1236,
+      "step": 1524
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590996162164428e-05,
+      "loss": 0.1237,
+      "step": 1525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584053843606456e-05,
+      "loss": 0.1306,
+      "step": 1526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457710872683215e-05,
+      "loss": 0.1275,
+      "step": 1527
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4570160816080978e-05,
+      "loss": 0.1351,
+      "step": 1528
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563210115594109e-05,
+      "loss": 0.1305,
+      "step": 1529
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556256629614417e-05,
+      "loss": 0.1339,
+      "step": 1530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454930036238648e-05,
+      "loss": 0.1278,
+      "step": 1531
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542341318156568e-05,
+      "loss": 0.1317,
+      "step": 1532
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4535379501172646e-05,
+      "loss": 0.1283,
+      "step": 1533
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528414915684376e-05,
+      "loss": 0.1352,
+      "step": 1534
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452144756594311e-05,
+      "loss": 0.128,
+      "step": 1535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514477456201881e-05,
+      "loss": 0.1346,
+      "step": 1536
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507504590715413e-05,
+      "loss": 0.1362,
+      "step": 1537
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450052897374011e-05,
+      "loss": 0.1269,
+      "step": 1538
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449355060953406e-05,
+      "loss": 0.1224,
+      "step": 1539
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486569502357018e-05,
+      "loss": 0.1127,
+      "step": 1540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.447958565647042e-05,
+      "loss": 0.1402,
+      "step": 1541
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4472599076137374e-05,
+      "loss": 0.1348,
+      "step": 1542
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465609765622656e-05,
+      "loss": 0.1256,
+      "step": 1543
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458617729192713e-05,
+      "loss": 0.127,
+      "step": 1544
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451622971115641e-05,
+      "loss": 0.1288,
+      "step": 1545
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444462549566122e-05,
+      "loss": 0.1278,
+      "step": 1546
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437625307100869e-05,
+      "loss": 0.1232,
+      "step": 1547
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4430622409707674e-05,
+      "loss": 0.1215,
+      "step": 1548
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423616807756367e-05,
+      "loss": 0.1299,
+      "step": 1549
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4416608505523341e-05,
+      "loss": 0.1339,
+      "step": 1550
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440959750728663e-05,
+      "loss": 0.1271,
+      "step": 1551
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402583817325915e-05,
+      "loss": 0.1297,
+      "step": 1552
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395567439922517e-05,
+      "loss": 0.1283,
+      "step": 1553
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388548379359407e-05,
+      "loss": 0.122,
+      "step": 1554
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4381526639921183e-05,
+      "loss": 0.1374,
+      "step": 1555
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4374502225894085e-05,
+      "loss": 0.1357,
+      "step": 1556
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4367475141565984e-05,
+      "loss": 0.1253,
+      "step": 1557
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360445391226384e-05,
+      "loss": 0.1239,
+      "step": 1558
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4353412979166404e-05,
+      "loss": 0.1294,
+      "step": 1559
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4346377909678805e-05,
+      "loss": 0.1283,
+      "step": 1560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433934018705796e-05,
+      "loss": 0.1216,
+      "step": 1561
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332299815599862e-05,
+      "loss": 0.1157,
+      "step": 1562
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325256799602122e-05,
+      "loss": 0.1326,
+      "step": 1563
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431821114336397e-05,
+      "loss": 0.1266,
+      "step": 1564
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431116285118624e-05,
+      "loss": 0.1269,
+      "step": 1565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304111927371379e-05,
+      "loss": 0.1244,
+      "step": 1566
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4297058376223438e-05,
+      "loss": 0.1213,
+      "step": 1567
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4290002202048077e-05,
+      "loss": 0.1432,
+      "step": 1568
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4282943409152553e-05,
+      "loss": 0.1373,
+      "step": 1569
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427588200184572e-05,
+      "loss": 0.1379,
+      "step": 1570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426881798443803e-05,
+      "loss": 0.1273,
+      "step": 1571
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261751361241536e-05,
+      "loss": 0.1367,
+      "step": 1572
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254682136569865e-05,
+      "loss": 0.124,
+      "step": 1573
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247610314738247e-05,
+      "loss": 0.1272,
+      "step": 1574
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240535900063487e-05,
+      "loss": 0.1372,
+      "step": 1575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4233458896863984e-05,
+      "loss": 0.1291,
+      "step": 1576
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4226379309459702e-05,
+      "loss": 0.1298,
+      "step": 1577
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219297142172196e-05,
+      "loss": 0.1297,
+      "step": 1578
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4212212399324591e-05,
+      "loss": 0.1288,
+      "step": 1579
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4205125085241583e-05,
+      "loss": 0.1334,
+      "step": 1580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198035204249436e-05,
+      "loss": 0.1212,
+      "step": 1581
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4190942760675985e-05,
+      "loss": 0.1208,
+      "step": 1582
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4183847758850629e-05,
+      "loss": 0.1237,
+      "step": 1583
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4176750203104324e-05,
+      "loss": 0.1293,
+      "step": 1584
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4169650097769592e-05,
+      "loss": 0.1316,
+      "step": 1585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162547447180501e-05,
+      "loss": 0.1202,
+      "step": 1586
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155442255672687e-05,
+      "loss": 0.1322,
+      "step": 1587
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4148334527583325e-05,
+      "loss": 0.1287,
+      "step": 1588
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4141224267251143e-05,
+      "loss": 0.1296,
+      "step": 1589
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134111479016416e-05,
+      "loss": 0.1289,
+      "step": 1590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412699616722096e-05,
+      "loss": 0.1299,
+      "step": 1591
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4119878336208129e-05,
+      "loss": 0.1293,
+      "step": 1592
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112757990322816e-05,
+      "loss": 0.1305,
+      "step": 1593
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105635133911459e-05,
+      "loss": 0.1224,
+      "step": 1594
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098509771322015e-05,
+      "loss": 0.1341,
+      "step": 1595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091381906903975e-05,
+      "loss": 0.1253,
+      "step": 1596
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4084251545008363e-05,
+      "loss": 0.128,
+      "step": 1597
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4077118689987716e-05,
+      "loss": 0.1286,
+      "step": 1598
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069983346196105e-05,
+      "loss": 0.1267,
+      "step": 1599
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062845517989114e-05,
+      "loss": 0.1136,
+      "step": 1600
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055705209723845e-05,
+      "loss": 0.1323,
+      "step": 1601
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404856242575891e-05,
+      "loss": 0.1301,
+      "step": 1602
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041417170454436e-05,
+      "loss": 0.117,
+      "step": 1603
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034269448172062e-05,
+      "loss": 0.1331,
+      "step": 1604
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4027119263274927e-05,
+      "loss": 0.121,
+      "step": 1605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4019966620127674e-05,
+      "loss": 0.1274,
+      "step": 1606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4012811523096448e-05,
+      "loss": 0.1238,
+      "step": 1607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005653976548895e-05,
+      "loss": 0.1267,
+      "step": 1608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998493984854146e-05,
+      "loss": 0.1238,
+      "step": 1609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991331552382839e-05,
+      "loss": 0.1338,
+      "step": 1610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.398416668350709e-05,
+      "loss": 0.1286,
+      "step": 1611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976999382600509e-05,
+      "loss": 0.1224,
+      "step": 1612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3969829654038187e-05,
+      "loss": 0.1285,
+      "step": 1613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962657502196697e-05,
+      "loss": 0.1218,
+      "step": 1614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3955482931454095e-05,
+      "loss": 0.1248,
+      "step": 1615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3948305946189909e-05,
+      "loss": 0.1214,
+      "step": 1616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941126550785145e-05,
+      "loss": 0.1153,
+      "step": 1617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3933944749622275e-05,
+      "loss": 0.1327,
+      "step": 1618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926760547085242e-05,
+      "loss": 0.1146,
+      "step": 1619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3919573947559462e-05,
+      "loss": 0.1389,
+      "step": 1620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912384955431801e-05,
+      "loss": 0.1218,
+      "step": 1621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905193575090595e-05,
+      "loss": 0.122,
+      "step": 1622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3897999810925634e-05,
+      "loss": 0.1331,
+      "step": 1623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3890803667328162e-05,
+      "loss": 0.1355,
+      "step": 1624
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883605148690883e-05,
+      "loss": 0.1373,
+      "step": 1625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387640425940794e-05,
+      "loss": 0.1193,
+      "step": 1626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386920100387493e-05,
+      "loss": 0.1274,
+      "step": 1627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3861995386488893e-05,
+      "loss": 0.1229,
+      "step": 1628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854787411648311e-05,
+      "loss": 0.1358,
+      "step": 1629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3847577083753105e-05,
+      "loss": 0.125,
+      "step": 1630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840364407204634e-05,
+      "loss": 0.1231,
+      "step": 1631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383314938640568e-05,
+      "loss": 0.1211,
+      "step": 1632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825932025760476e-05,
+      "loss": 0.1278,
+      "step": 1633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818712329674664e-05,
+      "loss": 0.1304,
+      "step": 1634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811490302555321e-05,
+      "loss": 0.1271,
+      "step": 1635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804265948810951e-05,
+      "loss": 0.1216,
+      "step": 1636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797039272851464e-05,
+      "loss": 0.1232,
+      "step": 1637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789810279088203e-05,
+      "loss": 0.1342,
+      "step": 1638
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782578971933912e-05,
+      "loss": 0.1301,
+      "step": 1639
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3775345355802766e-05,
+      "loss": 0.1302,
+      "step": 1640
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768109435110324e-05,
+      "loss": 0.1238,
+      "step": 1641
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760871214273575e-05,
+      "loss": 0.1312,
+      "step": 1642
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753630697710895e-05,
+      "loss": 0.1304,
+      "step": 1643
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746387889842076e-05,
+      "loss": 0.1368,
+      "step": 1644
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739142795088298e-05,
+      "loss": 0.1344,
+      "step": 1645
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731895417872144e-05,
+      "loss": 0.1273,
+      "step": 1646
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3724645762617578e-05,
+      "loss": 0.1386,
+      "step": 1647
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3717393833749974e-05,
+      "loss": 0.1255,
+      "step": 1648
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710139635696077e-05,
+      "loss": 0.1339,
+      "step": 1649
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702883172884023e-05,
+      "loss": 0.1281,
+      "step": 1650
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695624449743335e-05,
+      "loss": 0.1258,
+      "step": 1651
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3688363470704906e-05,
+      "loss": 0.1289,
+      "step": 1652
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3681100240201017e-05,
+      "loss": 0.1248,
+      "step": 1653
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673834762665314e-05,
+      "loss": 0.1209,
+      "step": 1654
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666567042532823e-05,
+      "loss": 0.1284,
+      "step": 1655
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659297084239929e-05,
+      "loss": 0.1248,
+      "step": 1656
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3652024892224394e-05,
+      "loss": 0.1384,
+      "step": 1657
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3644750470925331e-05,
+      "loss": 0.132,
+      "step": 1658
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637473824783229e-05,
+      "loss": 0.123,
+      "step": 1659
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630194958239926e-05,
+      "loss": 0.1283,
+      "step": 1660
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622913875738613e-05,
+      "loss": 0.1227,
+      "step": 1661
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361563058172384e-05,
+      "loss": 0.1223,
+      "step": 1662
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608345080641504e-05,
+      "loss": 0.1282,
+      "step": 1663
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3601057376938847e-05,
+      "loss": 0.1373,
+      "step": 1664
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359376747506446e-05,
+      "loss": 0.123,
+      "step": 1665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586475379468271e-05,
+      "loss": 0.131,
+      "step": 1666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3579181094601557e-05,
+      "loss": 0.1178,
+      "step": 1667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571884624916918e-05,
+      "loss": 0.1258,
+      "step": 1668
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3564585974868296e-05,
+      "loss": 0.1266,
+      "step": 1669
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355728514891096e-05,
+      "loss": 0.1239,
+      "step": 1670
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549982151501512e-05,
+      "loss": 0.1283,
+      "step": 1671
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542676987097878e-05,
+      "loss": 0.1244,
+      "step": 1672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535369660159301e-05,
+      "loss": 0.1269,
+      "step": 1673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352806017514635e-05,
+      "loss": 0.138,
+      "step": 1674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520748536520908e-05,
+      "loss": 0.1322,
+      "step": 1675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513434748746181e-05,
+      "loss": 0.1256,
+      "step": 1676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506118816286675e-05,
+      "loss": 0.1279,
+      "step": 1677
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498800743608212e-05,
+      "loss": 0.1225,
+      "step": 1678
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3491480535177916e-05,
+      "loss": 0.1201,
+      "step": 1679
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484158195464224e-05,
+      "loss": 0.1252,
+      "step": 1680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3476833728936862e-05,
+      "loss": 0.1364,
+      "step": 1681
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469507140066862e-05,
+      "loss": 0.1219,
+      "step": 1682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462178433326548e-05,
+      "loss": 0.1316,
+      "step": 1683
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454847613189542e-05,
+      "loss": 0.1322,
+      "step": 1684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447514684130747e-05,
+      "loss": 0.1212,
+      "step": 1685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440179650626364e-05,
+      "loss": 0.1233,
+      "step": 1686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3432842517153868e-05,
+      "loss": 0.1255,
+      "step": 1687
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425503288192024e-05,
+      "loss": 0.1245,
+      "step": 1688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3418161968220873e-05,
+      "loss": 0.1183,
+      "step": 1689
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410818561721735e-05,
+      "loss": 0.1323,
+      "step": 1690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3403473073177198e-05,
+      "loss": 0.1257,
+      "step": 1691
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3396125507071128e-05,
+      "loss": 0.1289,
+      "step": 1692
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3388775867888648e-05,
+      "loss": 0.1433,
+      "step": 1693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338142416011616e-05,
+      "loss": 0.1323,
+      "step": 1694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374070388241325e-05,
+      "loss": 0.1295,
+      "step": 1695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366714556753056e-05,
+      "loss": 0.1259,
+      "step": 1696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3359356670141528e-05,
+      "loss": 0.1168,
+      "step": 1697
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3351996732898176e-05,
+      "loss": 0.131,
+      "step": 1698
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.334463474951568e-05,
+      "loss": 0.1364,
+      "step": 1699
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333727072448797e-05,
+      "loss": 0.1332,
+      "step": 1700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3329904662310227e-05,
+      "loss": 0.1272,
+      "step": 1701
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.332253656747887e-05,
+      "loss": 0.1328,
+      "step": 1702
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315166444491557e-05,
+      "loss": 0.1258,
+      "step": 1703
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307794297847191e-05,
+      "loss": 0.1263,
+      "step": 1704
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330042013204591e-05,
+      "loss": 0.1223,
+      "step": 1705
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3293043951589077e-05,
+      "loss": 0.1261,
+      "step": 1706
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285665760979292e-05,
+      "loss": 0.113,
+      "step": 1707
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278285564720376e-05,
+      "loss": 0.1364,
+      "step": 1708
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270903367317383e-05,
+      "loss": 0.126,
+      "step": 1709
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3263519173276578e-05,
+      "loss": 0.1352,
+      "step": 1710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256132987105452e-05,
+      "loss": 0.1213,
+      "step": 1711
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324874481331271e-05,
+      "loss": 0.1341,
+      "step": 1712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241354656408275e-05,
+      "loss": 0.1344,
+      "step": 1713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233962520903265e-05,
+      "loss": 0.1262,
+      "step": 1714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3226568411310026e-05,
+      "loss": 0.1238,
+      "step": 1715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219172332142099e-05,
+      "loss": 0.1251,
+      "step": 1716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211774287914224e-05,
+      "loss": 0.1315,
+      "step": 1717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3204374283142342e-05,
+      "loss": 0.1229,
+      "step": 1718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3196972322343602e-05,
+      "loss": 0.1269,
+      "step": 1719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189568410036333e-05,
+      "loss": 0.1391,
+      "step": 1720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3182162550740063e-05,
+      "loss": 0.1272,
+      "step": 1721
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174754748975503e-05,
+      "loss": 0.1326,
+      "step": 1722
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167345009264553e-05,
+      "loss": 0.1324,
+      "step": 1723
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31599333361303e-05,
+      "loss": 0.1365,
+      "step": 1724
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3152519734097003e-05,
+      "loss": 0.1279,
+      "step": 1725
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145104207690102e-05,
+      "loss": 0.1253,
+      "step": 1726
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137686761436212e-05,
+      "loss": 0.1244,
+      "step": 1727
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3130267399863124e-05,
+      "loss": 0.1302,
+      "step": 1728
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312284612749979e-05,
+      "loss": 0.1334,
+      "step": 1729
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115422948876333e-05,
+      "loss": 0.1231,
+      "step": 1730
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107997868524042e-05,
+      "loss": 0.1301,
+      "step": 1731
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100570890975359e-05,
+      "loss": 0.1255,
+      "step": 1732
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093142020763895e-05,
+      "loss": 0.1301,
+      "step": 1733
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085711262424406e-05,
+      "loss": 0.1307,
+      "step": 1734
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078278620492806e-05,
+      "loss": 0.1203,
+      "step": 1735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070844099506158e-05,
+      "loss": 0.1263,
+      "step": 1736
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063407704002676e-05,
+      "loss": 0.1276,
+      "step": 1737
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055969438521703e-05,
+      "loss": 0.1266,
+      "step": 1738
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048529307603745e-05,
+      "loss": 0.1305,
+      "step": 1739
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041087315790434e-05,
+      "loss": 0.1279,
+      "step": 1740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033643467624541e-05,
+      "loss": 0.1324,
+      "step": 1741
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026197767649964e-05,
+      "loss": 0.1342,
+      "step": 1742
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3018750220411742e-05,
+      "loss": 0.1371,
+      "step": 1743
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011300830456035e-05,
+      "loss": 0.1425,
+      "step": 1744
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003849602330126e-05,
+      "loss": 0.1186,
+      "step": 1745
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996396540582428e-05,
+      "loss": 0.131,
+      "step": 1746
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988941649762463e-05,
+      "loss": 0.1187,
+      "step": 1747
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981484934420881e-05,
+      "loss": 0.1215,
+      "step": 1748
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2974026399109437e-05,
+      "loss": 0.1179,
+      "step": 1749
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966566048380998e-05,
+      "loss": 0.1334,
+      "step": 1750
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959103886789546e-05,
+      "loss": 0.1275,
+      "step": 1751
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951639918890157e-05,
+      "loss": 0.1194,
+      "step": 1752
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944174149239022e-05,
+      "loss": 0.134,
+      "step": 1753
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293670658239342e-05,
+      "loss": 0.1299,
+      "step": 1754
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2929237222911737e-05,
+      "loss": 0.1313,
+      "step": 1755
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2921766075353446e-05,
+      "loss": 0.1425,
+      "step": 1756
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914293144279117e-05,
+      "loss": 0.1346,
+      "step": 1757
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906818434250403e-05,
+      "loss": 0.1229,
+      "step": 1758
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289934194983005e-05,
+      "loss": 0.1207,
+      "step": 1759
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891863695581884e-05,
+      "loss": 0.1321,
+      "step": 1760
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884383676070806e-05,
+      "loss": 0.1211,
+      "step": 1761
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876901895862799e-05,
+      "loss": 0.1286,
+      "step": 1762
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2869418359524929e-05,
+      "loss": 0.1266,
+      "step": 1763
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2861933071625314e-05,
+      "loss": 0.1307,
+      "step": 1764
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854446036733164e-05,
+      "loss": 0.1227,
+      "step": 1765
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.284695725941874e-05,
+      "loss": 0.1284,
+      "step": 1766
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839466744253368e-05,
+      "loss": 0.1237,
+      "step": 1767
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831974495809445e-05,
+      "loss": 0.1187,
+      "step": 1768
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824480518660414e-05,
+      "loss": 0.1285,
+      "step": 1769
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816984817380779e-05,
+      "loss": 0.131,
+      "step": 1770
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2809487396546097e-05,
+      "loss": 0.1247,
+      "step": 1771
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280198826073297e-05,
+      "loss": 0.143,
+      "step": 1772
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794487414519053e-05,
+      "loss": 0.1169,
+      "step": 1773
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2786984862483043e-05,
+      "loss": 0.1248,
+      "step": 1774
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779480609204677e-05,
+      "loss": 0.1265,
+      "step": 1775
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2771974659264731e-05,
+      "loss": 0.1242,
+      "step": 1776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2764467017245014e-05,
+      "loss": 0.1266,
+      "step": 1777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756957687728377e-05,
+      "loss": 0.1365,
+      "step": 1778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749446675298689e-05,
+      "loss": 0.1198,
+      "step": 1779
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2741933984540853e-05,
+      "loss": 0.1234,
+      "step": 1780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2734419620040797e-05,
+      "loss": 0.1242,
+      "step": 1781
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2726903586385468e-05,
+      "loss": 0.1235,
+      "step": 1782
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719385888162836e-05,
+      "loss": 0.1267,
+      "step": 1783
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271186652996188e-05,
+      "loss": 0.1264,
+      "step": 1784
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704345516372599e-05,
+      "loss": 0.118,
+      "step": 1785
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2696822851986e-05,
+      "loss": 0.1252,
+      "step": 1786
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2689298541394093e-05,
+      "loss": 0.129,
+      "step": 1787
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2681772589189904e-05,
+      "loss": 0.1224,
+      "step": 1788
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2674244999967451e-05,
+      "loss": 0.1216,
+      "step": 1789
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2666715778321756e-05,
+      "loss": 0.1301,
+      "step": 1790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2659184928848836e-05,
+      "loss": 0.1197,
+      "step": 1791
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2651652456145702e-05,
+      "loss": 0.1337,
+      "step": 1792
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644118364810355e-05,
+      "loss": 0.1198,
+      "step": 1793
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636582659441788e-05,
+      "loss": 0.1242,
+      "step": 1794
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629045344639975e-05,
+      "loss": 0.1259,
+      "step": 1795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621506425005872e-05,
+      "loss": 0.132,
+      "step": 1796
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613965905141414e-05,
+      "loss": 0.1145,
+      "step": 1797
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2606423789649523e-05,
+      "loss": 0.1168,
+      "step": 1798
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2598880083134077e-05,
+      "loss": 0.1207,
+      "step": 1799
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2591334790199942e-05,
+      "loss": 0.1235,
+      "step": 1800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583787915452947e-05,
+      "loss": 0.1271,
+      "step": 1801
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257623946349988e-05,
+      "loss": 0.126,
+      "step": 1802
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2568689438948498e-05,
+      "loss": 0.1263,
+      "step": 1803
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2561137846407519e-05,
+      "loss": 0.1356,
+      "step": 1804
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553584690486616e-05,
+      "loss": 0.124,
+      "step": 1805
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546029975796412e-05,
+      "loss": 0.1336,
+      "step": 1806
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538473706948494e-05,
+      "loss": 0.1305,
+      "step": 1807
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530915888555381e-05,
+      "loss": 0.1343,
+      "step": 1808
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2523356525230554e-05,
+      "loss": 0.1231,
+      "step": 1809
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515795621588428e-05,
+      "loss": 0.1306,
+      "step": 1810
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250823318224435e-05,
+      "loss": 0.1208,
+      "step": 1811
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500669211814626e-05,
+      "loss": 0.1335,
+      "step": 1812
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493103714916484e-05,
+      "loss": 0.1296,
+      "step": 1813
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485536696168081e-05,
+      "loss": 0.1242,
+      "step": 1814
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477968160188505e-05,
+      "loss": 0.1305,
+      "step": 1815
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470398111597776e-05,
+      "loss": 0.119,
+      "step": 1816
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246282655501683e-05,
+      "loss": 0.1295,
+      "step": 1817
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455253495067528e-05,
+      "loss": 0.131,
+      "step": 1818
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447678936372647e-05,
+      "loss": 0.1289,
+      "step": 1819
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440102883555878e-05,
+      "loss": 0.1264,
+      "step": 1820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2432525341241827e-05,
+      "loss": 0.1216,
+      "step": 1821
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424946314056008e-05,
+      "loss": 0.1386,
+      "step": 1822
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417365806624839e-05,
+      "loss": 0.122,
+      "step": 1823
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409783823575645e-05,
+      "loss": 0.1245,
+      "step": 1824
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402200369536647e-05,
+      "loss": 0.1362,
+      "step": 1825
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394615449136971e-05,
+      "loss": 0.1279,
+      "step": 1826
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2387029067006633e-05,
+      "loss": 0.1259,
+      "step": 1827
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237944122777654e-05,
+      "loss": 0.1333,
+      "step": 1828
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371851936078496e-05,
+      "loss": 0.1264,
+      "step": 1829
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2364261196545182e-05,
+      "loss": 0.128,
+      "step": 1830
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235666901381017e-05,
+      "loss": 0.1322,
+      "step": 1831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2349075392507907e-05,
+      "loss": 0.1308,
+      "step": 1832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2341480337273724e-05,
+      "loss": 0.1297,
+      "step": 1833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2333883852743824e-05,
+      "loss": 0.1252,
+      "step": 1834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326285943555282e-05,
+      "loss": 0.1286,
+      "step": 1835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318686614346049e-05,
+      "loss": 0.1321,
+      "step": 1836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2311085869754933e-05,
+      "loss": 0.1298,
+      "step": 1837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303483714421613e-05,
+      "loss": 0.1162,
+      "step": 1838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295880152986623e-05,
+      "loss": 0.1254,
+      "step": 1839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2288275190091363e-05,
+      "loss": 0.1245,
+      "step": 1840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280668830378084e-05,
+      "loss": 0.1224,
+      "step": 1841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273061078489889e-05,
+      "loss": 0.1355,
+      "step": 1842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265451939070733e-05,
+      "loss": 0.1315,
+      "step": 1843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2257841416765416e-05,
+      "loss": 0.1216,
+      "step": 1844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2250229516219586e-05,
+      "loss": 0.1236,
+      "step": 1845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242616242079723e-05,
+      "loss": 0.1301,
+      "step": 1846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2235001598993162e-05,
+      "loss": 0.1351,
+      "step": 1847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227385591608054e-05,
+      "loss": 0.1221,
+      "step": 1848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22197682245734e-05,
+      "loss": 0.1313,
+      "step": 1849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2212149502539017e-05,
+      "loss": 0.1382,
+      "step": 1850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2204529430155561e-05,
+      "loss": 0.1265,
+      "step": 1851
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196908012074502e-05,
+      "loss": 0.1351,
+      "step": 1852
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218928525294814e-05,
+      "loss": 0.1335,
+      "step": 1853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218166115742959e-05,
+      "loss": 0.1237,
+      "step": 1854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217403573017278e-05,
+      "loss": 0.1335,
+      "step": 1855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166408975832453e-05,
+      "loss": 0.1332,
+      "step": 1856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158780899064167e-05,
+      "loss": 0.1359,
+      "step": 1857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215115150452428e-05,
+      "loss": 0.1205,
+      "step": 1858
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2143520796869957e-05,
+      "loss": 0.1262,
+      "step": 1859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2135888780759164e-05,
+      "loss": 0.1313,
+      "step": 1860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212825546085067e-05,
+      "loss": 0.1281,
+      "step": 1861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2120620841804032e-05,
+      "loss": 0.1164,
+      "step": 1862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211298492827961e-05,
+      "loss": 0.1247,
+      "step": 1863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2105347724938543e-05,
+      "loss": 0.1364,
+      "step": 1864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097709236442763e-05,
+      "loss": 0.1375,
+      "step": 1865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209006946745499e-05,
+      "loss": 0.1278,
+      "step": 1866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082428422638724e-05,
+      "loss": 0.1255,
+      "step": 1867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074786106658238e-05,
+      "loss": 0.1158,
+      "step": 1868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2067142524178584e-05,
+      "loss": 0.132,
+      "step": 1869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205949767986559e-05,
+      "loss": 0.135,
+      "step": 1870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051851578385853e-05,
+      "loss": 0.1301,
+      "step": 1871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044204224406734e-05,
+      "loss": 0.1286,
+      "step": 1872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2036555622596365e-05,
+      "loss": 0.1284,
+      "step": 1873
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202890577762363e-05,
+      "loss": 0.121,
+      "step": 1874
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021254694158185e-05,
+      "loss": 0.1304,
+      "step": 1875
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2013602376870429e-05,
+      "loss": 0.1348,
+      "step": 1876
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200594883043152e-05,
+      "loss": 0.1358,
+      "step": 1877
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998294059513369e-05,
+      "loss": 0.123,
+      "step": 1878
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1990638068788627e-05,
+      "loss": 0.1328,
+      "step": 1879
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982980862930703e-05,
+      "loss": 0.1331,
+      "step": 1880
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1975322446613731e-05,
+      "loss": 0.1311,
+      "step": 1881
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967662824512596e-05,
+      "loss": 0.1315,
+      "step": 1882
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960002001302912e-05,
+      "loss": 0.1399,
+      "step": 1883
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195233998166103e-05,
+      "loss": 0.1361,
+      "step": 1884
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944676770264034e-05,
+      "loss": 0.1239,
+      "step": 1885
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1937012371789727e-05,
+      "loss": 0.1344,
+      "step": 1886
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1929346790916645e-05,
+      "loss": 0.1247,
+      "step": 1887
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192168003232404e-05,
+      "loss": 0.1275,
+      "step": 1888
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914012100691885e-05,
+      "loss": 0.1284,
+      "step": 1889
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1906343000700871e-05,
+      "loss": 0.1271,
+      "step": 1890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898672737032399e-05,
+      "loss": 0.1289,
+      "step": 1891
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1891001314368583e-05,
+      "loss": 0.1244,
+      "step": 1892
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1883328737392244e-05,
+      "loss": 0.1273,
+      "step": 1893
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875655010786901e-05,
+      "loss": 0.1284,
+      "step": 1894
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186798013923679e-05,
+      "loss": 0.1274,
+      "step": 1895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186030412742683e-05,
+      "loss": 0.1301,
+      "step": 1896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852626980042643e-05,
+      "loss": 0.1248,
+      "step": 1897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1844948701770546e-05,
+      "loss": 0.1242,
+      "step": 1898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837269297297544e-05,
+      "loss": 0.1347,
+      "step": 1899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1829588771311329e-05,
+      "loss": 0.1363,
+      "step": 1900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821907128500279e-05,
+      "loss": 0.1261,
+      "step": 1901
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181422437355345e-05,
+      "loss": 0.1208,
+      "step": 1902
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1806540511160581e-05,
+      "loss": 0.1255,
+      "step": 1903
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798855546012085e-05,
+      "loss": 0.1292,
+      "step": 1904
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791169482799053e-05,
+      "loss": 0.1277,
+      "step": 1905
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1783482326213236e-05,
+      "loss": 0.1212,
+      "step": 1906
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1775794080947065e-05,
+      "loss": 0.1245,
+      "step": 1907
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1768104751693625e-05,
+      "loss": 0.1267,
+      "step": 1908
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760414343146667e-05,
+      "loss": 0.1386,
+      "step": 1909
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1752722860000602e-05,
+      "loss": 0.1264,
+      "step": 1910
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745030306950495e-05,
+      "loss": 0.1415,
+      "step": 1911
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1737336688692068e-05,
+      "loss": 0.1342,
+      "step": 1912
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729642009921685e-05,
+      "loss": 0.1277,
+      "step": 1913
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721946275336365e-05,
+      "loss": 0.1357,
+      "step": 1914
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1714249489633766e-05,
+      "loss": 0.1283,
+      "step": 1915
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706551657512195e-05,
+      "loss": 0.1298,
+      "step": 1916
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698852783670591e-05,
+      "loss": 0.1211,
+      "step": 1917
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691152872808529e-05,
+      "loss": 0.1315,
+      "step": 1918
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1683451929626224e-05,
+      "loss": 0.1179,
+      "step": 1919
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1675749958824511e-05,
+      "loss": 0.1302,
+      "step": 1920
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668046965104863e-05,
+      "loss": 0.122,
+      "step": 1921
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660342953169368e-05,
+      "loss": 0.1297,
+      "step": 1922
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652637927720739e-05,
+      "loss": 0.1292,
+      "step": 1923
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164493189346231e-05,
+      "loss": 0.1324,
+      "step": 1924
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1637224855098029e-05,
+      "loss": 0.1289,
+      "step": 1925
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629516817332458e-05,
+      "loss": 0.1353,
+      "step": 1926
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621807784870767e-05,
+      "loss": 0.1184,
+      "step": 1927
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1614097762418733e-05,
+      "loss": 0.1402,
+      "step": 1928
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606386754682739e-05,
+      "loss": 0.1216,
+      "step": 1929
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598674766369768e-05,
+      "loss": 0.1296,
+      "step": 1930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590961802187406e-05,
+      "loss": 0.1337,
+      "step": 1931
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158324786684383e-05,
+      "loss": 0.1205,
+      "step": 1932
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1575532965047809e-05,
+      "loss": 0.123,
+      "step": 1933
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567817101508702e-05,
+      "loss": 0.1336,
+      "step": 1934
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560100280936464e-05,
+      "loss": 0.1227,
+      "step": 1935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552382508041623e-05,
+      "loss": 0.118,
+      "step": 1936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154466378753529e-05,
+      "loss": 0.127,
+      "step": 1937
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153694412412916e-05,
+      "loss": 0.1329,
+      "step": 1938
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529223522535502e-05,
+      "loss": 0.1305,
+      "step": 1939
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521501987467153e-05,
+      "loss": 0.1196,
+      "step": 1940
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513779523637527e-05,
+      "loss": 0.1294,
+      "step": 1941
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1506056135760597e-05,
+      "loss": 0.1238,
+      "step": 1942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498331828550906e-05,
+      "loss": 0.1277,
+      "step": 1943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490606606723559e-05,
+      "loss": 0.1288,
+      "step": 1944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148288047499421e-05,
+      "loss": 0.1245,
+      "step": 1945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1475153438079079e-05,
+      "loss": 0.1272,
+      "step": 1946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467425500694936e-05,
+      "loss": 0.1328,
+      "step": 1947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459696667559096e-05,
+      "loss": 0.1194,
+      "step": 1948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451966943389423e-05,
+      "loss": 0.136,
+      "step": 1949
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444236332904331e-05,
+      "loss": 0.141,
+      "step": 1950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436504840822767e-05,
+      "loss": 0.1252,
+      "step": 1951
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1428772471864216e-05,
+      "loss": 0.1286,
+      "step": 1952
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1421039230748705e-05,
+      "loss": 0.1288,
+      "step": 1953
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413305122196791e-05,
+      "loss": 0.1202,
+      "step": 1954
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140557015092956e-05,
+      "loss": 0.1221,
+      "step": 1955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1397834321668616e-05,
+      "loss": 0.1305,
+      "step": 1956
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390097639136101e-05,
+      "loss": 0.1339,
+      "step": 1957
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1382360108054674e-05,
+      "loss": 0.129,
+      "step": 1958
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374621733147501e-05,
+      "loss": 0.1312,
+      "step": 1959
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1366882519138281e-05,
+      "loss": 0.1205,
+      "step": 1960
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359142470751208e-05,
+      "loss": 0.1306,
+      "step": 1961
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351401592710997e-05,
+      "loss": 0.1393,
+      "step": 1962
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1343659889742865e-05,
+      "loss": 0.1279,
+      "step": 1963
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1335917366572531e-05,
+      "loss": 0.1295,
+      "step": 1964
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132817402792622e-05,
+      "loss": 0.1276,
+      "step": 1965
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1320429878530647e-05,
+      "loss": 0.1161,
+      "step": 1966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312684923113029e-05,
+      "loss": 0.1174,
+      "step": 1967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1304939166401075e-05,
+      "loss": 0.1203,
+      "step": 1968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1297192613122977e-05,
+      "loss": 0.1227,
+      "step": 1969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1289445268007417e-05,
+      "loss": 0.1147,
+      "step": 1970
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1281697135783563e-05,
+      "loss": 0.1322,
+      "step": 1971
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1273948221181058e-05,
+      "loss": 0.1251,
+      "step": 1972
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1266198528930026e-05,
+      "loss": 0.1246,
+      "step": 1973
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1258448063761063e-05,
+      "loss": 0.1291,
+      "step": 1974
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250696830405243e-05,
+      "loss": 0.1237,
+      "step": 1975
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1242944833594101e-05,
+      "loss": 0.1225,
+      "step": 1976
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123519207805964e-05,
+      "loss": 0.1303,
+      "step": 1977
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1227438568534334e-05,
+      "loss": 0.1227,
+      "step": 1978
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1219684309751107e-05,
+      "loss": 0.1299,
+      "step": 1979
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211929306443343e-05,
+      "loss": 0.1314,
+      "step": 1980
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204173563344884e-05,
+      "loss": 0.1248,
+      "step": 1981
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196417085190022e-05,
+      "loss": 0.1265,
+      "step": 1982
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188659876713495e-05,
+      "loss": 0.122,
+      "step": 1983
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180901942650493e-05,
+      "loss": 0.1281,
+      "step": 1984
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117314328773664e-05,
+      "loss": 0.1255,
+      "step": 1985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165383916708007e-05,
+      "loss": 0.1353,
+      "step": 1986
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157623834301103e-05,
+      "loss": 0.1197,
+      "step": 1987
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149863045252864e-05,
+      "loss": 0.1231,
+      "step": 1988
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142101554300664e-05,
+      "loss": 0.1185,
+      "step": 1989
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134339366182299e-05,
+      "loss": 0.1268,
+      "step": 1990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1126576485636e-05,
+      "loss": 0.1268,
+      "step": 1991
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1118812917400411e-05,
+      "loss": 0.1324,
+      "step": 1992
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111048666214607e-05,
+      "loss": 0.1355,
+      "step": 1993
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103283736818066e-05,
+      "loss": 0.1313,
+      "step": 1994
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109551813395069e-05,
+      "loss": 0.1209,
+      "step": 1995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1087751862352788e-05,
+      "loss": 0.1312,
+      "step": 1996
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079984926765077e-05,
+      "loss": 0.1238,
+      "step": 1997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1072217331928685e-05,
+      "loss": 0.1268,
+      "step": 1998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064449082585136e-05,
+      "loss": 0.1148,
+      "step": 1999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056680183476355e-05,
+      "loss": 0.1245,
+      "step": 2000
+    },
+    {
+      "epoch": 1.45,
+      "eval_loss": 0.12460336089134216,
+      "eval_runtime": 714.4442,
+      "eval_samples_per_second": 69.984,
+      "eval_steps_per_second": 2.188,
+      "step": 2000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.104891063934466e-05,
+      "loss": 0.131,
+      "step": 2001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1041140454932778e-05,
+      "loss": 0.1142,
+      "step": 2002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033369634983804e-05,
+      "loss": 0.1252,
+      "step": 2003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102559818424124e-05,
+      "loss": 0.1197,
+      "step": 2004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1017826107448963e-05,
+      "loss": 0.1274,
+      "step": 2005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010053409351236e-05,
+      "loss": 0.132,
+      "step": 2006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002280094692698e-05,
+      "loss": 0.1287,
+      "step": 2007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0994506168218371e-05,
+      "loss": 0.1259,
+      "step": 2008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986731634673641e-05,
+      "loss": 0.1295,
+      "step": 2009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0978956498804275e-05,
+      "loss": 0.1277,
+      "step": 2010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971180765356396e-05,
+      "loss": 0.1272,
+      "step": 2011
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0963404439076501e-05,
+      "loss": 0.1323,
+      "step": 2012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955627524711447e-05,
+      "loss": 0.119,
+      "step": 2013
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094785002700845e-05,
+      "loss": 0.1169,
+      "step": 2014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0940071950715079e-05,
+      "loss": 0.121,
+      "step": 2015
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932293300579258e-05,
+      "loss": 0.1256,
+      "step": 2016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924514081349263e-05,
+      "loss": 0.14,
+      "step": 2017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0916734297773712e-05,
+      "loss": 0.1308,
+      "step": 2018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908953954601576e-05,
+      "loss": 0.1334,
+      "step": 2019
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090117305658216e-05,
+      "loss": 0.1264,
+      "step": 2020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893391608465112e-05,
+      "loss": 0.1237,
+      "step": 2021
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885609615000414e-05,
+      "loss": 0.1272,
+      "step": 2022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087782708093838e-05,
+      "loss": 0.1249,
+      "step": 2023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870044011029656e-05,
+      "loss": 0.1205,
+      "step": 2024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0862260410025219e-05,
+      "loss": 0.1258,
+      "step": 2025
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085447628267636e-05,
+      "loss": 0.1211,
+      "step": 2026
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0846691633734699e-05,
+      "loss": 0.1196,
+      "step": 2027
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838906467952171e-05,
+      "loss": 0.1266,
+      "step": 2028
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831120790081027e-05,
+      "loss": 0.1382,
+      "step": 2029
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0823334604873836e-05,
+      "loss": 0.1212,
+      "step": 2030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815547917083468e-05,
+      "loss": 0.1396,
+      "step": 2031
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807760731463107e-05,
+      "loss": 0.1276,
+      "step": 2032
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799973052766232e-05,
+      "loss": 0.1251,
+      "step": 2033
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0792184885746635e-05,
+      "loss": 0.1293,
+      "step": 2034
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784396235158398e-05,
+      "loss": 0.1257,
+      "step": 2035
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776607105755899e-05,
+      "loss": 0.1282,
+      "step": 2036
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0768817502293806e-05,
+      "loss": 0.1339,
+      "step": 2037
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761027429527089e-05,
+      "loss": 0.1193,
+      "step": 2038
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0753236892210985e-05,
+      "loss": 0.1266,
+      "step": 2039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074544589510103e-05,
+      "loss": 0.128,
+      "step": 2040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0737654442953033e-05,
+      "loss": 0.1337,
+      "step": 2041
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729862540523086e-05,
+      "loss": 0.116,
+      "step": 2042
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0722070192567552e-05,
+      "loss": 0.1332,
+      "step": 2043
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714277403843065e-05,
+      "loss": 0.1231,
+      "step": 2044
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706484179106534e-05,
+      "loss": 0.1385,
+      "step": 2045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698690523115125e-05,
+      "loss": 0.1292,
+      "step": 2046
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690896440626277e-05,
+      "loss": 0.1281,
+      "step": 2047
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683101936397682e-05,
+      "loss": 0.1257,
+      "step": 2048
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675307015187292e-05,
+      "loss": 0.1165,
+      "step": 2049
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667511681753315e-05,
+      "loss": 0.129,
+      "step": 2050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.065971594085421e-05,
+      "loss": 0.123,
+      "step": 2051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651919797248685e-05,
+      "loss": 0.1225,
+      "step": 2052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0644123255695686e-05,
+      "loss": 0.1235,
+      "step": 2053
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0636326320954417e-05,
+      "loss": 0.1312,
+      "step": 2054
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628528997784308e-05,
+      "loss": 0.1365,
+      "step": 2055
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062073129094504e-05,
+      "loss": 0.1376,
+      "step": 2056
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061293320519651e-05,
+      "loss": 0.1319,
+      "step": 2057
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605134745298863e-05,
+      "loss": 0.1301,
+      "step": 2058
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597335916012463e-05,
+      "loss": 0.1126,
+      "step": 2059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589536722097906e-05,
+      "loss": 0.1355,
+      "step": 2060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581737168316003e-05,
+      "loss": 0.1312,
+      "step": 2061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057393725942779e-05,
+      "loss": 0.1281,
+      "step": 2062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566137000194514e-05,
+      "loss": 0.1381,
+      "step": 2063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558336395377648e-05,
+      "loss": 0.1267,
+      "step": 2064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550535449738866e-05,
+      "loss": 0.1292,
+      "step": 2065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054273416804005e-05,
+      "loss": 0.1377,
+      "step": 2066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.053493255504329e-05,
+      "loss": 0.1164,
+      "step": 2067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0527130615510877e-05,
+      "loss": 0.1188,
+      "step": 2068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05193283542053e-05,
+      "loss": 0.1187,
+      "step": 2069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511525775889248e-05,
+      "loss": 0.1406,
+      "step": 2070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503722885325607e-05,
+      "loss": 0.1263,
+      "step": 2071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495919687277442e-05,
+      "loss": 0.1286,
+      "step": 2072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488116186508014e-05,
+      "loss": 0.1349,
+      "step": 2073
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480312387780768e-05,
+      "loss": 0.1233,
+      "step": 2074
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472508295859329e-05,
+      "loss": 0.1349,
+      "step": 2075
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0464703915507502e-05,
+      "loss": 0.1235,
+      "step": 2076
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0456899251489265e-05,
+      "loss": 0.1262,
+      "step": 2077
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0449094308568776e-05,
+      "loss": 0.1237,
+      "step": 2078
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0441289091510356e-05,
+      "loss": 0.1224,
+      "step": 2079
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.04334836050785e-05,
+      "loss": 0.1301,
+      "step": 2080
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425677854037862e-05,
+      "loss": 0.1248,
+      "step": 2081
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0417871843153258e-05,
+      "loss": 0.1357,
+      "step": 2082
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0410065577189665e-05,
+      "loss": 0.1255,
+      "step": 2083
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040225906091222e-05,
+      "loss": 0.1384,
+      "step": 2084
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0394452299086196e-05,
+      "loss": 0.1229,
+      "step": 2085
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0386645296477039e-05,
+      "loss": 0.1274,
+      "step": 2086
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0378838057850326e-05,
+      "loss": 0.1215,
+      "step": 2087
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371030587971783e-05,
+      "loss": 0.1302,
+      "step": 2088
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363222891607276e-05,
+      "loss": 0.1258,
+      "step": 2089
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035541497352281e-05,
+      "loss": 0.1198,
+      "step": 2090
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347606838484527e-05,
+      "loss": 0.1302,
+      "step": 2091
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03397984912587e-05,
+      "loss": 0.127,
+      "step": 2092
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331989936611728e-05,
+      "loss": 0.1354,
+      "step": 2093
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0324181179310143e-05,
+      "loss": 0.1253,
+      "step": 2094
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0316372224120595e-05,
+      "loss": 0.1179,
+      "step": 2095
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308563075809859e-05,
+      "loss": 0.1252,
+      "step": 2096
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0300753739144824e-05,
+      "loss": 0.1292,
+      "step": 2097
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02929442188925e-05,
+      "loss": 0.1226,
+      "step": 2098
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285134519820005e-05,
+      "loss": 0.1262,
+      "step": 2099
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0277324646694564e-05,
+      "loss": 0.1261,
+      "step": 2100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0269514604283512e-05,
+      "loss": 0.1195,
+      "step": 2101
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261704397354287e-05,
+      "loss": 0.134,
+      "step": 2102
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0253894030674425e-05,
+      "loss": 0.1159,
+      "step": 2103
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246083509011561e-05,
+      "loss": 0.1198,
+      "step": 2104
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0238272837133426e-05,
+      "loss": 0.1231,
+      "step": 2105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023046201980784e-05,
+      "loss": 0.1247,
+      "step": 2106
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222651061802716e-05,
+      "loss": 0.133,
+      "step": 2107
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0214839967886044e-05,
+      "loss": 0.1292,
+      "step": 2108
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207028742825908e-05,
+      "loss": 0.1221,
+      "step": 2109
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199217391390466e-05,
+      "loss": 0.1107,
+      "step": 2110
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191405918347954e-05,
+      "loss": 0.1205,
+      "step": 2111
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0183594328466679e-05,
+      "loss": 0.1163,
+      "step": 2112
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017578262651503e-05,
+      "loss": 0.1253,
+      "step": 2113
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167970817261448e-05,
+      "loss": 0.1222,
+      "step": 2114
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160158905474455e-05,
+      "loss": 0.1338,
+      "step": 2115
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152346895922629e-05,
+      "loss": 0.1293,
+      "step": 2116
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144534793374605e-05,
+      "loss": 0.1232,
+      "step": 2117
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013672260259908e-05,
+      "loss": 0.1237,
+      "step": 2118
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128910328364803e-05,
+      "loss": 0.1242,
+      "step": 2119
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121097975440571e-05,
+      "loss": 0.1354,
+      "step": 2120
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113285548595231e-05,
+      "loss": 0.1328,
+      "step": 2121
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0105473052597676e-05,
+      "loss": 0.1303,
+      "step": 2122
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097660492216845e-05,
+      "loss": 0.1296,
+      "step": 2123
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089847872221708e-05,
+      "loss": 0.114,
+      "step": 2124
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0082035197381274e-05,
+      "loss": 0.1298,
+      "step": 2125
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074222472464593e-05,
+      "loss": 0.1381,
+      "step": 2126
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066409702240734e-05,
+      "loss": 0.1203,
+      "step": 2127
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00585968914788e-05,
+      "loss": 0.1422,
+      "step": 2128
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050784044947919e-05,
+      "loss": 0.1219,
+      "step": 2129
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042971167417239e-05,
+      "loss": 0.1272,
+      "step": 2130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003515826365593e-05,
+      "loss": 0.1314,
+      "step": 2131
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027345338433176e-05,
+      "loss": 0.1257,
+      "step": 2132
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001953239651817e-05,
+      "loss": 0.1462,
+      "step": 2133
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011719442680123e-05,
+      "loss": 0.127,
+      "step": 2134
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003906481688246e-05,
+      "loss": 0.1321,
+      "step": 2135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996093518311758e-06,
+      "loss": 0.1276,
+      "step": 2136
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98828055731988e-06,
+      "loss": 0.1199,
+      "step": 2137
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980467603481832e-06,
+      "loss": 0.1319,
+      "step": 2138
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972654661566826e-06,
+      "loss": 0.1287,
+      "step": 2139
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.964841736344071e-06,
+      "loss": 0.13,
+      "step": 2140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95702883258276e-06,
+      "loss": 0.1337,
+      "step": 2141
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949215955052083e-06,
+      "loss": 0.1289,
+      "step": 2142
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941403108521203e-06,
+      "loss": 0.1237,
+      "step": 2143
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933590297759271e-06,
+      "loss": 0.1331,
+      "step": 2144
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92577752753541e-06,
+      "loss": 0.1282,
+      "step": 2145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.917964802618728e-06,
+      "loss": 0.1239,
+      "step": 2146
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910152127778297e-06,
+      "loss": 0.134,
+      "step": 2147
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902339507783159e-06,
+      "loss": 0.1136,
+      "step": 2148
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.894526947402326e-06,
+      "loss": 0.1335,
+      "step": 2149
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886714451404772e-06,
+      "loss": 0.1349,
+      "step": 2150
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878902024559434e-06,
+      "loss": 0.1391,
+      "step": 2151
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871089671635202e-06,
+      "loss": 0.1341,
+      "step": 2152
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863277397400923e-06,
+      "loss": 0.1347,
+      "step": 2153
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855465206625397e-06,
+      "loss": 0.124,
+      "step": 2154
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847653104077373e-06,
+      "loss": 0.1291,
+      "step": 2155
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839841094525547e-06,
+      "loss": 0.1389,
+      "step": 2156
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832029182738555e-06,
+      "loss": 0.13,
+      "step": 2157
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.824217373484975e-06,
+      "loss": 0.1271,
+      "step": 2158
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.816405671533325e-06,
+      "loss": 0.1212,
+      "step": 2159
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808594081652051e-06,
+      "loss": 0.1251,
+      "step": 2160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.800782608609539e-06,
+      "loss": 0.1292,
+      "step": 2161
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792971257174097e-06,
+      "loss": 0.1389,
+      "step": 2162
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.785160032113961e-06,
+      "loss": 0.1212,
+      "step": 2163
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77734893819729e-06,
+      "loss": 0.1275,
+      "step": 2164
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769537980192159e-06,
+      "loss": 0.1231,
+      "step": 2165
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.761727162866574e-06,
+      "loss": 0.1284,
+      "step": 2166
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75391649098844e-06,
+      "loss": 0.1369,
+      "step": 2167
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.746105969325577e-06,
+      "loss": 0.1242,
+      "step": 2168
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.738295602645715e-06,
+      "loss": 0.1293,
+      "step": 2169
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73048539571649e-06,
+      "loss": 0.1285,
+      "step": 2170
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72267535330544e-06,
+      "loss": 0.1236,
+      "step": 2171
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.714865480179998e-06,
+      "loss": 0.1263,
+      "step": 2172
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707055781107503e-06,
+      "loss": 0.1333,
+      "step": 2173
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69924626085518e-06,
+      "loss": 0.1266,
+      "step": 2174
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691436924190146e-06,
+      "loss": 0.1338,
+      "step": 2175
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68362777587941e-06,
+      "loss": 0.124,
+      "step": 2176
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.675818820689862e-06,
+      "loss": 0.1252,
+      "step": 2177
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668010063388275e-06,
+      "loss": 0.1205,
+      "step": 2178
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660201508741304e-06,
+      "loss": 0.1225,
+      "step": 2179
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652393161515475e-06,
+      "loss": 0.1226,
+      "step": 2180
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644585026477193e-06,
+      "loss": 0.1366,
+      "step": 2181
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636777108392727e-06,
+      "loss": 0.1282,
+      "step": 2182
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62896941202822e-06,
+      "loss": 0.1281,
+      "step": 2183
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621161942149677e-06,
+      "loss": 0.1233,
+      "step": 2184
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.613354703522965e-06,
+      "loss": 0.123,
+      "step": 2185
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605547700913806e-06,
+      "loss": 0.1372,
+      "step": 2186
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597740939087787e-06,
+      "loss": 0.1327,
+      "step": 2187
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.589934422810339e-06,
+      "loss": 0.1318,
+      "step": 2188
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582128156846747e-06,
+      "loss": 0.1266,
+      "step": 2189
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574322145962143e-06,
+      "loss": 0.1216,
+      "step": 2190
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.566516394921506e-06,
+      "loss": 0.1283,
+      "step": 2191
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.558710908489648e-06,
+      "loss": 0.1183,
+      "step": 2192
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.550905691431229e-06,
+      "loss": 0.1205,
+      "step": 2193
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54310074851074e-06,
+      "loss": 0.1252,
+      "step": 2194
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.5352960844925e-06,
+      "loss": 0.1324,
+      "step": 2195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527491704140671e-06,
+      "loss": 0.1238,
+      "step": 2196
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519687612219232e-06,
+      "loss": 0.1243,
+      "step": 2197
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511883813491986e-06,
+      "loss": 0.1255,
+      "step": 2198
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504080312722558e-06,
+      "loss": 0.1205,
+      "step": 2199
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496277114674393e-06,
+      "loss": 0.1267,
+      "step": 2200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488474224110752e-06,
+      "loss": 0.1299,
+      "step": 2201
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.480671645794702e-06,
+      "loss": 0.1232,
+      "step": 2202
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.472869384489128e-06,
+      "loss": 0.1278,
+      "step": 2203
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465067444956716e-06,
+      "loss": 0.1209,
+      "step": 2204
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.457265831959955e-06,
+      "loss": 0.1333,
+      "step": 2205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449464550261137e-06,
+      "loss": 0.1358,
+      "step": 2206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441663604622354e-06,
+      "loss": 0.1261,
+      "step": 2207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433862999805487e-06,
+      "loss": 0.1309,
+      "step": 2208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426062740572214e-06,
+      "loss": 0.1284,
+      "step": 2209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418262831683999e-06,
+      "loss": 0.1238,
+      "step": 2210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410463277902097e-06,
+      "loss": 0.1237,
+      "step": 2211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402664083987539e-06,
+      "loss": 0.1295,
+      "step": 2212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.39486525470114e-06,
+      "loss": 0.1263,
+      "step": 2213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387066794803494e-06,
+      "loss": 0.1331,
+      "step": 2214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379268709054966e-06,
+      "loss": 0.1268,
+      "step": 2215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.371471002215695e-06,
+      "loss": 0.1341,
+      "step": 2216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.363673679045588e-06,
+      "loss": 0.1256,
+      "step": 2217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355876744304318e-06,
+      "loss": 0.127,
+      "step": 2218
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.348080202751322e-06,
+      "loss": 0.1242,
+      "step": 2219
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340284059145794e-06,
+      "loss": 0.1182,
+      "step": 2220
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33248831824669e-06,
+      "loss": 0.1292,
+      "step": 2221
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324692984812713e-06,
+      "loss": 0.1333,
+      "step": 2222
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316898063602325e-06,
+      "loss": 0.1392,
+      "step": 2223
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309103559373725e-06,
+      "loss": 0.1288,
+      "step": 2224
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301309476884875e-06,
+      "loss": 0.129,
+      "step": 2225
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293515820893468e-06,
+      "loss": 0.1225,
+      "step": 2226
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285722596156935e-06,
+      "loss": 0.1184,
+      "step": 2227
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.277929807432448e-06,
+      "loss": 0.1249,
+      "step": 2228
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270137459476914e-06,
+      "loss": 0.119,
+      "step": 2229
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262345557046968e-06,
+      "loss": 0.1216,
+      "step": 2230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.254554104898973e-06,
+      "loss": 0.1288,
+      "step": 2231
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246763107789018e-06,
+      "loss": 0.1247,
+      "step": 2232
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238972570472915e-06,
+      "loss": 0.122,
+      "step": 2233
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231182497706195e-06,
+      "loss": 0.1328,
+      "step": 2234
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223392894244106e-06,
+      "loss": 0.1234,
+      "step": 2235
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215603764841605e-06,
+      "loss": 0.1177,
+      "step": 2236
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.207815114253367e-06,
+      "loss": 0.1312,
+      "step": 2237
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20002694723377e-06,
+      "loss": 0.126,
+      "step": 2238
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192239268536898e-06,
+      "loss": 0.1298,
+      "step": 2239
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184452082916535e-06,
+      "loss": 0.1322,
+      "step": 2240
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.176665395126168e-06,
+      "loss": 0.1288,
+      "step": 2241
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168879209918976e-06,
+      "loss": 0.14,
+      "step": 2242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.161093532047832e-06,
+      "loss": 0.1288,
+      "step": 2243
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.153308366265304e-06,
+      "loss": 0.1293,
+      "step": 2244
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145523717323643e-06,
+      "loss": 0.1286,
+      "step": 2245
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.137739589974784e-06,
+      "loss": 0.1278,
+      "step": 2246
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129955988970346e-06,
+      "loss": 0.1222,
+      "step": 2247
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122172919061624e-06,
+      "loss": 0.1272,
+      "step": 2248
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114390384999591e-06,
+      "loss": 0.1255,
+      "step": 2249
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106608391534894e-06,
+      "loss": 0.1278,
+      "step": 2250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098826943417846e-06,
+      "loss": 0.1242,
+      "step": 2251
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091046045398429e-06,
+      "loss": 0.1279,
+      "step": 2252
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08326570222629e-06,
+      "loss": 0.1372,
+      "step": 2253
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075485918650739e-06,
+      "loss": 0.1349,
+      "step": 2254
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067706699420744e-06,
+      "loss": 0.1234,
+      "step": 2255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.059928049284921e-06,
+      "loss": 0.133,
+      "step": 2256
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05214997299155e-06,
+      "loss": 0.137,
+      "step": 2257
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044372475288551e-06,
+      "loss": 0.1309,
+      "step": 2258
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.036595560923499e-06,
+      "loss": 0.1241,
+      "step": 2259
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.028819234643606e-06,
+      "loss": 0.1259,
+      "step": 2260
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.021043501195729e-06,
+      "loss": 0.1232,
+      "step": 2261
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01326836532636e-06,
+      "loss": 0.1262,
+      "step": 2262
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005493831781632e-06,
+      "loss": 0.1182,
+      "step": 2263
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997719905307303e-06,
+      "loss": 0.1363,
+      "step": 2264
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989946590648767e-06,
+      "loss": 0.1278,
+      "step": 2265
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98217389255104e-06,
+      "loss": 0.1241,
+      "step": 2266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974401815758762e-06,
+      "loss": 0.127,
+      "step": 2267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.9666303650162e-06,
+      "loss": 0.1269,
+      "step": 2268
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.958859545067227e-06,
+      "loss": 0.1257,
+      "step": 2269
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951089360655341e-06,
+      "loss": 0.1244,
+      "step": 2270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94331981652365e-06,
+      "loss": 0.1258,
+      "step": 2271
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935550917414868e-06,
+      "loss": 0.127,
+      "step": 2272
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.927782668071318e-06,
+      "loss": 0.1223,
+      "step": 2273
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.920015073234926e-06,
+      "loss": 0.1241,
+      "step": 2274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.912248137647217e-06,
+      "loss": 0.1279,
+      "step": 2275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904481866049312e-06,
+      "loss": 0.1242,
+      "step": 2276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896716263181937e-06,
+      "loss": 0.1291,
+      "step": 2277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888951333785396e-06,
+      "loss": 0.1234,
+      "step": 2278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88118708259959e-06,
+      "loss": 0.1233,
+      "step": 2279
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.873423514364003e-06,
+      "loss": 0.1219,
+      "step": 2280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865660633817705e-06,
+      "loss": 0.1111,
+      "step": 2281
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.857898445699341e-06,
+      "loss": 0.1306,
+      "step": 2282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85013695474714e-06,
+      "loss": 0.1218,
+      "step": 2283
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.8423761656989e-06,
+      "loss": 0.1216,
+      "step": 2284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834616083291994e-06,
+      "loss": 0.1328,
+      "step": 2285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826856712263363e-06,
+      "loss": 0.1237,
+      "step": 2286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819098057349509e-06,
+      "loss": 0.126,
+      "step": 2287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.811340123286505e-06,
+      "loss": 0.1245,
+      "step": 2288
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80358291480998e-06,
+      "loss": 0.1241,
+      "step": 2289
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.795826436655118e-06,
+      "loss": 0.1236,
+      "step": 2290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78807069355666e-06,
+      "loss": 0.1295,
+      "step": 2291
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780315690248898e-06,
+      "loss": 0.1232,
+      "step": 2292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.772561431465669e-06,
+      "loss": 0.1436,
+      "step": 2293
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764807921940362e-06,
+      "loss": 0.1344,
+      "step": 2294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.757055166405904e-06,
+      "loss": 0.1283,
+      "step": 2295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74930316959476e-06,
+      "loss": 0.1222,
+      "step": 2296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.741551936238939e-06,
+      "loss": 0.1244,
+      "step": 2297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733801471069978e-06,
+      "loss": 0.1255,
+      "step": 2298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.726051778818945e-06,
+      "loss": 0.1317,
+      "step": 2299
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.718302864216438e-06,
+      "loss": 0.1299,
+      "step": 2300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710554731992585e-06,
+      "loss": 0.1187,
+      "step": 2301
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702807386877025e-06,
+      "loss": 0.1186,
+      "step": 2302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695060833598928e-06,
+      "loss": 0.12,
+      "step": 2303
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687315076886975e-06,
+      "loss": 0.1261,
+      "step": 2304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679570121469358e-06,
+      "loss": 0.12,
+      "step": 2305
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671825972073786e-06,
+      "loss": 0.1151,
+      "step": 2306
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.664082633427474e-06,
+      "loss": 0.1096,
+      "step": 2307
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65634011025714e-06,
+      "loss": 0.1235,
+      "step": 2308
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648598407289008e-06,
+      "loss": 0.1158,
+      "step": 2309
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640857529248794e-06,
+      "loss": 0.1218,
+      "step": 2310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.633117480861722e-06,
+      "loss": 0.1219,
+      "step": 2311
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625378266852497e-06,
+      "loss": 0.1261,
+      "step": 2312
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.617639891945328e-06,
+      "loss": 0.1272,
+      "step": 2313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609902360863899e-06,
+      "loss": 0.1243,
+      "step": 2314
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.602165678331385e-06,
+      "loss": 0.1264,
+      "step": 2315
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594429849070444e-06,
+      "loss": 0.1194,
+      "step": 2316
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58669487780321e-06,
+      "loss": 0.1386,
+      "step": 2317
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.578960769251297e-06,
+      "loss": 0.1188,
+      "step": 2318
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571227528135789e-06,
+      "loss": 0.1276,
+      "step": 2319
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563495159177237e-06,
+      "loss": 0.1255,
+      "step": 2320
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.555763667095672e-06,
+      "loss": 0.1272,
+      "step": 2321
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54803305661058e-06,
+      "loss": 0.1231,
+      "step": 2322
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54030333244091e-06,
+      "loss": 0.1242,
+      "step": 2323
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53257449930507e-06,
+      "loss": 0.1208,
+      "step": 2324
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.524846561920925e-06,
+      "loss": 0.1285,
+      "step": 2325
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517119525005793e-06,
+      "loss": 0.1229,
+      "step": 2326
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.509393393276446e-06,
+      "loss": 0.128,
+      "step": 2327
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501668171449095e-06,
+      "loss": 0.1184,
+      "step": 2328
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.493943864239406e-06,
+      "loss": 0.1345,
+      "step": 2329
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486220476362476e-06,
+      "loss": 0.1308,
+      "step": 2330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47849801253285e-06,
+      "loss": 0.1256,
+      "step": 2331
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.470776477464501e-06,
+      "loss": 0.1313,
+      "step": 2332
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463055875870844e-06,
+      "loss": 0.1209,
+      "step": 2333
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455336212464715e-06,
+      "loss": 0.129,
+      "step": 2334
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447617491958382e-06,
+      "loss": 0.1204,
+      "step": 2335
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43989971906354e-06,
+      "loss": 0.1385,
+      "step": 2336
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432182898491301e-06,
+      "loss": 0.1308,
+      "step": 2337
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424467034952198e-06,
+      "loss": 0.1158,
+      "step": 2338
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.416752133156177e-06,
+      "loss": 0.1323,
+      "step": 2339
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.409038197812598e-06,
+      "loss": 0.1256,
+      "step": 2340
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401325233630232e-06,
+      "loss": 0.1258,
+      "step": 2341
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393613245317263e-06,
+      "loss": 0.1277,
+      "step": 2342
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.385902237581269e-06,
+      "loss": 0.1277,
+      "step": 2343
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378192215129235e-06,
+      "loss": 0.1254,
+      "step": 2344
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370483182667543e-06,
+      "loss": 0.1301,
+      "step": 2345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.362775144901971e-06,
+      "loss": 0.1273,
+      "step": 2346
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355068106537691e-06,
+      "loss": 0.1285,
+      "step": 2347
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347362072279263e-06,
+      "loss": 0.119,
+      "step": 2348
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339657046830635e-06,
+      "loss": 0.1306,
+      "step": 2349
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331953034895139e-06,
+      "loss": 0.1289,
+      "step": 2350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32425004117549e-06,
+      "loss": 0.1263,
+      "step": 2351
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31654807037378e-06,
+      "loss": 0.1282,
+      "step": 2352
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308847127191473e-06,
+      "loss": 0.117,
+      "step": 2353
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.30114721632941e-06,
+      "loss": 0.1261,
+      "step": 2354
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293448342487807e-06,
+      "loss": 0.1258,
+      "step": 2355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285750510366236e-06,
+      "loss": 0.1254,
+      "step": 2356
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.278053724663638e-06,
+      "loss": 0.1259,
+      "step": 2357
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270357990078318e-06,
+      "loss": 0.1177,
+      "step": 2358
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.262663311307936e-06,
+      "loss": 0.1235,
+      "step": 2359
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254969693049506e-06,
+      "loss": 0.118,
+      "step": 2360
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247277139999401e-06,
+      "loss": 0.1159,
+      "step": 2361
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239585656853338e-06,
+      "loss": 0.1245,
+      "step": 2362
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231895248306382e-06,
+      "loss": 0.1234,
+      "step": 2363
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.22420591905294e-06,
+      "loss": 0.1249,
+      "step": 2364
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.216517673786767e-06,
+      "loss": 0.1333,
+      "step": 2365
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208830517200952e-06,
+      "loss": 0.1185,
+      "step": 2366
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.20114445398792e-06,
+      "loss": 0.1247,
+      "step": 2367
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.193459488839426e-06,
+      "loss": 0.1285,
+      "step": 2368
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185775626446557e-06,
+      "loss": 0.1212,
+      "step": 2369
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178092871499728e-06,
+      "loss": 0.1288,
+      "step": 2370
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.170411228688671e-06,
+      "loss": 0.1163,
+      "step": 2371
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162730702702456e-06,
+      "loss": 0.1275,
+      "step": 2372
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.155051298229454e-06,
+      "loss": 0.1346,
+      "step": 2373
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147373019957357e-06,
+      "loss": 0.1286,
+      "step": 2374
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139695872573172e-06,
+      "loss": 0.1337,
+      "step": 2375
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132019860763212e-06,
+      "loss": 0.1261,
+      "step": 2376
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.124344989213099e-06,
+      "loss": 0.1174,
+      "step": 2377
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11667126260776e-06,
+      "loss": 0.1198,
+      "step": 2378
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10899868563142e-06,
+      "loss": 0.1332,
+      "step": 2379
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101327262967603e-06,
+      "loss": 0.1177,
+      "step": 2380
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.093656999299132e-06,
+      "loss": 0.1277,
+      "step": 2381
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085987899308117e-06,
+      "loss": 0.1262,
+      "step": 2382
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078319967675964e-06,
+      "loss": 0.122,
+      "step": 2383
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07065320908336e-06,
+      "loss": 0.1158,
+      "step": 2384
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.062987628210276e-06,
+      "loss": 0.1223,
+      "step": 2385
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05532322973597e-06,
+      "loss": 0.1239,
+      "step": 2386
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047660018338971e-06,
+      "loss": 0.12,
+      "step": 2387
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03999799869709e-06,
+      "loss": 0.1238,
+      "step": 2388
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032337175487407e-06,
+      "loss": 0.1296,
+      "step": 2389
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024677553386272e-06,
+      "loss": 0.1286,
+      "step": 2390
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0170191370693e-06,
+      "loss": 0.1323,
+      "step": 2391
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.009361931211375e-06,
+      "loss": 0.1234,
+      "step": 2392
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001705940486636e-06,
+      "loss": 0.1157,
+      "step": 2393
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.994051169568486e-06,
+      "loss": 0.1255,
+      "step": 2394
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.986397623129578e-06,
+      "loss": 0.1256,
+      "step": 2395
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978745305841822e-06,
+      "loss": 0.133,
+      "step": 2396
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971094222376375e-06,
+      "loss": 0.1262,
+      "step": 2397
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96344437740364e-06,
+      "loss": 0.1313,
+      "step": 2398
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.955795775593271e-06,
+      "loss": 0.1165,
+      "step": 2399
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948148421614148e-06,
+      "loss": 0.118,
+      "step": 2400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94050232013441e-06,
+      "loss": 0.1242,
+      "step": 2401
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.932857475821416e-06,
+      "loss": 0.1238,
+      "step": 2402
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925213893341764e-06,
+      "loss": 0.1259,
+      "step": 2403
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917571577361276e-06,
+      "loss": 0.1201,
+      "step": 2404
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909930532545009e-06,
+      "loss": 0.1342,
+      "step": 2405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902290763557237e-06,
+      "loss": 0.1328,
+      "step": 2406
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89465227506146e-06,
+      "loss": 0.1239,
+      "step": 2407
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887015071720394e-06,
+      "loss": 0.1304,
+      "step": 2408
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87937915819597e-06,
+      "loss": 0.1263,
+      "step": 2409
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871744539149332e-06,
+      "loss": 0.1292,
+      "step": 2410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86411121924084e-06,
+      "loss": 0.1318,
+      "step": 2411
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.856479203130047e-06,
+      "loss": 0.1294,
+      "step": 2412
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848848495475724e-06,
+      "loss": 0.119,
+      "step": 2413
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841219100935835e-06,
+      "loss": 0.132,
+      "step": 2414
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833591024167549e-06,
+      "loss": 0.124,
+      "step": 2415
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825964269827223e-06,
+      "loss": 0.1247,
+      "step": 2416
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818338842570413e-06,
+      "loss": 0.1285,
+      "step": 2417
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810714747051861e-06,
+      "loss": 0.1282,
+      "step": 2418
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803091987925501e-06,
+      "loss": 0.1256,
+      "step": 2419
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795470569844444e-06,
+      "loss": 0.1256,
+      "step": 2420
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.787850497460987e-06,
+      "loss": 0.1295,
+      "step": 2421
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780231775426603e-06,
+      "loss": 0.1229,
+      "step": 2422
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772614408391947e-06,
+      "loss": 0.1282,
+      "step": 2423
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764998401006841e-06,
+      "loss": 0.1129,
+      "step": 2424
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757383757920278e-06,
+      "loss": 0.1352,
+      "step": 2425
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.74977048378042e-06,
+      "loss": 0.1161,
+      "step": 2426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742158583234587e-06,
+      "loss": 0.1251,
+      "step": 2427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.734548060929272e-06,
+      "loss": 0.1363,
+      "step": 2428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726938921510116e-06,
+      "loss": 0.1244,
+      "step": 2429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719331169621918e-06,
+      "loss": 0.1234,
+      "step": 2430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711724809908638e-06,
+      "loss": 0.1194,
+      "step": 2431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.704119847013377e-06,
+      "loss": 0.1187,
+      "step": 2432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696516285578388e-06,
+      "loss": 0.1219,
+      "step": 2433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688914130245067e-06,
+      "loss": 0.1173,
+      "step": 2434
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681313385653951e-06,
+      "loss": 0.1186,
+      "step": 2435
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673714056444716e-06,
+      "loss": 0.1296,
+      "step": 2436
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666116147256178e-06,
+      "loss": 0.1348,
+      "step": 2437
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658519662726277e-06,
+      "loss": 0.1265,
+      "step": 2438
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.650924607492096e-06,
+      "loss": 0.1273,
+      "step": 2439
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.643330986189834e-06,
+      "loss": 0.1248,
+      "step": 2440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.635738803454821e-06,
+      "loss": 0.1235,
+      "step": 2441
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628148063921507e-06,
+      "loss": 0.1254,
+      "step": 2442
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620558772223461e-06,
+      "loss": 0.1273,
+      "step": 2443
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6129709329933695e-06,
+      "loss": 0.1216,
+      "step": 2444
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605384550863031e-06,
+      "loss": 0.1253,
+      "step": 2445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597799630463355e-06,
+      "loss": 0.1262,
+      "step": 2446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5902161764243584e-06,
+      "loss": 0.1293,
+      "step": 2447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5826341933751635e-06,
+      "loss": 0.1328,
+      "step": 2448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575053685943995e-06,
+      "loss": 0.1296,
+      "step": 2449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.567474658758176e-06,
+      "loss": 0.1216,
+      "step": 2450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559897116444125e-06,
+      "loss": 0.1264,
+      "step": 2451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.552321063627357e-06,
+      "loss": 0.1268,
+      "step": 2452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.544746504932477e-06,
+      "loss": 0.1252,
+      "step": 2453
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.537173444983174e-06,
+      "loss": 0.1287,
+      "step": 2454
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52960188840223e-06,
+      "loss": 0.1208,
+      "step": 2455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522031839811497e-06,
+      "loss": 0.1249,
+      "step": 2456
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514463303831923e-06,
+      "loss": 0.1164,
+      "step": 2457
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506896285083518e-06,
+      "loss": 0.1221,
+      "step": 2458
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.499330788185373e-06,
+      "loss": 0.1304,
+      "step": 2459
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.49176681775565e-06,
+      "loss": 0.1224,
+      "step": 2460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484204378411577e-06,
+      "loss": 0.1248,
+      "step": 2461
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.476643474769449e-06,
+      "loss": 0.1298,
+      "step": 2462
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46908411144462e-06,
+      "loss": 0.1269,
+      "step": 2463
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461526293051511e-06,
+      "loss": 0.1313,
+      "step": 2464
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453970024203591e-06,
+      "loss": 0.1341,
+      "step": 2465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4464153095133874e-06,
+      "loss": 0.122,
+      "step": 2466
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.438862153592483e-06,
+      "loss": 0.134,
+      "step": 2467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431310561051505e-06,
+      "loss": 0.1195,
+      "step": 2468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423760536500124e-06,
+      "loss": 0.132,
+      "step": 2469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416212084547057e-06,
+      "loss": 0.128,
+      "step": 2470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408665209800059e-06,
+      "loss": 0.1335,
+      "step": 2471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.401119916865924e-06,
+      "loss": 0.1206,
+      "step": 2472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393576210350482e-06,
+      "loss": 0.1432,
+      "step": 2473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386034094858588e-06,
+      "loss": 0.119,
+      "step": 2474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3784935749941324e-06,
+      "loss": 0.1295,
+      "step": 2475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3709546553600286e-06,
+      "loss": 0.1271,
+      "step": 2476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3634173405582145e-06,
+      "loss": 0.121,
+      "step": 2477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355881635189647e-06,
+      "loss": 0.1273,
+      "step": 2478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.348347543854302e-06,
+      "loss": 0.1197,
+      "step": 2479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3408150711511685e-06,
+      "loss": 0.1198,
+      "step": 2480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.333284221678248e-06,
+      "loss": 0.1313,
+      "step": 2481
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.325755000032553e-06,
+      "loss": 0.1363,
+      "step": 2482
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.318227410810101e-06,
+      "loss": 0.1237,
+      "step": 2483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.310701458605911e-06,
+      "loss": 0.1315,
+      "step": 2484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.303177148014007e-06,
+      "loss": 0.1361,
+      "step": 2485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295654483627405e-06,
+      "loss": 0.1187,
+      "step": 2486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288133470038124e-06,
+      "loss": 0.1217,
+      "step": 2487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280614111837165e-06,
+      "loss": 0.1321,
+      "step": 2488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.273096413614531e-06,
+      "loss": 0.132,
+      "step": 2489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.265580379959203e-06,
+      "loss": 0.1261,
+      "step": 2490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.258066015459146e-06,
+      "loss": 0.1277,
+      "step": 2491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250553324701313e-06,
+      "loss": 0.1269,
+      "step": 2492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243042312271624e-06,
+      "loss": 0.1261,
+      "step": 2493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.235532982754986e-06,
+      "loss": 0.1394,
+      "step": 2494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.228025340735272e-06,
+      "loss": 0.1248,
+      "step": 2495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220519390795325e-06,
+      "loss": 0.1257,
+      "step": 2496
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.213015137516959e-06,
+      "loss": 0.1278,
+      "step": 2497
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20551258548095e-06,
+      "loss": 0.1241,
+      "step": 2498
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.198011739267035e-06,
+      "loss": 0.1256,
+      "step": 2499
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190512603453907e-06,
+      "loss": 0.122,
+      "step": 2500
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 0.12310730665922165,
+      "eval_runtime": 717.8764,
+      "eval_samples_per_second": 69.65,
+      "eval_steps_per_second": 2.177,
+      "step": 2500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183015182619224e-06,
+      "loss": 0.1294,
+      "step": 2501
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17551948133959e-06,
+      "loss": 0.1244,
+      "step": 2502
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168025504190558e-06,
+      "loss": 0.138,
+      "step": 2503
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160533255746634e-06,
+      "loss": 0.1154,
+      "step": 2504
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153042740581264e-06,
+      "loss": 0.1346,
+      "step": 2505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145553963266838e-06,
+      "loss": 0.1283,
+      "step": 2506
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.138066928374687e-06,
+      "loss": 0.1306,
+      "step": 2507
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130581640475076e-06,
+      "loss": 0.1188,
+      "step": 2508
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.123098104137203e-06,
+      "loss": 0.1344,
+      "step": 2509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1156163239291995e-06,
+      "loss": 0.1354,
+      "step": 2510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108136304418121e-06,
+      "loss": 0.1285,
+      "step": 2511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.100658050169953e-06,
+      "loss": 0.1306,
+      "step": 2512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0931815657496005e-06,
+      "loss": 0.1217,
+      "step": 2513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.085706855720888e-06,
+      "loss": 0.1177,
+      "step": 2514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078233924646559e-06,
+      "loss": 0.1314,
+      "step": 2515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.070762777088269e-06,
+      "loss": 0.1211,
+      "step": 2516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063293417606585e-06,
+      "loss": 0.1259,
+      "step": 2517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05582585076098e-06,
+      "loss": 0.1245,
+      "step": 2518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048360081109844e-06,
+      "loss": 0.1483,
+      "step": 2519
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040896113210456e-06,
+      "loss": 0.1284,
+      "step": 2520
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033433951619002e-06,
+      "loss": 0.1282,
+      "step": 2521
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0259736008905655e-06,
+      "loss": 0.1247,
+      "step": 2522
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.018515065579121e-06,
+      "loss": 0.1328,
+      "step": 2523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.011058350237538e-06,
+      "loss": 0.1241,
+      "step": 2524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003603459417576e-06,
+      "loss": 0.1213,
+      "step": 2525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.996150397669876e-06,
+      "loss": 0.1239,
+      "step": 2526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988699169543969e-06,
+      "loss": 0.1252,
+      "step": 2527
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.981249779588261e-06,
+      "loss": 0.1209,
+      "step": 2528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973802232350038e-06,
+      "loss": 0.1251,
+      "step": 2529
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966356532375462e-06,
+      "loss": 0.1285,
+      "step": 2530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958912684209566e-06,
+      "loss": 0.1317,
+      "step": 2531
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951470692396256e-06,
+      "loss": 0.1319,
+      "step": 2532
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9440305614782985e-06,
+      "loss": 0.1165,
+      "step": 2533
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936592295997328e-06,
+      "loss": 0.1219,
+      "step": 2534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.929155900493843e-06,
+      "loss": 0.1253,
+      "step": 2535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921721379507197e-06,
+      "loss": 0.1243,
+      "step": 2536
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914288737575597e-06,
+      "loss": 0.1317,
+      "step": 2537
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.906857979236109e-06,
+      "loss": 0.1251,
+      "step": 2538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.899429109024644e-06,
+      "loss": 0.1231,
+      "step": 2539
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8920021314759635e-06,
+      "loss": 0.1282,
+      "step": 2540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884577051123671e-06,
+      "loss": 0.1274,
+      "step": 2541
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.877153872500215e-06,
+      "loss": 0.1227,
+      "step": 2542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.869732600136881e-06,
+      "loss": 0.1357,
+      "step": 2543
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862313238563791e-06,
+      "loss": 0.1344,
+      "step": 2544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854895792309903e-06,
+      "loss": 0.1332,
+      "step": 2545
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.847480265903002e-06,
+      "loss": 0.1177,
+      "step": 2546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.840066663869701e-06,
+      "loss": 0.1196,
+      "step": 2547
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832654990735446e-06,
+      "loss": 0.1283,
+      "step": 2548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825245251024497e-06,
+      "loss": 0.1364,
+      "step": 2549
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.817837449259938e-06,
+      "loss": 0.1233,
+      "step": 2550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.810431589963668e-06,
+      "loss": 0.1376,
+      "step": 2551
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803027677656398e-06,
+      "loss": 0.1195,
+      "step": 2552
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.795625716857659e-06,
+      "loss": 0.1303,
+      "step": 2553
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7882257120857805e-06,
+      "loss": 0.1185,
+      "step": 2554
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7808276678579055e-06,
+      "loss": 0.119,
+      "step": 2555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773431588689976e-06,
+      "loss": 0.1215,
+      "step": 2556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7660374790967365e-06,
+      "loss": 0.1393,
+      "step": 2557
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75864534359173e-06,
+      "loss": 0.1244,
+      "step": 2558
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751255186687291e-06,
+      "loss": 0.1313,
+      "step": 2559
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743867012894551e-06,
+      "loss": 0.136,
+      "step": 2560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.736480826723425e-06,
+      "loss": 0.1299,
+      "step": 2561
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7290966326826215e-06,
+      "loss": 0.1237,
+      "step": 2562
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.721714435279626e-06,
+      "loss": 0.1204,
+      "step": 2563
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.714334239020712e-06,
+      "loss": 0.1265,
+      "step": 2564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.706956048410926e-06,
+      "loss": 0.125,
+      "step": 2565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699579867954094e-06,
+      "loss": 0.1317,
+      "step": 2566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692205702152812e-06,
+      "loss": 0.1394,
+      "step": 2567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684833555508446e-06,
+      "loss": 0.1212,
+      "step": 2568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677463432521134e-06,
+      "loss": 0.1256,
+      "step": 2569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.670095337689776e-06,
+      "loss": 0.1181,
+      "step": 2570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6627292755120324e-06,
+      "loss": 0.1215,
+      "step": 2571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.655365250484325e-06,
+      "loss": 0.1295,
+      "step": 2572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6480032671018276e-06,
+      "loss": 0.1269,
+      "step": 2573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.640643329858476e-06,
+      "loss": 0.1341,
+      "step": 2574
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633285443246951e-06,
+      "loss": 0.1211,
+      "step": 2575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.625929611758679e-06,
+      "loss": 0.1183,
+      "step": 2576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6185758398838405e-06,
+      "loss": 0.1335,
+      "step": 2577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.611224132111351e-06,
+      "loss": 0.1322,
+      "step": 2578
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.603874492928875e-06,
+      "loss": 0.1305,
+      "step": 2579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.596526926822802e-06,
+      "loss": 0.1332,
+      "step": 2580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589181438278265e-06,
+      "loss": 0.1188,
+      "step": 2581
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.581838031779126e-06,
+      "loss": 0.1213,
+      "step": 2582
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574496711807976e-06,
+      "loss": 0.1373,
+      "step": 2583
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5671574828461335e-06,
+      "loss": 0.1229,
+      "step": 2584
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55982034937364e-06,
+      "loss": 0.1299,
+      "step": 2585
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552485315869256e-06,
+      "loss": 0.1273,
+      "step": 2586
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545152386810463e-06,
+      "loss": 0.1259,
+      "step": 2587
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537821566673455e-06,
+      "loss": 0.1319,
+      "step": 2588
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5304928599331415e-06,
+      "loss": 0.1286,
+      "step": 2589
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523166271063142e-06,
+      "loss": 0.1365,
+      "step": 2590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.51584180453578e-06,
+      "loss": 0.129,
+      "step": 2591
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.508519464822085e-06,
+      "loss": 0.1332,
+      "step": 2592
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501199256391791e-06,
+      "loss": 0.1231,
+      "step": 2593
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493881183713328e-06,
+      "loss": 0.1302,
+      "step": 2594
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.486565251253821e-06,
+      "loss": 0.1223,
+      "step": 2595
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479251463479093e-06,
+      "loss": 0.1317,
+      "step": 2596
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4719398248536545e-06,
+      "loss": 0.1292,
+      "step": 2597
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464630339840703e-06,
+      "loss": 0.1265,
+      "step": 2598
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4573230129021265e-06,
+      "loss": 0.1231,
+      "step": 2599
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450017848498491e-06,
+      "loss": 0.115,
+      "step": 2600
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442714851089042e-06,
+      "loss": 0.1364,
+      "step": 2601
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4354140251317075e-06,
+      "loss": 0.1332,
+      "step": 2602
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428115375083086e-06,
+      "loss": 0.1316,
+      "step": 2603
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.420818905398447e-06,
+      "loss": 0.1169,
+      "step": 2604
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413524620531731e-06,
+      "loss": 0.1257,
+      "step": 2605
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.406232524935543e-06,
+      "loss": 0.1208,
+      "step": 2606
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3989426230611574e-06,
+      "loss": 0.1269,
+      "step": 2607
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3916549193585005e-06,
+      "loss": 0.1339,
+      "step": 2608
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.384369418276164e-06,
+      "loss": 0.1265,
+      "step": 2609
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377086124261389e-06,
+      "loss": 0.1286,
+      "step": 2610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369805041760078e-06,
+      "loss": 0.1223,
+      "step": 2611
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362526175216771e-06,
+      "loss": 0.1218,
+      "step": 2612
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3552495290746695e-06,
+      "loss": 0.1191,
+      "step": 2613
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34797510777561e-06,
+      "loss": 0.1229,
+      "step": 2614
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340702915760073e-06,
+      "loss": 0.1166,
+      "step": 2615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33343295746718e-06,
+      "loss": 0.1221,
+      "step": 2616
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326165237334687e-06,
+      "loss": 0.1281,
+      "step": 2617
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.318899759798986e-06,
+      "loss": 0.127,
+      "step": 2618
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.311636529295096e-06,
+      "loss": 0.1225,
+      "step": 2619
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304375550256669e-06,
+      "loss": 0.1375,
+      "step": 2620
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.297116827115979e-06,
+      "loss": 0.1274,
+      "step": 2621
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289860364303927e-06,
+      "loss": 0.1258,
+      "step": 2622
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.282606166250029e-06,
+      "loss": 0.1207,
+      "step": 2623
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2753542373824235e-06,
+      "loss": 0.1245,
+      "step": 2624
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2681045821278605e-06,
+      "loss": 0.1298,
+      "step": 2625
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260857204911705e-06,
+      "loss": 0.1244,
+      "step": 2626
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.253612110157927e-06,
+      "loss": 0.1376,
+      "step": 2627
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246369302289108e-06,
+      "loss": 0.1212,
+      "step": 2628
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239128785726431e-06,
+      "loss": 0.1278,
+      "step": 2629
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.23189056488968e-06,
+      "loss": 0.1202,
+      "step": 2630
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.224654644197241e-06,
+      "loss": 0.1308,
+      "step": 2631
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217421028066091e-06,
+      "loss": 0.124,
+      "step": 2632
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.210189720911803e-06,
+      "loss": 0.1248,
+      "step": 2633
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.20296072714854e-06,
+      "loss": 0.1197,
+      "step": 2634
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.195734051189051e-06,
+      "loss": 0.119,
+      "step": 2635
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.188509697444678e-06,
+      "loss": 0.1326,
+      "step": 2636
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181287670325337e-06,
+      "loss": 0.1252,
+      "step": 2637
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174067974239526e-06,
+      "loss": 0.12,
+      "step": 2638
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1668506135943195e-06,
+      "loss": 0.1281,
+      "step": 2639
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.15963559279537e-06,
+      "loss": 0.1185,
+      "step": 2640
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152422916246896e-06,
+      "loss": 0.1154,
+      "step": 2641
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14521258835169e-06,
+      "loss": 0.1244,
+      "step": 2642
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.138004613511109e-06,
+      "loss": 0.1259,
+      "step": 2643
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130798996125074e-06,
+      "loss": 0.1185,
+      "step": 2644
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123595740592063e-06,
+      "loss": 0.1374,
+      "step": 2645
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.11639485130912e-06,
+      "loss": 0.118,
+      "step": 2646
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.109196332671839e-06,
+      "loss": 0.1256,
+      "step": 2647
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.102000189074369e-06,
+      "loss": 0.1308,
+      "step": 2648
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094806424909408e-06,
+      "loss": 0.1203,
+      "step": 2649
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0876150445682015e-06,
+      "loss": 0.1215,
+      "step": 2650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0804260524405405e-06,
+      "loss": 0.1338,
+      "step": 2651
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073239452914758e-06,
+      "loss": 0.1137,
+      "step": 2652
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.06605525037773e-06,
+      "loss": 0.1217,
+      "step": 2653
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0588734492148595e-06,
+      "loss": 0.1264,
+      "step": 2654
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051694053810094e-06,
+      "loss": 0.1297,
+      "step": 2655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.04451706854591e-06,
+      "loss": 0.1178,
+      "step": 2656
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037342497803307e-06,
+      "loss": 0.1229,
+      "step": 2657
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.030170345961819e-06,
+      "loss": 0.1281,
+      "step": 2658
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023000617399496e-06,
+      "loss": 0.1231,
+      "step": 2659
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.015833316492914e-06,
+      "loss": 0.1245,
+      "step": 2660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.008668447617165e-06,
+      "loss": 0.1139,
+      "step": 2661
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001506015145858e-06,
+      "loss": 0.1332,
+      "step": 2662
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994346023451112e-06,
+      "loss": 0.1322,
+      "step": 2663
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987188476903551e-06,
+      "loss": 0.1264,
+      "step": 2664
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9800333798723275e-06,
+      "loss": 0.1286,
+      "step": 2665
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972880736725074e-06,
+      "loss": 0.1362,
+      "step": 2666
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.965730551827938e-06,
+      "loss": 0.1209,
+      "step": 2667
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.958582829545564e-06,
+      "loss": 0.1279,
+      "step": 2668
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.951437574241092e-06,
+      "loss": 0.1202,
+      "step": 2669
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944294790276157e-06,
+      "loss": 0.1244,
+      "step": 2670
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937154482010887e-06,
+      "loss": 0.1261,
+      "step": 2671
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.930016653803896e-06,
+      "loss": 0.1277,
+      "step": 2672
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.922881310012285e-06,
+      "loss": 0.1167,
+      "step": 2673
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.915748454991641e-06,
+      "loss": 0.1408,
+      "step": 2674
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.908618093096027e-06,
+      "loss": 0.1245,
+      "step": 2675
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901490228677989e-06,
+      "loss": 0.1261,
+      "step": 2676
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.894364866088545e-06,
+      "loss": 0.1331,
+      "step": 2677
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887242009677186e-06,
+      "loss": 0.1255,
+      "step": 2678
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.880121663791876e-06,
+      "loss": 0.1214,
+      "step": 2679
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873003832779045e-06,
+      "loss": 0.1297,
+      "step": 2680
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.865888520983587e-06,
+      "loss": 0.1116,
+      "step": 2681
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8587757327488595e-06,
+      "loss": 0.1161,
+      "step": 2682
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.851665472416678e-06,
+      "loss": 0.127,
+      "step": 2683
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844557744327316e-06,
+      "loss": 0.1261,
+      "step": 2684
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.837452552819505e-06,
+      "loss": 0.1233,
+      "step": 2685
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.830349902230414e-06,
+      "loss": 0.1236,
+      "step": 2686
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8232497968956825e-06,
+      "loss": 0.1191,
+      "step": 2687
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.816152241149378e-06,
+      "loss": 0.1206,
+      "step": 2688
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809057239324022e-06,
+      "loss": 0.1248,
+      "step": 2689
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8019647957505694e-06,
+      "loss": 0.1281,
+      "step": 2690
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.794874914758422e-06,
+      "loss": 0.1261,
+      "step": 2691
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787787600675414e-06,
+      "loss": 0.1165,
+      "step": 2692
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780702857827805e-06,
+      "loss": 0.1245,
+      "step": 2693
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.773620690540298e-06,
+      "loss": 0.1266,
+      "step": 2694
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766541103136019e-06,
+      "loss": 0.117,
+      "step": 2695
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759464099936513e-06,
+      "loss": 0.1296,
+      "step": 2696
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752389685261757e-06,
+      "loss": 0.1201,
+      "step": 2697
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.745317863430135e-06,
+      "loss": 0.1224,
+      "step": 2698
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738248638758468e-06,
+      "loss": 0.1318,
+      "step": 2699
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731182015561969e-06,
+      "loss": 0.122,
+      "step": 2700
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7241179981542805e-06,
+      "loss": 0.1275,
+      "step": 2701
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.717056590847451e-06,
+      "loss": 0.131,
+      "step": 2702
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7099977979519235e-06,
+      "loss": 0.1234,
+      "step": 2703
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702941623776566e-06,
+      "loss": 0.1302,
+      "step": 2704
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695888072628624e-06,
+      "loss": 0.1261,
+      "step": 2705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688837148813764e-06,
+      "loss": 0.1336,
+      "step": 2706
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68178885663603e-06,
+      "loss": 0.1202,
+      "step": 2707
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674743200397879e-06,
+      "loss": 0.1208,
+      "step": 2708
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667700184400139e-06,
+      "loss": 0.1168,
+      "step": 2709
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660659812942043e-06,
+      "loss": 0.1223,
+      "step": 2710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.653622090321196e-06,
+      "loss": 0.1231,
+      "step": 2711
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.646587020833598e-06,
+      "loss": 0.1282,
+      "step": 2712
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.63955460877362e-06,
+      "loss": 0.1125,
+      "step": 2713
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632524858434019e-06,
+      "loss": 0.144,
+      "step": 2714
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.625497774105917e-06,
+      "loss": 0.1194,
+      "step": 2715
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.618473360078822e-06,
+      "loss": 0.1185,
+      "step": 2716
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611451620640598e-06,
+      "loss": 0.1291,
+      "step": 2717
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604432560077489e-06,
+      "loss": 0.1258,
+      "step": 2718
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.597416182674091e-06,
+      "loss": 0.121,
+      "step": 2719
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.590402492713377e-06,
+      "loss": 0.126,
+      "step": 2720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583391494476664e-06,
+      "loss": 0.122,
+      "step": 2721
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576383192243639e-06,
+      "loss": 0.1307,
+      "step": 2722
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.569377590292329e-06,
+      "loss": 0.1179,
+      "step": 2723
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562374692899135e-06,
+      "loss": 0.1156,
+      "step": 2724
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.555374504338782e-06,
+      "loss": 0.1287,
+      "step": 2725
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548377028884359e-06,
+      "loss": 0.1286,
+      "step": 2726
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.54138227080729e-06,
+      "loss": 0.1342,
+      "step": 2727
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.534390234377346e-06,
+      "loss": 0.1152,
+      "step": 2728
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.527400923862627e-06,
+      "loss": 0.1299,
+      "step": 2729
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.520414343529583e-06,
+      "loss": 0.118,
+      "step": 2730
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.513430497642984e-06,
+      "loss": 0.1266,
+      "step": 2731
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5064493904659445e-06,
+      "loss": 0.1176,
+      "step": 2732
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49947102625989e-06,
+      "loss": 0.1304,
+      "step": 2733
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4924954092845905e-06,
+      "loss": 0.132,
+      "step": 2734
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.485522543798122e-06,
+      "loss": 0.1243,
+      "step": 2735
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478552434056892e-06,
+      "loss": 0.1278,
+      "step": 2736
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471585084315627e-06,
+      "loss": 0.1263,
+      "step": 2737
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464620498827355e-06,
+      "loss": 0.1233,
+      "step": 2738
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457658681843436e-06,
+      "loss": 0.1355,
+      "step": 2739
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450699637613522e-06,
+      "loss": 0.1299,
+      "step": 2740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.443743370385584e-06,
+      "loss": 0.1216,
+      "step": 2741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436789884405892e-06,
+      "loss": 0.1198,
+      "step": 2742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.429839183919025e-06,
+      "loss": 0.1331,
+      "step": 2743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.422891273167851e-06,
+      "loss": 0.1216,
+      "step": 2744
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415946156393548e-06,
+      "loss": 0.1174,
+      "step": 2745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409003837835574e-06,
+      "loss": 0.1277,
+      "step": 2746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402064321731695e-06,
+      "loss": 0.1257,
+      "step": 2747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.39512761231795e-06,
+      "loss": 0.1238,
+      "step": 2748
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.38819371382868e-06,
+      "loss": 0.1272,
+      "step": 2749
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381262630496496e-06,
+      "loss": 0.136,
+      "step": 2750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374334366552305e-06,
+      "loss": 0.117,
+      "step": 2751
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367408926225272e-06,
+      "loss": 0.118,
+      "step": 2752
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360486313742869e-06,
+      "loss": 0.1242,
+      "step": 2753
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353566533330812e-06,
+      "loss": 0.1219,
+      "step": 2754
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346649589213108e-06,
+      "loss": 0.1365,
+      "step": 2755
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339735485612021e-06,
+      "loss": 0.1224,
+      "step": 2756
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3328242267480915e-06,
+      "loss": 0.139,
+      "step": 2757
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.325915816840108e-06,
+      "loss": 0.1351,
+      "step": 2758
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3190102601051386e-06,
+      "loss": 0.125,
+      "step": 2759
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312107560758492e-06,
+      "loss": 0.1299,
+      "step": 2760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305207723013752e-06,
+      "loss": 0.1353,
+      "step": 2761
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298310751082733e-06,
+      "loss": 0.125,
+      "step": 2762
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2914166491755234e-06,
+      "loss": 0.1274,
+      "step": 2763
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2845254215004395e-06,
+      "loss": 0.1318,
+      "step": 2764
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2776370722640565e-06,
+      "loss": 0.1253,
+      "step": 2765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.270751605671185e-06,
+      "loss": 0.1282,
+      "step": 2766
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.263869025924884e-06,
+      "loss": 0.1237,
+      "step": 2767
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.256989337226437e-06,
+      "loss": 0.1192,
+      "step": 2768
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250112543775374e-06,
+      "loss": 0.1263,
+      "step": 2769
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2432386497694625e-06,
+      "loss": 0.1253,
+      "step": 2770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.236367659404679e-06,
+      "loss": 0.1232,
+      "step": 2771
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.229499576875253e-06,
+      "loss": 0.1241,
+      "step": 2772
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.222634406373616e-06,
+      "loss": 0.1189,
+      "step": 2773
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.215772152090443e-06,
+      "loss": 0.1282,
+      "step": 2774
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2089128182146076e-06,
+      "loss": 0.1296,
+      "step": 2775
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.202056408933222e-06,
+      "loss": 0.1215,
+      "step": 2776
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.195202928431593e-06,
+      "loss": 0.1212,
+      "step": 2777
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.188352380893258e-06,
+      "loss": 0.1155,
+      "step": 2778
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181504770499948e-06,
+      "loss": 0.1189,
+      "step": 2779
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1746601014316156e-06,
+      "loss": 0.1339,
+      "step": 2780
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167818377866403e-06,
+      "loss": 0.1142,
+      "step": 2781
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.160979603980667e-06,
+      "loss": 0.1275,
+      "step": 2782
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.154143783948959e-06,
+      "loss": 0.1235,
+      "step": 2783
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.147310921944031e-06,
+      "loss": 0.1292,
+      "step": 2784
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.140481022136817e-06,
+      "loss": 0.1137,
+      "step": 2785
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.13365408869646e-06,
+      "loss": 0.1322,
+      "step": 2786
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.126830125790277e-06,
+      "loss": 0.1297,
+      "step": 2787
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120009137583786e-06,
+      "loss": 0.1337,
+      "step": 2788
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113191128240672e-06,
+      "loss": 0.1195,
+      "step": 2789
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106376101922821e-06,
+      "loss": 0.1287,
+      "step": 2790
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099564062790281e-06,
+      "loss": 0.1253,
+      "step": 2791
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.09275501500129e-06,
+      "loss": 0.1236,
+      "step": 2792
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.085948962712246e-06,
+      "loss": 0.1194,
+      "step": 2793
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0791459100777365e-06,
+      "loss": 0.1274,
+      "step": 2794
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.072345861250498e-06,
+      "loss": 0.1176,
+      "step": 2795
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065548820381452e-06,
+      "loss": 0.1297,
+      "step": 2796
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.058754791619669e-06,
+      "loss": 0.1237,
+      "step": 2797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0519637791123895e-06,
+      "loss": 0.1236,
+      "step": 2798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.04517578700501e-06,
+      "loss": 0.1284,
+      "step": 2799
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0383908194410845e-06,
+      "loss": 0.1186,
+      "step": 2800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.031608880562315e-06,
+      "loss": 0.1247,
+      "step": 2801
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.024829974508569e-06,
+      "loss": 0.1313,
+      "step": 2802
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.018054105417842e-06,
+      "loss": 0.1263,
+      "step": 2803
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011281277426294e-06,
+      "loss": 0.1227,
+      "step": 2804
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.004511494668223e-06,
+      "loss": 0.1219,
+      "step": 2805
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9977447612760596e-06,
+      "loss": 0.124,
+      "step": 2806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.990981081380389e-06,
+      "loss": 0.1265,
+      "step": 2807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.984220459109914e-06,
+      "loss": 0.1211,
+      "step": 2808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977462898591488e-06,
+      "loss": 0.1252,
+      "step": 2809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.970708403950083e-06,
+      "loss": 0.1249,
+      "step": 2810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.963956979308805e-06,
+      "loss": 0.1308,
+      "step": 2811
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957208628788891e-06,
+      "loss": 0.1259,
+      "step": 2812
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9504633565096884e-06,
+      "loss": 0.1221,
+      "step": 2813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943721166588675e-06,
+      "loss": 0.1307,
+      "step": 2814
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936982063141452e-06,
+      "loss": 0.1291,
+      "step": 2815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9302460502817196e-06,
+      "loss": 0.1318,
+      "step": 2816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923513132121309e-06,
+      "loss": 0.1308,
+      "step": 2817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.916783312770149e-06,
+      "loss": 0.127,
+      "step": 2818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.910056596336289e-06,
+      "loss": 0.1222,
+      "step": 2819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.903332986925871e-06,
+      "loss": 0.1288,
+      "step": 2820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896612488643154e-06,
+      "loss": 0.1193,
+      "step": 2821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889895105590484e-06,
+      "loss": 0.1182,
+      "step": 2822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.883180841868321e-06,
+      "loss": 0.1295,
+      "step": 2823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.876469701575204e-06,
+      "loss": 0.1226,
+      "step": 2824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869761688807782e-06,
+      "loss": 0.135,
+      "step": 2825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.863056807660779e-06,
+      "loss": 0.1233,
+      "step": 2826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.856355062227021e-06,
+      "loss": 0.1302,
+      "step": 2827
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.849656456597411e-06,
+      "loss": 0.1208,
+      "step": 2828
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842960994860941e-06,
+      "loss": 0.1287,
+      "step": 2829
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836268681104676e-06,
+      "loss": 0.1342,
+      "step": 2830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829579519413773e-06,
+      "loss": 0.123,
+      "step": 2831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8228935138714475e-06,
+      "loss": 0.1311,
+      "step": 2832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.816210668559006e-06,
+      "loss": 0.1233,
+      "step": 2833
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8095309875558105e-06,
+      "loss": 0.1329,
+      "step": 2834
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.802854474939304e-06,
+      "loss": 0.1234,
+      "step": 2835
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796181134784983e-06,
+      "loss": 0.1268,
+      "step": 2836
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789510971166419e-06,
+      "loss": 0.124,
+      "step": 2837
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7828439881552434e-06,
+      "loss": 0.1256,
+      "step": 2838
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.776180189821134e-06,
+      "loss": 0.1285,
+      "step": 2839
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769519580231837e-06,
+      "loss": 0.125,
+      "step": 2840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.762862163453152e-06,
+      "loss": 0.119,
+      "step": 2841
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75620794354892e-06,
+      "loss": 0.1261,
+      "step": 2842
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74955692458104e-06,
+      "loss": 0.1165,
+      "step": 2843
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742909110609448e-06,
+      "loss": 0.1196,
+      "step": 2844
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.736264505692139e-06,
+      "loss": 0.1322,
+      "step": 2845
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7296231138851264e-06,
+      "loss": 0.1228,
+      "step": 2846
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722984939242482e-06,
+      "loss": 0.1213,
+      "step": 2847
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.716349985816311e-06,
+      "loss": 0.1329,
+      "step": 2848
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709718257656738e-06,
+      "loss": 0.1216,
+      "step": 2849
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703089758811938e-06,
+      "loss": 0.133,
+      "step": 2850
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696464493328098e-06,
+      "loss": 0.1259,
+      "step": 2851
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.689842465249446e-06,
+      "loss": 0.1311,
+      "step": 2852
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.68322367861822e-06,
+      "loss": 0.1239,
+      "step": 2853
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.67660813747469e-06,
+      "loss": 0.1202,
+      "step": 2854
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669995845857139e-06,
+      "loss": 0.1198,
+      "step": 2855
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663386807801872e-06,
+      "loss": 0.1272,
+      "step": 2856
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6567810273431965e-06,
+      "loss": 0.1242,
+      "step": 2857
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.650178508513449e-06,
+      "loss": 0.1296,
+      "step": 2858
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643579255342956e-06,
+      "loss": 0.1159,
+      "step": 2859
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636983271860069e-06,
+      "loss": 0.1247,
+      "step": 2860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630390562091124e-06,
+      "loss": 0.123,
+      "step": 2861
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623801130060478e-06,
+      "loss": 0.1316,
+      "step": 2862
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.617214979790472e-06,
+      "loss": 0.1303,
+      "step": 2863
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6106321153014566e-06,
+      "loss": 0.1184,
+      "step": 2864
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.604052540611762e-06,
+      "loss": 0.1312,
+      "step": 2865
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.597476259737727e-06,
+      "loss": 0.1304,
+      "step": 2866
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.590903276693662e-06,
+      "loss": 0.1378,
+      "step": 2867
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5843335954918825e-06,
+      "loss": 0.1279,
+      "step": 2868
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5777672201426705e-06,
+      "loss": 0.1143,
+      "step": 2869
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571204154654303e-06,
+      "loss": 0.1382,
+      "step": 2870
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.564644403033033e-06,
+      "loss": 0.13,
+      "step": 2871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.558087969283092e-06,
+      "loss": 0.1246,
+      "step": 2872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551534857406681e-06,
+      "loss": 0.1196,
+      "step": 2873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54498507140398e-06,
+      "loss": 0.1239,
+      "step": 2874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53843861527313e-06,
+      "loss": 0.1212,
+      "step": 2875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53189549301025e-06,
+      "loss": 0.1182,
+      "step": 2876
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525355708609414e-06,
+      "loss": 0.1195,
+      "step": 2877
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518819266062669e-06,
+      "loss": 0.1121,
+      "step": 2878
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.512286169360007e-06,
+      "loss": 0.1305,
+      "step": 2879
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505756422489398e-06,
+      "loss": 0.1196,
+      "step": 2880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.499230029436747e-06,
+      "loss": 0.1186,
+      "step": 2881
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.492706994185924e-06,
+      "loss": 0.1295,
+      "step": 2882
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.486187320718749e-06,
+      "loss": 0.1253,
+      "step": 2883
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479671013014981e-06,
+      "loss": 0.1225,
+      "step": 2884
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473158075052338e-06,
+      "loss": 0.1214,
+      "step": 2885
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466648510806465e-06,
+      "loss": 0.1317,
+      "step": 2886
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460142324250964e-06,
+      "loss": 0.116,
+      "step": 2887
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.453639519357359e-06,
+      "loss": 0.1137,
+      "step": 2888
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4471401000951255e-06,
+      "loss": 0.1239,
+      "step": 2889
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440644070431659e-06,
+      "loss": 0.1188,
+      "step": 2890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434151434332297e-06,
+      "loss": 0.122,
+      "step": 2891
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4276621957602935e-06,
+      "loss": 0.1329,
+      "step": 2892
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421176358676844e-06,
+      "loss": 0.1164,
+      "step": 2893
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4146939270410506e-06,
+      "loss": 0.1164,
+      "step": 2894
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408214904809951e-06,
+      "loss": 0.1353,
+      "step": 2895
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.401739295938489e-06,
+      "loss": 0.1256,
+      "step": 2896
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.39526710437954e-06,
+      "loss": 0.1157,
+      "step": 2897
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388798334083877e-06,
+      "loss": 0.129,
+      "step": 2898
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382332989000196e-06,
+      "loss": 0.1288,
+      "step": 2899
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375871073075097e-06,
+      "loss": 0.1206,
+      "step": 2900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3694125902530945e-06,
+      "loss": 0.1243,
+      "step": 2901
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362957544476594e-06,
+      "loss": 0.131,
+      "step": 2902
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356505939685915e-06,
+      "loss": 0.1187,
+      "step": 2903
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350057779819264e-06,
+      "loss": 0.1152,
+      "step": 2904
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3436130688127635e-06,
+      "loss": 0.1264,
+      "step": 2905
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337171810600409e-06,
+      "loss": 0.1401,
+      "step": 2906
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330734009114108e-06,
+      "loss": 0.1333,
+      "step": 2907
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.32429966828364e-06,
+      "loss": 0.1212,
+      "step": 2908
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.317868792036689e-06,
+      "loss": 0.1259,
+      "step": 2909
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.311441384298809e-06,
+      "loss": 0.1308,
+      "step": 2910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.30501744899345e-06,
+      "loss": 0.1201,
+      "step": 2911
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29859699004193e-06,
+      "loss": 0.1279,
+      "step": 2912
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292180011363458e-06,
+      "loss": 0.122,
+      "step": 2913
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2857665168751025e-06,
+      "loss": 0.1303,
+      "step": 2914
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279356510491821e-06,
+      "loss": 0.1303,
+      "step": 2915
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2729499961264365e-06,
+      "loss": 0.1252,
+      "step": 2916
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266546977689633e-06,
+      "loss": 0.1242,
+      "step": 2917
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.260147459089971e-06,
+      "loss": 0.1185,
+      "step": 2918
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253751444233867e-06,
+      "loss": 0.1276,
+      "step": 2919
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247358937025606e-06,
+      "loss": 0.1256,
+      "step": 2920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24096994136732e-06,
+      "loss": 0.1254,
+      "step": 2921
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234584461159014e-06,
+      "loss": 0.1296,
+      "step": 2922
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.22820250029853e-06,
+      "loss": 0.1222,
+      "step": 2923
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221824062681578e-06,
+      "loss": 0.1289,
+      "step": 2924
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2154491522017005e-06,
+      "loss": 0.1217,
+      "step": 2925
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209077772750304e-06,
+      "loss": 0.1186,
+      "step": 2926
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202709928216624e-06,
+      "loss": 0.1242,
+      "step": 2927
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196345622487748e-06,
+      "loss": 0.1293,
+      "step": 2928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189984859448603e-06,
+      "loss": 0.1167,
+      "step": 2929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.183627642981952e-06,
+      "loss": 0.1191,
+      "step": 2930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177273976968388e-06,
+      "loss": 0.1241,
+      "step": 2931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170923865286345e-06,
+      "loss": 0.1308,
+      "step": 2932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164577311812078e-06,
+      "loss": 0.1339,
+      "step": 2933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158234320419683e-06,
+      "loss": 0.132,
+      "step": 2934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.151894894981065e-06,
+      "loss": 0.1169,
+      "step": 2935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1455590393659715e-06,
+      "loss": 0.1361,
+      "step": 2936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.13922675744195e-06,
+      "loss": 0.1234,
+      "step": 2937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132898053074387e-06,
+      "loss": 0.1346,
+      "step": 2938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.126572930126466e-06,
+      "loss": 0.1224,
+      "step": 2939
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.120251392459201e-06,
+      "loss": 0.1291,
+      "step": 2940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.113933443931405e-06,
+      "loss": 0.132,
+      "step": 2941
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.10761908839971e-06,
+      "loss": 0.1277,
+      "step": 2942
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101308329718545e-06,
+      "loss": 0.1329,
+      "step": 2943
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095001171740154e-06,
+      "loss": 0.1186,
+      "step": 2944
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.088697618314571e-06,
+      "loss": 0.1288,
+      "step": 2945
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082397673289642e-06,
+      "loss": 0.1255,
+      "step": 2946
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076101340510998e-06,
+      "loss": 0.1242,
+      "step": 2947
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.069808623822077e-06,
+      "loss": 0.1319,
+      "step": 2948
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.063519527064104e-06,
+      "loss": 0.1294,
+      "step": 2949
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05723405407609e-06,
+      "loss": 0.1275,
+      "step": 2950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.050952208694844e-06,
+      "loss": 0.1308,
+      "step": 2951
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0446739947549485e-06,
+      "loss": 0.1285,
+      "step": 2952
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038399416088783e-06,
+      "loss": 0.1266,
+      "step": 2953
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.032128476526493e-06,
+      "loss": 0.1292,
+      "step": 2954
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0258611798960175e-06,
+      "loss": 0.1353,
+      "step": 2955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0195975300230585e-06,
+      "loss": 0.1337,
+      "step": 2956
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.013337530731104e-06,
+      "loss": 0.1196,
+      "step": 2957
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0070811858413985e-06,
+      "loss": 0.1292,
+      "step": 2958
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.000828499172976e-06,
+      "loss": 0.1245,
+      "step": 2959
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.994579474542618e-06,
+      "loss": 0.1153,
+      "step": 2960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.988334115764887e-06,
+      "loss": 0.1254,
+      "step": 2961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98209242665209e-06,
+      "loss": 0.1199,
+      "step": 2962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975854411014312e-06,
+      "loss": 0.1254,
+      "step": 2963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96962007265938e-06,
+      "loss": 0.1086,
+      "step": 2964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.96338941539289e-06,
+      "loss": 0.1312,
+      "step": 2965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957162443018178e-06,
+      "loss": 0.1204,
+      "step": 2966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.950939159336346e-06,
+      "loss": 0.1292,
+      "step": 2967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944719568146226e-06,
+      "loss": 0.1316,
+      "step": 2968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.938503673244413e-06,
+      "loss": 0.1249,
+      "step": 2969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932291478425232e-06,
+      "loss": 0.1286,
+      "step": 2970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.926082987480761e-06,
+      "loss": 0.1207,
+      "step": 2971
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9198782042008074e-06,
+      "loss": 0.1235,
+      "step": 2972
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913677132372925e-06,
+      "loss": 0.1259,
+      "step": 2973
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.907479775782389e-06,
+      "loss": 0.1289,
+      "step": 2974
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901286138212222e-06,
+      "loss": 0.1245,
+      "step": 2975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895096223443162e-06,
+      "loss": 0.1216,
+      "step": 2976
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888910035253689e-06,
+      "loss": 0.1194,
+      "step": 2977
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882727577419993e-06,
+      "loss": 0.1247,
+      "step": 2978
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.876548853716001e-06,
+      "loss": 0.1267,
+      "step": 2979
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870373867913348e-06,
+      "loss": 0.1168,
+      "step": 2980
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864202623781398e-06,
+      "loss": 0.1118,
+      "step": 2981
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858035125087222e-06,
+      "loss": 0.1152,
+      "step": 2982
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85187137559561e-06,
+      "loss": 0.118,
+      "step": 2983
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845711379069069e-06,
+      "loss": 0.1198,
+      "step": 2984
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8395551392678e-06,
+      "loss": 0.1167,
+      "step": 2985
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.833402659949726e-06,
+      "loss": 0.1228,
+      "step": 2986
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827253944870461e-06,
+      "loss": 0.1235,
+      "step": 2987
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.821108997783332e-06,
+      "loss": 0.1182,
+      "step": 2988
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.814967822439364e-06,
+      "loss": 0.1097,
+      "step": 2989
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.808830422587273e-06,
+      "loss": 0.1208,
+      "step": 2990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.80269680197348e-06,
+      "loss": 0.1144,
+      "step": 2991
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7965669643420876e-06,
+      "loss": 0.1297,
+      "step": 2992
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7904409134348996e-06,
+      "loss": 0.12,
+      "step": 2993
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7843186529914066e-06,
+      "loss": 0.1258,
+      "step": 2994
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.778200186748778e-06,
+      "loss": 0.1305,
+      "step": 2995
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.772085518441878e-06,
+      "loss": 0.1261,
+      "step": 2996
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7659746518032393e-06,
+      "loss": 0.1256,
+      "step": 2997
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.75986759056309e-06,
+      "loss": 0.135,
+      "step": 2998
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.753764338449317e-06,
+      "loss": 0.1293,
+      "step": 2999
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.747664899187501e-06,
+      "loss": 0.114,
+      "step": 3000
+    },
+    {
+      "epoch": 2.17,
+      "eval_loss": 0.12288390845060349,
+      "eval_runtime": 716.6812,
+      "eval_samples_per_second": 69.766,
+      "eval_steps_per_second": 2.181,
+      "step": 3000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7415692765008784e-06,
+      "loss": 0.1178,
+      "step": 3001
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7354774741103704e-06,
+      "loss": 0.1275,
+      "step": 3002
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729389495734551e-06,
+      "loss": 0.1193,
+      "step": 3003
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7233053450896773e-06,
+      "loss": 0.1268,
+      "step": 3004
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.717225025889651e-06,
+      "loss": 0.128,
+      "step": 3005
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7111485418460535e-06,
+      "loss": 0.1223,
+      "step": 3006
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.705075896668109e-06,
+      "loss": 0.1212,
+      "step": 3007
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6990070940627132e-06,
+      "loss": 0.1322,
+      "step": 3008
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6929421377344e-06,
+      "loss": 0.1224,
+      "step": 3009
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.686881031385372e-06,
+      "loss": 0.1256,
+      "step": 3010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6808237787154676e-06,
+      "loss": 0.1128,
+      "step": 3011
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6747703834221847e-06,
+      "loss": 0.1176,
+      "step": 3012
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6687208492006556e-06,
+      "loss": 0.1253,
+      "step": 3013
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6626751797436657e-06,
+      "loss": 0.1202,
+      "step": 3014
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6566333787416307e-06,
+      "loss": 0.1302,
+      "step": 3015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6505954498826144e-06,
+      "loss": 0.1434,
+      "step": 3016
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.644561396852313e-06,
+      "loss": 0.1259,
+      "step": 3017
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6385312233340607e-06,
+      "loss": 0.1272,
+      "step": 3018
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6325049330088138e-06,
+      "loss": 0.1175,
+      "step": 3019
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6264825295551687e-06,
+      "loss": 0.123,
+      "step": 3020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.620464016649341e-06,
+      "loss": 0.1223,
+      "step": 3021
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6144493979651805e-06,
+      "loss": 0.1245,
+      "step": 3022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6084386771741475e-06,
+      "loss": 0.1284,
+      "step": 3023
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6024318579453377e-06,
+      "loss": 0.1253,
+      "step": 3024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.59642894394545e-06,
+      "loss": 0.1381,
+      "step": 3025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5904299388388107e-06,
+      "loss": 0.1174,
+      "step": 3026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.58443484628736e-06,
+      "loss": 0.132,
+      "step": 3027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5784436699506397e-06,
+      "loss": 0.1235,
+      "step": 3028
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.572456413485813e-06,
+      "loss": 0.1245,
+      "step": 3029
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5664730805476375e-06,
+      "loss": 0.1248,
+      "step": 3030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.560493674788492e-06,
+      "loss": 0.1181,
+      "step": 3031
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5545181998583413e-06,
+      "loss": 0.1151,
+      "step": 3032
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.548546659404765e-06,
+      "loss": 0.1194,
+      "step": 3033
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.542579057072927e-06,
+      "loss": 0.1281,
+      "step": 3034
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5366153965056025e-06,
+      "loss": 0.1286,
+      "step": 3035
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.530655681343147e-06,
+      "loss": 0.1248,
+      "step": 3036
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5246999152235186e-06,
+      "loss": 0.1113,
+      "step": 3037
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5187481017822533e-06,
+      "loss": 0.123,
+      "step": 3038
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.512800244652488e-06,
+      "loss": 0.1272,
+      "step": 3039
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5068563474649296e-06,
+      "loss": 0.1308,
+      "step": 3040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5009164138478825e-06,
+      "loss": 0.137,
+      "step": 3041
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4949804474272188e-06,
+      "loss": 0.1236,
+      "step": 3042
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4890484518264e-06,
+      "loss": 0.1243,
+      "step": 3043
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4831204306664535e-06,
+      "loss": 0.1202,
+      "step": 3044
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4771963875659896e-06,
+      "loss": 0.1159,
+      "step": 3045
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4712763261411775e-06,
+      "loss": 0.1359,
+      "step": 3046
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.465360250005779e-06,
+      "loss": 0.129,
+      "step": 3047
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4594481627710976e-06,
+      "loss": 0.1258,
+      "step": 3048
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4535400680460194e-06,
+      "loss": 0.1354,
+      "step": 3049
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447635969436981e-06,
+      "loss": 0.1221,
+      "step": 3050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.441735870547992e-06,
+      "loss": 0.124,
+      "step": 3051
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4358397749806074e-06,
+      "loss": 0.123,
+      "step": 3052
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4299476863339533e-06,
+      "loss": 0.1319,
+      "step": 3053
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4240596082046927e-06,
+      "loss": 0.1225,
+      "step": 3054
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.418175544187058e-06,
+      "loss": 0.1207,
+      "step": 3055
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.412295497872816e-06,
+      "loss": 0.132,
+      "step": 3056
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4064194728512944e-06,
+      "loss": 0.1335,
+      "step": 3057
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4005474727093534e-06,
+      "loss": 0.1318,
+      "step": 3058
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3946795010314083e-06,
+      "loss": 0.1168,
+      "step": 3059
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3888155613994045e-06,
+      "loss": 0.1245,
+      "step": 3060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3829556573928323e-06,
+      "loss": 0.1244,
+      "step": 3061
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.377099792588724e-06,
+      "loss": 0.1262,
+      "step": 3062
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.371247970561631e-06,
+      "loss": 0.1252,
+      "step": 3063
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3654001948836524e-06,
+      "loss": 0.1353,
+      "step": 3064
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359556469124404e-06,
+      "loss": 0.1158,
+      "step": 3065
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.353716796851044e-06,
+      "loss": 0.1154,
+      "step": 3066
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.347881181628241e-06,
+      "loss": 0.123,
+      "step": 3067
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3420496270182003e-06,
+      "loss": 0.1195,
+      "step": 3068
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.336222136580637e-06,
+      "loss": 0.1327,
+      "step": 3069
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3303987138727965e-06,
+      "loss": 0.1272,
+      "step": 3070
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3245793624494304e-06,
+      "loss": 0.1245,
+      "step": 3071
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3187640858628157e-06,
+      "loss": 0.1373,
+      "step": 3072
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.31295288766273e-06,
+      "loss": 0.1261,
+      "step": 3073
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.307145771396475e-06,
+      "loss": 0.1258,
+      "step": 3074
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301342740608844e-06,
+      "loss": 0.1195,
+      "step": 3075
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.295543798842159e-06,
+      "loss": 0.1172,
+      "step": 3076
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2897489496362234e-06,
+      "loss": 0.1274,
+      "step": 3077
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2839581965283573e-06,
+      "loss": 0.1165,
+      "step": 3078
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.27817154305337e-06,
+      "loss": 0.1263,
+      "step": 3079
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2723889927435802e-06,
+      "loss": 0.1267,
+      "step": 3080
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2666105491287882e-06,
+      "loss": 0.1214,
+      "step": 3081
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2608362157362994e-06,
+      "loss": 0.1174,
+      "step": 3082
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2550659960909013e-06,
+      "loss": 0.1306,
+      "step": 3083
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24929989371488e-06,
+      "loss": 0.1271,
+      "step": 3084
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2435379121279962e-06,
+      "loss": 0.1142,
+      "step": 3085
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237780054847507e-06,
+      "loss": 0.1168,
+      "step": 3086
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2320263253881412e-06,
+      "loss": 0.1184,
+      "step": 3087
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.226276727262119e-06,
+      "loss": 0.1225,
+      "step": 3088
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2205312639791264e-06,
+      "loss": 0.116,
+      "step": 3089
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.214789939046338e-06,
+      "loss": 0.1347,
+      "step": 3090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2090527559683913e-06,
+      "loss": 0.1228,
+      "step": 3091
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2033197182474053e-06,
+      "loss": 0.1374,
+      "step": 3092
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1975908293829573e-06,
+      "loss": 0.1173,
+      "step": 3093
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1918660928721023e-06,
+      "loss": 0.1279,
+      "step": 3094
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.186145512209359e-06,
+      "loss": 0.1344,
+      "step": 3095
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.180429090886702e-06,
+      "loss": 0.1145,
+      "step": 3096
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1747168323935763e-06,
+      "loss": 0.1195,
+      "step": 3097
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.169008740216877e-06,
+      "loss": 0.1302,
+      "step": 3098
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1633048178409642e-06,
+      "loss": 0.1217,
+      "step": 3099
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1576050687476457e-06,
+      "loss": 0.1349,
+      "step": 3100
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1519094964161888e-06,
+      "loss": 0.1262,
+      "step": 3101
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.146218104323302e-06,
+      "loss": 0.123,
+      "step": 3102
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1405308959431536e-06,
+      "loss": 0.1214,
+      "step": 3103
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1348478747473442e-06,
+      "loss": 0.1202,
+      "step": 3104
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.129169044204933e-06,
+      "loss": 0.1353,
+      "step": 3105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1234944077824102e-06,
+      "loss": 0.1243,
+      "step": 3106
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.117823968943716e-06,
+      "loss": 0.1161,
+      "step": 3107
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.112157731150214e-06,
+      "loss": 0.1206,
+      "step": 3108
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1064956978607187e-06,
+      "loss": 0.1188,
+      "step": 3109
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1008378725314647e-06,
+      "loss": 0.1285,
+      "step": 3110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0951842586161295e-06,
+      "loss": 0.1205,
+      "step": 3111
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0895348595658093e-06,
+      "loss": 0.1342,
+      "step": 3112
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0838896788290384e-06,
+      "loss": 0.1323,
+      "step": 3113
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078248719851765e-06,
+      "loss": 0.1155,
+      "step": 3114
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0726119860773706e-06,
+      "loss": 0.1298,
+      "step": 3115
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.066979480946648e-06,
+      "loss": 0.1232,
+      "step": 3116
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.061351207897818e-06,
+      "loss": 0.1188,
+      "step": 3117
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.055727170366508e-06,
+      "loss": 0.1382,
+      "step": 3118
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.050107371785771e-06,
+      "loss": 0.114,
+      "step": 3119
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0444918155860603e-06,
+      "loss": 0.1288,
+      "step": 3120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038880505195253e-06,
+      "loss": 0.1287,
+      "step": 3121
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0332734440386214e-06,
+      "loss": 0.1167,
+      "step": 3122
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0276706355388554e-06,
+      "loss": 0.1308,
+      "step": 3123
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.022072083116038e-06,
+      "loss": 0.1354,
+      "step": 3124
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.016477790187665e-06,
+      "loss": 0.1194,
+      "step": 3125
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0108877601686226e-06,
+      "loss": 0.1272,
+      "step": 3126
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0053019964712026e-06,
+      "loss": 0.1234,
+      "step": 3127
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9997205025050856e-06,
+      "loss": 0.1202,
+      "step": 3128
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9941432816773507e-06,
+      "loss": 0.1187,
+      "step": 3129
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.98857033739247e-06,
+      "loss": 0.1149,
+      "step": 3130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.983001673052298e-06,
+      "loss": 0.1178,
+      "step": 3131
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9774372920560857e-06,
+      "loss": 0.1292,
+      "step": 3132
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.971877197800458e-06,
+      "loss": 0.126,
+      "step": 3133
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9663213936794334e-06,
+      "loss": 0.1134,
+      "step": 3134
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9607698830844113e-06,
+      "loss": 0.1263,
+      "step": 3135
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9552226694041597e-06,
+      "loss": 0.1227,
+      "step": 3136
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.949679756024838e-06,
+      "loss": 0.1328,
+      "step": 3137
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944141146329966e-06,
+      "loss": 0.1231,
+      "step": 3138
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9386068437004476e-06,
+      "loss": 0.1282,
+      "step": 3139
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.933076851514558e-06,
+      "loss": 0.1185,
+      "step": 3140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9275511731479287e-06,
+      "loss": 0.1284,
+      "step": 3141
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9220298119735746e-06,
+      "loss": 0.1368,
+      "step": 3142
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9165127713618578e-06,
+      "loss": 0.1276,
+      "step": 3143
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9110000546805217e-06,
+      "loss": 0.1268,
+      "step": 3144
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.905491665294652e-06,
+      "loss": 0.123,
+      "step": 3145
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8999876065667077e-06,
+      "loss": 0.118,
+      "step": 3146
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.894487881856494e-06,
+      "loss": 0.1343,
+      "step": 3147
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.888992494521179e-06,
+      "loss": 0.1245,
+      "step": 3148
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8835014479152744e-06,
+      "loss": 0.1219,
+      "step": 3149
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8780147453906506e-06,
+      "loss": 0.1181,
+      "step": 3150
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8725323902965174e-06,
+      "loss": 0.1256,
+      "step": 3151
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.867054385979443e-06,
+      "loss": 0.12,
+      "step": 3152
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8615807357833246e-06,
+      "loss": 0.1181,
+      "step": 3153
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8561114430494165e-06,
+      "loss": 0.1272,
+      "step": 3154
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8506465111162986e-06,
+      "loss": 0.1157,
+      "step": 3155
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8451859433199047e-06,
+      "loss": 0.121,
+      "step": 3156
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8397297429934877e-06,
+      "loss": 0.1263,
+      "step": 3157
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.834277913467651e-06,
+      "loss": 0.1204,
+      "step": 3158
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8288304580703165e-06,
+      "loss": 0.1248,
+      "step": 3159
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8233873801267452e-06,
+      "loss": 0.1281,
+      "step": 3160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.817948682959518e-06,
+      "loss": 0.1201,
+      "step": 3161
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8125143698885483e-06,
+      "loss": 0.1353,
+      "step": 3162
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8070844442310707e-06,
+      "loss": 0.111,
+      "step": 3163
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8016589093016468e-06,
+      "loss": 0.1293,
+      "step": 3164
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7962377684121457e-06,
+      "loss": 0.1274,
+      "step": 3165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7908210248717672e-06,
+      "loss": 0.112,
+      "step": 3166
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7854086819870162e-06,
+      "loss": 0.1292,
+      "step": 3167
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7800007430617228e-06,
+      "loss": 0.1206,
+      "step": 3168
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7745972113970164e-06,
+      "loss": 0.1194,
+      "step": 3169
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7691980902913473e-06,
+      "loss": 0.12,
+      "step": 3170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.763803383040463e-06,
+      "loss": 0.1281,
+      "step": 3171
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7584130929374264e-06,
+      "loss": 0.1306,
+      "step": 3172
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753027223272601e-06,
+      "loss": 0.135,
+      "step": 3173
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.747645777333645e-06,
+      "loss": 0.1284,
+      "step": 3174
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7422687584055284e-06,
+      "loss": 0.1196,
+      "step": 3175
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7368961697705065e-06,
+      "loss": 0.1237,
+      "step": 3176
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7315280147081426e-06,
+      "loss": 0.1226,
+      "step": 3177
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7261642964952796e-06,
+      "loss": 0.1221,
+      "step": 3178
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7208050184060677e-06,
+      "loss": 0.1187,
+      "step": 3179
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7154501837119307e-06,
+      "loss": 0.1314,
+      "step": 3180
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7100997956815954e-06,
+      "loss": 0.1221,
+      "step": 3181
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7047538575810605e-06,
+      "loss": 0.1375,
+      "step": 3182
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.699412372673621e-06,
+      "loss": 0.1188,
+      "step": 3183
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6940753442198407e-06,
+      "loss": 0.1125,
+      "step": 3184
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6887427754775763e-06,
+      "loss": 0.1295,
+      "step": 3185
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6834146697019482e-06,
+      "loss": 0.1153,
+      "step": 3186
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.678091030145368e-06,
+      "loss": 0.1234,
+      "step": 3187
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.672771860057506e-06,
+      "loss": 0.1314,
+      "step": 3188
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6674571626853163e-06,
+      "loss": 0.1282,
+      "step": 3189
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.662146941273013e-06,
+      "loss": 0.1151,
+      "step": 3190
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.656841199062087e-06,
+      "loss": 0.1322,
+      "step": 3191
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6515399392912845e-06,
+      "loss": 0.1283,
+      "step": 3192
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.646243165196627e-06,
+      "loss": 0.1259,
+      "step": 3193
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.640950880011389e-06,
+      "loss": 0.1255,
+      "step": 3194
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6356630869661137e-06,
+      "loss": 0.1261,
+      "step": 3195
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6303797892885897e-06,
+      "loss": 0.1303,
+      "step": 3196
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.625100990203875e-06,
+      "loss": 0.1237,
+      "step": 3197
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6198266929342673e-06,
+      "loss": 0.122,
+      "step": 3198
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.614556900699333e-06,
+      "loss": 0.131,
+      "step": 3199
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.609291616715871e-06,
+      "loss": 0.1253,
+      "step": 3200
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6040308441979444e-06,
+      "loss": 0.108,
+      "step": 3201
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5987745863568494e-06,
+      "loss": 0.1272,
+      "step": 3202
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5935228464011352e-06,
+      "loss": 0.1265,
+      "step": 3203
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.588275627536586e-06,
+      "loss": 0.1188,
+      "step": 3204
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.583032932966235e-06,
+      "loss": 0.1268,
+      "step": 3205
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5777947658903422e-06,
+      "loss": 0.1305,
+      "step": 3206
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5725611295064144e-06,
+      "loss": 0.1262,
+      "step": 3207
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5673320270091918e-06,
+      "loss": 0.1172,
+      "step": 3208
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.562107461590637e-06,
+      "loss": 0.1362,
+      "step": 3209
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5568874364399563e-06,
+      "loss": 0.1237,
+      "step": 3210
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551671954743572e-06,
+      "loss": 0.1414,
+      "step": 3211
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5464610196851457e-06,
+      "loss": 0.1327,
+      "step": 3212
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.54125463444555e-06,
+      "loss": 0.1291,
+      "step": 3213
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.536052802202894e-06,
+      "loss": 0.1199,
+      "step": 3214
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.530855526132494e-06,
+      "loss": 0.1266,
+      "step": 3215
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5256628094069e-06,
+      "loss": 0.1179,
+      "step": 3216
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5204746551958627e-06,
+      "loss": 0.1215,
+      "step": 3217
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.515291066666362e-06,
+      "loss": 0.1151,
+      "step": 3218
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.510112046982579e-06,
+      "loss": 0.1399,
+      "step": 3219
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5049375993059166e-06,
+      "loss": 0.1231,
+      "step": 3220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.499767726794977e-06,
+      "loss": 0.1246,
+      "step": 3221
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4946024326055775e-06,
+      "loss": 0.1264,
+      "step": 3222
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.489441719890736e-06,
+      "loss": 0.1374,
+      "step": 3223
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.484285591800679e-06,
+      "loss": 0.127,
+      "step": 3224
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4791340514828245e-06,
+      "loss": 0.1174,
+      "step": 3225
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4739871020818017e-06,
+      "loss": 0.1251,
+      "step": 3226
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.468844746739427e-06,
+      "loss": 0.1251,
+      "step": 3227
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4637069885947228e-06,
+      "loss": 0.1167,
+      "step": 3228
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4585738307838936e-06,
+      "loss": 0.1234,
+      "step": 3229
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4534452764403483e-06,
+      "loss": 0.1254,
+      "step": 3230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4483213286946727e-06,
+      "loss": 0.1253,
+      "step": 3231
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.443201990674655e-06,
+      "loss": 0.1315,
+      "step": 3232
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4380872655052533e-06,
+      "loss": 0.1235,
+      "step": 3233
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.432977156308627e-06,
+      "loss": 0.1142,
+      "step": 3234
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4278716662041015e-06,
+      "loss": 0.1244,
+      "step": 3235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.422770798308196e-06,
+      "loss": 0.1329,
+      "step": 3236
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4176745557345994e-06,
+      "loss": 0.1266,
+      "step": 3237
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4125829415941824e-06,
+      "loss": 0.1306,
+      "step": 3238
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4074959589949844e-06,
+      "loss": 0.1293,
+      "step": 3239
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4024136110422246e-06,
+      "loss": 0.1289,
+      "step": 3240
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3973359008382934e-06,
+      "loss": 0.1269,
+      "step": 3241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.392262831482739e-06,
+      "loss": 0.1221,
+      "step": 3242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3871944060722917e-06,
+      "loss": 0.1288,
+      "step": 3243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.382130627700835e-06,
+      "loss": 0.1344,
+      "step": 3244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3770714994594236e-06,
+      "loss": 0.1193,
+      "step": 3245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.372017024436267e-06,
+      "loss": 0.1157,
+      "step": 3246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3669672057167435e-06,
+      "loss": 0.1169,
+      "step": 3247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3619220463833768e-06,
+      "loss": 0.1196,
+      "step": 3248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.35688154951586e-06,
+      "loss": 0.1257,
+      "step": 3249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3518457181910272e-06,
+      "loss": 0.1137,
+      "step": 3250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3468145554828724e-06,
+      "loss": 0.1302,
+      "step": 3251
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.341788064462538e-06,
+      "loss": 0.1261,
+      "step": 3252
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.336766248198319e-06,
+      "loss": 0.123,
+      "step": 3253
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3317491097556455e-06,
+      "loss": 0.1264,
+      "step": 3254
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3267366521971046e-06,
+      "loss": 0.1335,
+      "step": 3255
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3217288785824144e-06,
+      "loss": 0.1329,
+      "step": 3256
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3167257919684473e-06,
+      "loss": 0.1238,
+      "step": 3257
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3117273954092014e-06,
+      "loss": 0.1255,
+      "step": 3258
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3067336919558214e-06,
+      "loss": 0.1166,
+      "step": 3259
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3017446846565795e-06,
+      "loss": 0.1293,
+      "step": 3260
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296760376556891e-06,
+      "loss": 0.1167,
+      "step": 3261
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291780770699291e-06,
+      "loss": 0.1184,
+      "step": 3262
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2868058701234565e-06,
+      "loss": 0.1118,
+      "step": 3263
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2818356778661797e-06,
+      "loss": 0.123,
+      "step": 3264
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2768701969613905e-06,
+      "loss": 0.1232,
+      "step": 3265
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271909430440132e-06,
+      "loss": 0.1245,
+      "step": 3266
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2669533813305824e-06,
+      "loss": 0.1246,
+      "step": 3267
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2620020526580243e-06,
+      "loss": 0.1193,
+      "step": 3268
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2570554474448746e-06,
+      "loss": 0.113,
+      "step": 3269
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.252113568710653e-06,
+      "loss": 0.1219,
+      "step": 3270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247176419472009e-06,
+      "loss": 0.1259,
+      "step": 3271
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2422440027426883e-06,
+      "loss": 0.1335,
+      "step": 3272
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2373163215335604e-06,
+      "loss": 0.1188,
+      "step": 3273
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.232393378852602e-06,
+      "loss": 0.1211,
+      "step": 3274
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.227475177704891e-06,
+      "loss": 0.1281,
+      "step": 3275
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222561721092621e-06,
+      "loss": 0.1196,
+      "step": 3276
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2176530120150753e-06,
+      "loss": 0.1296,
+      "step": 3277
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.212749053468657e-06,
+      "loss": 0.1258,
+      "step": 3278
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2078498484468512e-06,
+      "loss": 0.1308,
+      "step": 3279
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2029553999402574e-06,
+      "loss": 0.1317,
+      "step": 3280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1980657109365577e-06,
+      "loss": 0.1243,
+      "step": 3281
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.193180784420539e-06,
+      "loss": 0.1034,
+      "step": 3282
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.188300623374078e-06,
+      "loss": 0.1259,
+      "step": 3283
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1834252307761426e-06,
+      "loss": 0.1244,
+      "step": 3284
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1785546096027865e-06,
+      "loss": 0.1373,
+      "step": 3285
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1736887628271575e-06,
+      "loss": 0.1234,
+      "step": 3286
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.16882769341948e-06,
+      "loss": 0.1318,
+      "step": 3287
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1639714043470717e-06,
+      "loss": 0.1228,
+      "step": 3288
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1591198985743233e-06,
+      "loss": 0.1208,
+      "step": 3289
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154273179062717e-06,
+      "loss": 0.122,
+      "step": 3290
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149431248770799e-06,
+      "loss": 0.1219,
+      "step": 3291
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1445941106542057e-06,
+      "loss": 0.124,
+      "step": 3292
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1397617676656368e-06,
+      "loss": 0.1251,
+      "step": 3293
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1349342227548753e-06,
+      "loss": 0.1256,
+      "step": 3294
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.130111478868766e-06,
+      "loss": 0.1294,
+      "step": 3295
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.125293538951232e-06,
+      "loss": 0.1081,
+      "step": 3296
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1204804059432536e-06,
+      "loss": 0.1166,
+      "step": 3297
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1156720827828893e-06,
+      "loss": 0.1182,
+      "step": 3298
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1108685724052467e-06,
+      "loss": 0.1338,
+      "step": 3299
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1060698777425124e-06,
+      "loss": 0.1229,
+      "step": 3300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1012760017239177e-06,
+      "loss": 0.1313,
+      "step": 3301
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0964869472757634e-06,
+      "loss": 0.123,
+      "step": 3302
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0917027173214e-06,
+      "loss": 0.1124,
+      "step": 3303
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0869233147812396e-06,
+      "loss": 0.1244,
+      "step": 3304
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0821487425727403e-06,
+      "loss": 0.1301,
+      "step": 3305
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.07737900361042e-06,
+      "loss": 0.1201,
+      "step": 3306
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0726141008058364e-06,
+      "loss": 0.1159,
+      "step": 3307
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0678540370676036e-06,
+      "loss": 0.1262,
+      "step": 3308
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.063098815301381e-06,
+      "loss": 0.1271,
+      "step": 3309
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0583484384098627e-06,
+      "loss": 0.1264,
+      "step": 3310
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.053602909292799e-06,
+      "loss": 0.1229,
+      "step": 3311
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.048862230846974e-06,
+      "loss": 0.1147,
+      "step": 3312
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.044126405966207e-06,
+      "loss": 0.1351,
+      "step": 3313
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039395437541366e-06,
+      "loss": 0.1209,
+      "step": 3314
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.034669328460339e-06,
+      "loss": 0.1158,
+      "step": 3315
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.029948081608065e-06,
+      "loss": 0.1177,
+      "step": 3316
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025231699866499e-06,
+      "loss": 0.1128,
+      "step": 3317
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0205201861146363e-06,
+      "loss": 0.1241,
+      "step": 3318
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.015813543228501e-06,
+      "loss": 0.1281,
+      "step": 3319
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0111117740811337e-06,
+      "loss": 0.1288,
+      "step": 3320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0064148815426144e-06,
+      "loss": 0.1174,
+      "step": 3321
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0017228684800337e-06,
+      "loss": 0.1248,
+      "step": 3322
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997035737757512e-06,
+      "loss": 0.1367,
+      "step": 3323
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9923534922361833e-06,
+      "loss": 0.1194,
+      "step": 3324
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9876761347742056e-06,
+      "loss": 0.113,
+      "step": 3325
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.983003668226746e-06,
+      "loss": 0.1239,
+      "step": 3326
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978336095445995e-06,
+      "loss": 0.135,
+      "step": 3327
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9736734192811456e-06,
+      "loss": 0.1299,
+      "step": 3328
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9690156425784133e-06,
+      "loss": 0.1274,
+      "step": 3329
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9643627681810095e-06,
+      "loss": 0.119,
+      "step": 3330
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9597147989291678e-06,
+      "loss": 0.1219,
+      "step": 3331
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9550717376601127e-06,
+      "loss": 0.1282,
+      "step": 3332
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.950433587208086e-06,
+      "loss": 0.1316,
+      "step": 3333
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.945800350404321e-06,
+      "loss": 0.1166,
+      "step": 3334
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.941172030077062e-06,
+      "loss": 0.1271,
+      "step": 3335
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93654862905154e-06,
+      "loss": 0.1228,
+      "step": 3336
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9319301501499964e-06,
+      "loss": 0.124,
+      "step": 3337
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9273165961916563e-06,
+      "loss": 0.132,
+      "step": 3338
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9227079699927453e-06,
+      "loss": 0.1227,
+      "step": 3339
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.918104274366479e-06,
+      "loss": 0.1417,
+      "step": 3340
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.913505512123067e-06,
+      "loss": 0.1291,
+      "step": 3341
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9089116860697e-06,
+      "loss": 0.122,
+      "step": 3342
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9043227990105618e-06,
+      "loss": 0.1184,
+      "step": 3343
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8997388537468165e-06,
+      "loss": 0.1182,
+      "step": 3344
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.895159853076618e-06,
+      "loss": 0.1273,
+      "step": 3345
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8905857997950927e-06,
+      "loss": 0.1258,
+      "step": 3346
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8860166966943583e-06,
+      "loss": 0.1267,
+      "step": 3347
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8814525465634981e-06,
+      "loss": 0.1241,
+      "step": 3348
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8768933521885868e-06,
+      "loss": 0.1222,
+      "step": 3349
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8723391163526582e-06,
+      "loss": 0.1224,
+      "step": 3350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.86778984183573e-06,
+      "loss": 0.1188,
+      "step": 3351
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8632455314147924e-06,
+      "loss": 0.1212,
+      "step": 3352
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8587061878637947e-06,
+      "loss": 0.1234,
+      "step": 3353
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8541718139536669e-06,
+      "loss": 0.1152,
+      "step": 3354
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8496424124522937e-06,
+      "loss": 0.1323,
+      "step": 3355
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8451179861245373e-06,
+      "loss": 0.1255,
+      "step": 3356
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8405985377322078e-06,
+      "loss": 0.1261,
+      "step": 3357
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8360840700340932e-06,
+      "loss": 0.1253,
+      "step": 3358
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8315745857859246e-06,
+      "loss": 0.1287,
+      "step": 3359
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8270700877404068e-06,
+      "loss": 0.1186,
+      "step": 3360
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.822570578647187e-06,
+      "loss": 0.1238,
+      "step": 3361
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8180760612528792e-06,
+      "loss": 0.1292,
+      "step": 3362
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8135865383010387e-06,
+      "loss": 0.1268,
+      "step": 3363
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8091020125321835e-06,
+      "loss": 0.1274,
+      "step": 3364
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8046224866837702e-06,
+      "loss": 0.124,
+      "step": 3365
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8001479634902141e-06,
+      "loss": 0.1202,
+      "step": 3366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7956784456828658e-06,
+      "loss": 0.1219,
+      "step": 3367
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7912139359900326e-06,
+      "loss": 0.122,
+      "step": 3368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7867544371369483e-06,
+      "loss": 0.1221,
+      "step": 3369
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7822999518458107e-06,
+      "loss": 0.1249,
+      "step": 3370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7778504828357345e-06,
+      "loss": 0.1198,
+      "step": 3371
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7734060328227886e-06,
+      "loss": 0.1226,
+      "step": 3372
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7689666045199661e-06,
+      "loss": 0.1282,
+      "step": 3373
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7645322006372057e-06,
+      "loss": 0.1302,
+      "step": 3374
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.76010282388137e-06,
+      "loss": 0.1287,
+      "step": 3375
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7556784769562596e-06,
+      "loss": 0.1277,
+      "step": 3376
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7512591625625973e-06,
+      "loss": 0.1296,
+      "step": 3377
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7468448833980445e-06,
+      "loss": 0.1238,
+      "step": 3378
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7424356421571775e-06,
+      "loss": 0.1193,
+      "step": 3379
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.738031441531508e-06,
+      "loss": 0.1207,
+      "step": 3380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.733632284209461e-06,
+      "loss": 0.1264,
+      "step": 3381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7292381728763918e-06,
+      "loss": 0.1169,
+      "step": 3382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.724849110214567e-06,
+      "loss": 0.1409,
+      "step": 3383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.720465098903179e-06,
+      "loss": 0.1237,
+      "step": 3384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7160861416183304e-06,
+      "loss": 0.1268,
+      "step": 3385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7117122410330433e-06,
+      "loss": 0.1184,
+      "step": 3386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.707343399817254e-06,
+      "loss": 0.1254,
+      "step": 3387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7029796206378035e-06,
+      "loss": 0.1191,
+      "step": 3388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6986209061584502e-06,
+      "loss": 0.1285,
+      "step": 3389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6942672590398556e-06,
+      "loss": 0.122,
+      "step": 3390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.689918681939593e-06,
+      "loss": 0.1223,
+      "step": 3391
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6855751775121355e-06,
+      "loss": 0.1275,
+      "step": 3392
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6812367484088643e-06,
+      "loss": 0.127,
+      "step": 3393
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.676903397278057e-06,
+      "loss": 0.1241,
+      "step": 3394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6725751267649005e-06,
+      "loss": 0.1244,
+      "step": 3395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6682519395114694e-06,
+      "loss": 0.1404,
+      "step": 3396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.663933838156746e-06,
+      "loss": 0.1217,
+      "step": 3397
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6596208253365952e-06,
+      "loss": 0.1278,
+      "step": 3398
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6553129036837934e-06,
+      "loss": 0.1139,
+      "step": 3399
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6510100758279912e-06,
+      "loss": 0.1187,
+      "step": 3400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6467123443957433e-06,
+      "loss": 0.1217,
+      "step": 3401
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6424197120104834e-06,
+      "loss": 0.1215,
+      "step": 3402
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.638132181292541e-06,
+      "loss": 0.119,
+      "step": 3403
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6338497548591248e-06,
+      "loss": 0.1201,
+      "step": 3404
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6295724353243326e-06,
+      "loss": 0.1226,
+      "step": 3405
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.62530022529914e-06,
+      "loss": 0.1293,
+      "step": 3406
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6210331273914103e-06,
+      "loss": 0.1279,
+      "step": 3407
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6167711442058786e-06,
+      "loss": 0.1331,
+      "step": 3408
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6125142783441649e-06,
+      "loss": 0.1232,
+      "step": 3409
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6082625324047585e-06,
+      "loss": 0.1259,
+      "step": 3410
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.604015908983031e-06,
+      "loss": 0.1252,
+      "step": 3411
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5997744106712188e-06,
+      "loss": 0.1278,
+      "step": 3412
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5955380400584386e-06,
+      "loss": 0.1186,
+      "step": 3413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5913067997306685e-06,
+      "loss": 0.1264,
+      "step": 3414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5870806922707626e-06,
+      "loss": 0.1261,
+      "step": 3415
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.582859720258434e-06,
+      "loss": 0.1243,
+      "step": 3416
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5786438862702702e-06,
+      "loss": 0.1185,
+      "step": 3417
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.574433192879713e-06,
+      "loss": 0.1208,
+      "step": 3418
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5702276426570717e-06,
+      "loss": 0.1357,
+      "step": 3419
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5660272381695184e-06,
+      "loss": 0.1291,
+      "step": 3420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5618319819810745e-06,
+      "loss": 0.1265,
+      "step": 3421
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5576418766526313e-06,
+      "loss": 0.1161,
+      "step": 3422
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.553456924741924e-06,
+      "loss": 0.1225,
+      "step": 3423
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5492771288035534e-06,
+      "loss": 0.1254,
+      "step": 3424
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5451024913889601e-06,
+      "loss": 0.1045,
+      "step": 3425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5409330150464498e-06,
+      "loss": 0.1183,
+      "step": 3426
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5367687023211642e-06,
+      "loss": 0.1271,
+      "step": 3427
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5326095557551023e-06,
+      "loss": 0.1164,
+      "step": 3428
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.528455577887109e-06,
+      "loss": 0.1269,
+      "step": 3429
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5243067712528714e-06,
+      "loss": 0.1242,
+      "step": 3430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.520163138384918e-06,
+      "loss": 0.1241,
+      "step": 3431
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5160246818126246e-06,
+      "loss": 0.1209,
+      "step": 3432
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5118914040622013e-06,
+      "loss": 0.1326,
+      "step": 3433
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5077633076567033e-06,
+      "loss": 0.1279,
+      "step": 3434
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.503640395116015e-06,
+      "loss": 0.1299,
+      "step": 3435
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4995226689568665e-06,
+      "loss": 0.1142,
+      "step": 3436
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4954101316928117e-06,
+      "loss": 0.1173,
+      "step": 3437
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4913027858342456e-06,
+      "loss": 0.1333,
+      "step": 3438
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.487200633888386e-06,
+      "loss": 0.1252,
+      "step": 3439
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.483103678359291e-06,
+      "loss": 0.118,
+      "step": 3440
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4790119217478338e-06,
+      "loss": 0.1255,
+      "step": 3441
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4749253665517272e-06,
+      "loss": 0.1284,
+      "step": 3442
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4708440152654968e-06,
+      "loss": 0.1217,
+      "step": 3443
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4667678703805022e-06,
+      "loss": 0.1322,
+      "step": 3444
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.462696934384914e-06,
+      "loss": 0.1193,
+      "step": 3445
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4586312097637356e-06,
+      "loss": 0.1169,
+      "step": 3446
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4545706989987774e-06,
+      "loss": 0.1197,
+      "step": 3447
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4505154045686764e-06,
+      "loss": 0.1344,
+      "step": 3448
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4464653289488783e-06,
+      "loss": 0.1252,
+      "step": 3449
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.44242047461165e-06,
+      "loss": 0.1211,
+      "step": 3450
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4383808440260626e-06,
+      "loss": 0.1172,
+      "step": 3451
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.43434643965801e-06,
+      "loss": 0.1275,
+      "step": 3452
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4303172639701824e-06,
+      "loss": 0.1336,
+      "step": 3453
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4262933194220873e-06,
+      "loss": 0.1307,
+      "step": 3454
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4222746084700412e-06,
+      "loss": 0.1312,
+      "step": 3455
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4182611335671559e-06,
+      "loss": 0.138,
+      "step": 3456
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4142528971633563e-06,
+      "loss": 0.1343,
+      "step": 3457
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.410249901705365e-06,
+      "loss": 0.1337,
+      "step": 3458
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4062521496367055e-06,
+      "loss": 0.1109,
+      "step": 3459
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.402259643397703e-06,
+      "loss": 0.1136,
+      "step": 3460
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3982723854254777e-06,
+      "loss": 0.1307,
+      "step": 3461
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3942903781539484e-06,
+      "loss": 0.1256,
+      "step": 3462
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.390313624013826e-06,
+      "loss": 0.1273,
+      "step": 3463
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.386342125432617e-06,
+      "loss": 0.1219,
+      "step": 3464
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3823758848346236e-06,
+      "loss": 0.14,
+      "step": 3465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3784149046409279e-06,
+      "loss": 0.1245,
+      "step": 3466
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3744591872694113e-06,
+      "loss": 0.1351,
+      "step": 3467
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3705087351347357e-06,
+      "loss": 0.1259,
+      "step": 3468
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3665635506483543e-06,
+      "loss": 0.1228,
+      "step": 3469
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3626236362184996e-06,
+      "loss": 0.1226,
+      "step": 3470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3586889942501935e-06,
+      "loss": 0.1245,
+      "step": 3471
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3547596271452313e-06,
+      "loss": 0.1255,
+      "step": 3472
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3508355373021975e-06,
+      "loss": 0.1199,
+      "step": 3473
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3469167271164452e-06,
+      "loss": 0.1251,
+      "step": 3474
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3430031989801163e-06,
+      "loss": 0.1233,
+      "step": 3475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3390949552821164e-06,
+      "loss": 0.1247,
+      "step": 3476
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3351919984081351e-06,
+      "loss": 0.1292,
+      "step": 3477
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3312943307406279e-06,
+      "loss": 0.1338,
+      "step": 3478
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327401954658828e-06,
+      "loss": 0.1346,
+      "step": 3479
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3235148725387304e-06,
+      "loss": 0.1173,
+      "step": 3480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3196330867531093e-06,
+      "loss": 0.1359,
+      "step": 3481
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3157565996714927e-06,
+      "loss": 0.1292,
+      "step": 3482
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3118854136601877e-06,
+      "loss": 0.1198,
+      "step": 3483
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3080195310822541e-06,
+      "loss": 0.1173,
+      "step": 3484
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.304158954297524e-06,
+      "loss": 0.1221,
+      "step": 3485
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3003036856625784e-06,
+      "loss": 0.1224,
+      "step": 3486
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2964537275307754e-06,
+      "loss": 0.1285,
+      "step": 3487
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.292609082252214e-06,
+      "loss": 0.1288,
+      "step": 3488
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.288769752173763e-06,
+      "loss": 0.1248,
+      "step": 3489
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2849357396390382e-06,
+      "loss": 0.1149,
+      "step": 3490
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2811070469884147e-06,
+      "loss": 0.1224,
+      "step": 3491
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2772836765590158e-06,
+      "loss": 0.128,
+      "step": 3492
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2734656306847226e-06,
+      "loss": 0.1211,
+      "step": 3493
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2696529116961565e-06,
+      "loss": 0.1192,
+      "step": 3494
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2658455219206978e-06,
+      "loss": 0.1275,
+      "step": 3495
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262043463682463e-06,
+      "loss": 0.1237,
+      "step": 3496
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2582467393023235e-06,
+      "loss": 0.1166,
+      "step": 3497
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2544553510978918e-06,
+      "loss": 0.1218,
+      "step": 3498
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.250669301383518e-06,
+      "loss": 0.1307,
+      "step": 3499
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.246888592470301e-06,
+      "loss": 0.1248,
+      "step": 3500
+    },
+    {
+      "epoch": 2.53,
+      "eval_loss": 0.1225692555308342,
+      "eval_runtime": 716.0296,
+      "eval_samples_per_second": 69.83,
+      "eval_steps_per_second": 2.183,
+      "step": 3500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.243113226666074e-06,
+      "loss": 0.1256,
+      "step": 3501
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.239343206275414e-06,
+      "loss": 0.1289,
+      "step": 3502
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2355785335996263e-06,
+      "loss": 0.1259,
+      "step": 3503
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2318192109367632e-06,
+      "loss": 0.1295,
+      "step": 3504
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2280652405816006e-06,
+      "loss": 0.1277,
+      "step": 3505
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2243166248256566e-06,
+      "loss": 0.1274,
+      "step": 3506
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2205733659571706e-06,
+      "loss": 0.1275,
+      "step": 3507
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2168354662611225e-06,
+      "loss": 0.1313,
+      "step": 3508
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.213102928019213e-06,
+      "loss": 0.1168,
+      "step": 3509
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.209375753509875e-06,
+      "loss": 0.1228,
+      "step": 3510
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2056539450082616e-06,
+      "loss": 0.1192,
+      "step": 3511
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2019375047862592e-06,
+      "loss": 0.1174,
+      "step": 3512
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1982264351124662e-06,
+      "loss": 0.1315,
+      "step": 3513
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.194520738252214e-06,
+      "loss": 0.1171,
+      "step": 3514
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1908204164675408e-06,
+      "loss": 0.117,
+      "step": 3515
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.187125472017222e-06,
+      "loss": 0.1262,
+      "step": 3516
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1834359071567325e-06,
+      "loss": 0.1236,
+      "step": 3517
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1797517241382772e-06,
+      "loss": 0.1225,
+      "step": 3518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1760729252107628e-06,
+      "loss": 0.132,
+      "step": 3519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1723995126198228e-06,
+      "loss": 0.1177,
+      "step": 3520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1687314886077904e-06,
+      "loss": 0.1141,
+      "step": 3521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1650688554137212e-06,
+      "loss": 0.1169,
+      "step": 3522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.16141161527337e-06,
+      "loss": 0.121,
+      "step": 3523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1577597704192056e-06,
+      "loss": 0.1381,
+      "step": 3524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1541133230804002e-06,
+      "loss": 0.126,
+      "step": 3525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1504722754828356e-06,
+      "loss": 0.1272,
+      "step": 3526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1468366298490918e-06,
+      "loss": 0.1123,
+      "step": 3527
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1432063883984557e-06,
+      "loss": 0.1235,
+      "step": 3528
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1395815533469101e-06,
+      "loss": 0.1174,
+      "step": 3529
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1359621269071452e-06,
+      "loss": 0.1209,
+      "step": 3530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1323481112885459e-06,
+      "loss": 0.1219,
+      "step": 3531
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1287395086971908e-06,
+      "loss": 0.1204,
+      "step": 3532
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1251363213358612e-06,
+      "loss": 0.1195,
+      "step": 3533
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1215385514040245e-06,
+      "loss": 0.1201,
+      "step": 3534
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1179462010978504e-06,
+      "loss": 0.1242,
+      "step": 3535
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1143592726101915e-06,
+      "loss": 0.1187,
+      "step": 3536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1107777681305975e-06,
+      "loss": 0.1264,
+      "step": 3537
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1072016898453031e-06,
+      "loss": 0.1273,
+      "step": 3538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.103631039937234e-06,
+      "loss": 0.1222,
+      "step": 3539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.100065820585998e-06,
+      "loss": 0.1252,
+      "step": 3540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0965060339678935e-06,
+      "loss": 0.125,
+      "step": 3541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.092951682255896e-06,
+      "loss": 0.1306,
+      "step": 3542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0894027676196716e-06,
+      "loss": 0.1258,
+      "step": 3543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0858592922255572e-06,
+      "loss": 0.1258,
+      "step": 3544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0823212582365782e-06,
+      "loss": 0.1382,
+      "step": 3545
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.078788667812436e-06,
+      "loss": 0.1334,
+      "step": 3546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0752615231095086e-06,
+      "loss": 0.1265,
+      "step": 3547
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0717398262808466e-06,
+      "loss": 0.1186,
+      "step": 3548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.068223579476183e-06,
+      "loss": 0.1315,
+      "step": 3549
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0647127848419136e-06,
+      "loss": 0.1259,
+      "step": 3550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0612074445211173e-06,
+      "loss": 0.1285,
+      "step": 3551
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0577075606535314e-06,
+      "loss": 0.1176,
+      "step": 3552
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0542131353755758e-06,
+      "loss": 0.1114,
+      "step": 3553
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0507241708203253e-06,
+      "loss": 0.1201,
+      "step": 3554
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.047240669117533e-06,
+      "loss": 0.1222,
+      "step": 3555
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0437626323936067e-06,
+      "loss": 0.1237,
+      "step": 3556
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0402900627716273e-06,
+      "loss": 0.1252,
+      "step": 3557
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.036822962371331e-06,
+      "loss": 0.1189,
+      "step": 3558
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.033361333309123e-06,
+      "loss": 0.1276,
+      "step": 3559
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0299051776980607e-06,
+      "loss": 0.1214,
+      "step": 3560
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0264544976478684e-06,
+      "loss": 0.1246,
+      "step": 3561
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0230092952649196e-06,
+      "loss": 0.1156,
+      "step": 3562
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0195695726522525e-06,
+      "loss": 0.1292,
+      "step": 3563
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0161353319095523e-06,
+      "loss": 0.1228,
+      "step": 3564
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0127065751331645e-06,
+      "loss": 0.1312,
+      "step": 3565
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0092833044160844e-06,
+      "loss": 0.1159,
+      "step": 3566
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0058655218479564e-06,
+      "loss": 0.125,
+      "step": 3567
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.00245322951508e-06,
+      "loss": 0.122,
+      "step": 3568
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990464295003966e-07,
+      "loss": 0.124,
+      "step": 3569
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95645123883503e-07,
+      "loss": 0.1173,
+      "step": 3570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.92249314740632e-07,
+      "loss": 0.1307,
+      "step": 3571
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.888590041446733e-07,
+      "loss": 0.1217,
+      "step": 3572
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.85474194165148e-07,
+      "loss": 0.1278,
+      "step": 3573
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.820948868682278e-07,
+      "loss": 0.1244,
+      "step": 3574
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.787210843167226e-07,
+      "loss": 0.1288,
+      "step": 3575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.753527885700852e-07,
+      "loss": 0.1248,
+      "step": 3576
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.719900016843986e-07,
+      "loss": 0.1267,
+      "step": 3577
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.68632725712394e-07,
+      "loss": 0.121,
+      "step": 3578
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.652809627034277e-07,
+      "loss": 0.1316,
+      "step": 3579
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.619347147035007e-07,
+      "loss": 0.1207,
+      "step": 3580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.585939837552405e-07,
+      "loss": 0.1223,
+      "step": 3581
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55258771897911e-07,
+      "loss": 0.1345,
+      "step": 3582
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.519290811674021e-07,
+      "loss": 0.1281,
+      "step": 3583
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.486049135962417e-07,
+      "loss": 0.1279,
+      "step": 3584
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.452862712135769e-07,
+      "loss": 0.12,
+      "step": 3585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.419731560451906e-07,
+      "loss": 0.1234,
+      "step": 3586
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.386655701134839e-07,
+      "loss": 0.1254,
+      "step": 3587
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.353635154374896e-07,
+      "loss": 0.1179,
+      "step": 3588
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.320669940328575e-07,
+      "loss": 0.1362,
+      "step": 3589
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.287760079118679e-07,
+      "loss": 0.1215,
+      "step": 3590
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.254905590834129e-07,
+      "loss": 0.1353,
+      "step": 3591
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.222106495530136e-07,
+      "loss": 0.1217,
+      "step": 3592
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.18936281322802e-07,
+      "loss": 0.1248,
+      "step": 3593
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.156674563915324e-07,
+      "loss": 0.1244,
+      "step": 3594
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.124041767545733e-07,
+      "loss": 0.1204,
+      "step": 3595
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.0914644440391e-07,
+      "loss": 0.1266,
+      "step": 3596
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.058942613281385e-07,
+      "loss": 0.1251,
+      "step": 3597
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.026476295124698e-07,
+      "loss": 0.131,
+      "step": 3598
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.994065509387285e-07,
+      "loss": 0.1239,
+      "step": 3599
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.961710275853419e-07,
+      "loss": 0.1228,
+      "step": 3600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.929410614273559e-07,
+      "loss": 0.114,
+      "step": 3601
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.897166544364144e-07,
+      "loss": 0.1281,
+      "step": 3602
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.864978085807785e-07,
+      "loss": 0.1273,
+      "step": 3603
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.832845258253031e-07,
+      "loss": 0.1232,
+      "step": 3604
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.800768081314548e-07,
+      "loss": 0.1199,
+      "step": 3605
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.76874657457305e-07,
+      "loss": 0.1279,
+      "step": 3606
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.73678075757517e-07,
+      "loss": 0.1202,
+      "step": 3607
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.704870649833652e-07,
+      "loss": 0.12,
+      "step": 3608
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.673016270827195e-07,
+      "loss": 0.1302,
+      "step": 3609
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.641217640000443e-07,
+      "loss": 0.1198,
+      "step": 3610
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.609474776764082e-07,
+      "loss": 0.1172,
+      "step": 3611
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.577787700494677e-07,
+      "loss": 0.1231,
+      "step": 3612
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.546156430534813e-07,
+      "loss": 0.1227,
+      "step": 3613
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.514580986192933e-07,
+      "loss": 0.1332,
+      "step": 3614
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.483061386743496e-07,
+      "loss": 0.1301,
+      "step": 3615
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.45159765142678e-07,
+      "loss": 0.1278,
+      "step": 3616
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.42018979944903e-07,
+      "loss": 0.1227,
+      "step": 3617
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.388837849982323e-07,
+      "loss": 0.1234,
+      "step": 3618
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.357541822164672e-07,
+      "loss": 0.1198,
+      "step": 3619
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.326301735099895e-07,
+      "loss": 0.1214,
+      "step": 3620
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.295117607857717e-07,
+      "loss": 0.1304,
+      "step": 3621
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.263989459473654e-07,
+      "loss": 0.1272,
+      "step": 3622
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.232917308949096e-07,
+      "loss": 0.1201,
+      "step": 3623
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.201901175251193e-07,
+      "loss": 0.1199,
+      "step": 3624
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.170941077312977e-07,
+      "loss": 0.1267,
+      "step": 3625
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.140037034033199e-07,
+      "loss": 0.1167,
+      "step": 3626
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.109189064276457e-07,
+      "loss": 0.1385,
+      "step": 3627
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.078397186873066e-07,
+      "loss": 0.1261,
+      "step": 3628
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.047661420619146e-07,
+      "loss": 0.1266,
+      "step": 3629
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.016981784276534e-07,
+      "loss": 0.1305,
+      "step": 3630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.986358296572827e-07,
+      "loss": 0.1281,
+      "step": 3631
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.955790976201305e-07,
+      "loss": 0.1222,
+      "step": 3632
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.925279841821032e-07,
+      "loss": 0.1344,
+      "step": 3633
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.894824912056698e-07,
+      "loss": 0.1228,
+      "step": 3634
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.864426205498776e-07,
+      "loss": 0.1202,
+      "step": 3635
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.834083740703313e-07,
+      "loss": 0.1259,
+      "step": 3636
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.803797536192103e-07,
+      "loss": 0.1218,
+      "step": 3637
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.773567610452559e-07,
+      "loss": 0.1248,
+      "step": 3638
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.743393981937764e-07,
+      "loss": 0.1254,
+      "step": 3639
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.713276669066405e-07,
+      "loss": 0.1218,
+      "step": 3640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.683215690222845e-07,
+      "loss": 0.1289,
+      "step": 3641
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.653211063756982e-07,
+      "loss": 0.1278,
+      "step": 3642
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.623262807984377e-07,
+      "loss": 0.1244,
+      "step": 3643
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.593370941186195e-07,
+      "loss": 0.1238,
+      "step": 3644
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.563535481609097e-07,
+      "loss": 0.1241,
+      "step": 3645
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.533756447465401e-07,
+      "loss": 0.1203,
+      "step": 3646
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.504033856932924e-07,
+      "loss": 0.1244,
+      "step": 3647
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.474367728155063e-07,
+      "loss": 0.1195,
+      "step": 3648
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.444758079240699e-07,
+      "loss": 0.1216,
+      "step": 3649
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.415204928264319e-07,
+      "loss": 0.1324,
+      "step": 3650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.385708293265837e-07,
+      "loss": 0.1219,
+      "step": 3651
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.356268192250726e-07,
+      "loss": 0.1327,
+      "step": 3652
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.326884643189913e-07,
+      "loss": 0.1269,
+      "step": 3653
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.297557664019849e-07,
+      "loss": 0.1138,
+      "step": 3654
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.268287272642393e-07,
+      "loss": 0.1358,
+      "step": 3655
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.239073486924919e-07,
+      "loss": 0.1254,
+      "step": 3656
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.209916324700206e-07,
+      "loss": 0.1253,
+      "step": 3657
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.180815803766494e-07,
+      "loss": 0.1262,
+      "step": 3658
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.15177194188742e-07,
+      "loss": 0.1277,
+      "step": 3659
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.122784756792079e-07,
+      "loss": 0.1236,
+      "step": 3660
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.093854266174916e-07,
+      "loss": 0.1209,
+      "step": 3661
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.064980487695816e-07,
+      "loss": 0.1228,
+      "step": 3662
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.036163438980015e-07,
+      "loss": 0.1242,
+      "step": 3663
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.007403137618163e-07,
+      "loss": 0.1249,
+      "step": 3664
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.978699601166183e-07,
+      "loss": 0.1318,
+      "step": 3665
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.950052847145449e-07,
+      "loss": 0.1272,
+      "step": 3666
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.921462893042586e-07,
+      "loss": 0.1247,
+      "step": 3667
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.892929756309629e-07,
+      "loss": 0.1209,
+      "step": 3668
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.864453454363839e-07,
+      "loss": 0.1148,
+      "step": 3669
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.836034004587888e-07,
+      "loss": 0.1101,
+      "step": 3670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.807671424329643e-07,
+      "loss": 0.1309,
+      "step": 3671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.779365730902343e-07,
+      "loss": 0.1169,
+      "step": 3672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.751116941584423e-07,
+      "loss": 0.129,
+      "step": 3673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.722925073619646e-07,
+      "loss": 0.121,
+      "step": 3674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.69479014421699e-07,
+      "loss": 0.1261,
+      "step": 3675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.666712170550693e-07,
+      "loss": 0.1219,
+      "step": 3676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.638691169760247e-07,
+      "loss": 0.1237,
+      "step": 3677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.610727158950303e-07,
+      "loss": 0.1193,
+      "step": 3678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.582820155190795e-07,
+      "loss": 0.1256,
+      "step": 3679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.554970175516806e-07,
+      "loss": 0.1278,
+      "step": 3680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.52717723692865e-07,
+      "loss": 0.1307,
+      "step": 3681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.499441356391778e-07,
+      "loss": 0.1205,
+      "step": 3682
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.471762550836868e-07,
+      "loss": 0.1366,
+      "step": 3683
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.444140837159696e-07,
+      "loss": 0.1185,
+      "step": 3684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.416576232221239e-07,
+      "loss": 0.121,
+      "step": 3685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.389068752847571e-07,
+      "loss": 0.1267,
+      "step": 3686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.361618415829951e-07,
+      "loss": 0.1316,
+      "step": 3687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.334225237924685e-07,
+      "loss": 0.1259,
+      "step": 3688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.306889235853253e-07,
+      "loss": 0.1265,
+      "step": 3689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.279610426302185e-07,
+      "loss": 0.1185,
+      "step": 3690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.252388825923139e-07,
+      "loss": 0.1448,
+      "step": 3691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.225224451332801e-07,
+      "loss": 0.1106,
+      "step": 3692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.198117319113006e-07,
+      "loss": 0.1223,
+      "step": 3693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.171067445810553e-07,
+      "loss": 0.122,
+      "step": 3694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.144074847937376e-07,
+      "loss": 0.1326,
+      "step": 3695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.117139541970351e-07,
+      "loss": 0.1205,
+      "step": 3696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.090261544351472e-07,
+      "loss": 0.1229,
+      "step": 3697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.063440871487703e-07,
+      "loss": 0.1335,
+      "step": 3698
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.036677539751024e-07,
+      "loss": 0.1248,
+      "step": 3699
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.009971565478412e-07,
+      "loss": 0.133,
+      "step": 3700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.983322964971838e-07,
+      "loss": 0.1175,
+      "step": 3701
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.956731754498246e-07,
+      "loss": 0.128,
+      "step": 3702
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.930197950289551e-07,
+      "loss": 0.1199,
+      "step": 3703
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.903721568542609e-07,
+      "loss": 0.1237,
+      "step": 3704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.877302625419256e-07,
+      "loss": 0.1181,
+      "step": 3705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.850941137046228e-07,
+      "loss": 0.1311,
+      "step": 3706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.824637119515242e-07,
+      "loss": 0.1263,
+      "step": 3707
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.798390588882863e-07,
+      "loss": 0.1211,
+      "step": 3708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.772201561170631e-07,
+      "loss": 0.1307,
+      "step": 3709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.746070052364938e-07,
+      "loss": 0.1322,
+      "step": 3710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.719996078417078e-07,
+      "loss": 0.1164,
+      "step": 3711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.693979655243265e-07,
+      "loss": 0.1336,
+      "step": 3712
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.668020798724505e-07,
+      "loss": 0.1263,
+      "step": 3713
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.642119524706746e-07,
+      "loss": 0.1242,
+      "step": 3714
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.616275849000707e-07,
+      "loss": 0.1299,
+      "step": 3715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.590489787382015e-07,
+      "loss": 0.1351,
+      "step": 3716
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.564761355591075e-07,
+      "loss": 0.1264,
+      "step": 3717
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.539090569333183e-07,
+      "loss": 0.1187,
+      "step": 3718
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.513477444278347e-07,
+      "loss": 0.1304,
+      "step": 3719
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.487921996061474e-07,
+      "loss": 0.1167,
+      "step": 3720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.462424240282183e-07,
+      "loss": 0.123,
+      "step": 3721
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.436984192504957e-07,
+      "loss": 0.1174,
+      "step": 3722
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.411601868258986e-07,
+      "loss": 0.1191,
+      "step": 3723
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.386277283038277e-07,
+      "loss": 0.1238,
+      "step": 3724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.361010452301518e-07,
+      "loss": 0.122,
+      "step": 3725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.335801391472228e-07,
+      "loss": 0.1276,
+      "step": 3726
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.310650115938598e-07,
+      "loss": 0.1265,
+      "step": 3727
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.28555664105359e-07,
+      "loss": 0.1241,
+      "step": 3728
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.260520982134831e-07,
+      "loss": 0.1208,
+      "step": 3729
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.235543154464728e-07,
+      "loss": 0.1227,
+      "step": 3730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.210623173290308e-07,
+      "loss": 0.1129,
+      "step": 3731
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.185761053823357e-07,
+      "loss": 0.133,
+      "step": 3732
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.160956811240269e-07,
+      "loss": 0.1274,
+      "step": 3733
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.136210460682201e-07,
+      "loss": 0.1257,
+      "step": 3734
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.111522017254866e-07,
+      "loss": 0.1207,
+      "step": 3735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.086891496028723e-07,
+      "loss": 0.1249,
+      "step": 3736
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0623189120388e-07,
+      "loss": 0.1207,
+      "step": 3737
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.037804280284808e-07,
+      "loss": 0.1239,
+      "step": 3738
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.013347615731068e-07,
+      "loss": 0.1272,
+      "step": 3739
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.988948933306515e-07,
+      "loss": 0.1242,
+      "step": 3740
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.964608247904667e-07,
+      "loss": 0.1255,
+      "step": 3741
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940325574383697e-07,
+      "loss": 0.1143,
+      "step": 3742
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.916100927566292e-07,
+      "loss": 0.1231,
+      "step": 3743
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.891934322239788e-07,
+      "loss": 0.1315,
+      "step": 3744
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.867825773156065e-07,
+      "loss": 0.1223,
+      "step": 3745
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.843775295031516e-07,
+      "loss": 0.1183,
+      "step": 3746
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.819782902547187e-07,
+      "loss": 0.123,
+      "step": 3747
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.795848610348553e-07,
+      "loss": 0.1249,
+      "step": 3748
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.771972433045735e-07,
+      "loss": 0.1288,
+      "step": 3749
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7481543852132663e-07,
+      "loss": 0.1201,
+      "step": 3750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.724394481390293e-07,
+      "loss": 0.1306,
+      "step": 3751
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7006927360804253e-07,
+      "loss": 0.1188,
+      "step": 3752
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.677049163751768e-07,
+      "loss": 0.1187,
+      "step": 3753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6534637788369196e-07,
+      "loss": 0.1169,
+      "step": 3754
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6299365957329866e-07,
+      "loss": 0.1268,
+      "step": 3755
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.606467628801503e-07,
+      "loss": 0.134,
+      "step": 3756
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.583056892368509e-07,
+      "loss": 0.1259,
+      "step": 3757
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5597044007244517e-07,
+      "loss": 0.1254,
+      "step": 3758
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5364101681242833e-07,
+      "loss": 0.1266,
+      "step": 3759
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.51317420878733e-07,
+      "loss": 0.1301,
+      "step": 3760
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4899965368974006e-07,
+      "loss": 0.1197,
+      "step": 3761
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.46687716660269e-07,
+      "loss": 0.1221,
+      "step": 3762
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.443816112015831e-07,
+      "loss": 0.1258,
+      "step": 3763
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4208133872138184e-07,
+      "loss": 0.1234,
+      "step": 3764
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.397869006238098e-07,
+      "loss": 0.125,
+      "step": 3765
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3749829830944444e-07,
+      "loss": 0.122,
+      "step": 3766
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3521553317530495e-07,
+      "loss": 0.1197,
+      "step": 3767
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.329386066148444e-07,
+      "loss": 0.1239,
+      "step": 3768
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3066752001795554e-07,
+      "loss": 0.1323,
+      "step": 3769
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2840227477096154e-07,
+      "loss": 0.1192,
+      "step": 3770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.261428722566241e-07,
+      "loss": 0.125,
+      "step": 3771
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2388931385413555e-07,
+      "loss": 0.1219,
+      "step": 3772
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.216416009391244e-07,
+      "loss": 0.1161,
+      "step": 3773
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1939973488364536e-07,
+      "loss": 0.1223,
+      "step": 3774
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.171637170561893e-07,
+      "loss": 0.1273,
+      "step": 3775
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1493354882167323e-07,
+      "loss": 0.1214,
+      "step": 3776
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1270923154144713e-07,
+      "loss": 0.1263,
+      "step": 3777
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.10490766573286e-07,
+      "loss": 0.1326,
+      "step": 3778
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.082781552713955e-07,
+      "loss": 0.1225,
+      "step": 3779
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0607139898640515e-07,
+      "loss": 0.1273,
+      "step": 3780
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0387049906537415e-07,
+      "loss": 0.1229,
+      "step": 3781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.016754568517822e-07,
+      "loss": 0.122,
+      "step": 3782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.994862736855376e-07,
+      "loss": 0.1331,
+      "step": 3783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.97302950902968e-07,
+      "loss": 0.126,
+      "step": 3784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9512548983682844e-07,
+      "loss": 0.1242,
+      "step": 3785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9295389181629006e-07,
+      "loss": 0.138,
+      "step": 3786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.907881581669526e-07,
+      "loss": 0.1221,
+      "step": 3787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.886282902108274e-07,
+      "loss": 0.1267,
+      "step": 3788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.86474289266352e-07,
+      "loss": 0.1261,
+      "step": 3789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8432615664838136e-07,
+      "loss": 0.1161,
+      "step": 3790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.821838936681854e-07,
+      "loss": 0.1287,
+      "step": 3791
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.800475016334526e-07,
+      "loss": 0.1259,
+      "step": 3792
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.779169818482886e-07,
+      "loss": 0.1212,
+      "step": 3793
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7579233561321427e-07,
+      "loss": 0.1171,
+      "step": 3794
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.736735642251621e-07,
+      "loss": 0.1355,
+      "step": 3795
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7156066897748424e-07,
+      "loss": 0.1189,
+      "step": 3796
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.694536511599389e-07,
+      "loss": 0.127,
+      "step": 3797
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.673525120587051e-07,
+      "loss": 0.1346,
+      "step": 3798
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.652572529563636e-07,
+      "loss": 0.1226,
+      "step": 3799
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6316787513191454e-07,
+      "loss": 0.1248,
+      "step": 3800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.610843798607611e-07,
+      "loss": 0.1095,
+      "step": 3801
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5900676841472137e-07,
+      "loss": 0.1168,
+      "step": 3802
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5693504206201656e-07,
+      "loss": 0.1233,
+      "step": 3803
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.548692020672817e-07,
+      "loss": 0.1338,
+      "step": 3804
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.528092496915514e-07,
+      "loss": 0.1308,
+      "step": 3805
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.50755186192272e-07,
+      "loss": 0.1153,
+      "step": 3806
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.487070128232917e-07,
+      "loss": 0.1237,
+      "step": 3807
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.46664730834867e-07,
+      "loss": 0.1157,
+      "step": 3808
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4462834147365177e-07,
+      "loss": 0.126,
+      "step": 3809
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.425978459827117e-07,
+      "loss": 0.1245,
+      "step": 3810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.405732456015054e-07,
+      "loss": 0.1183,
+      "step": 3811
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3855454156590195e-07,
+      "loss": 0.1157,
+      "step": 3812
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.365417351081635e-07,
+      "loss": 0.1285,
+      "step": 3813
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3453482745695596e-07,
+      "loss": 0.1249,
+      "step": 3814
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.32533819837344e-07,
+      "loss": 0.115,
+      "step": 3815
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.305387134707916e-07,
+      "loss": 0.1344,
+      "step": 3816
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2854950957515785e-07,
+      "loss": 0.1277,
+      "step": 3817
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2656620936470153e-07,
+      "loss": 0.1223,
+      "step": 3818
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.245888140500764e-07,
+      "loss": 0.1287,
+      "step": 3819
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2261732483833243e-07,
+      "loss": 0.1201,
+      "step": 3820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2065174293291147e-07,
+      "loss": 0.1286,
+      "step": 3821
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1869206953365484e-07,
+      "loss": 0.113,
+      "step": 3822
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1673830583679344e-07,
+      "loss": 0.125,
+      "step": 3823
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1479045303494993e-07,
+      "loss": 0.1247,
+      "step": 3824
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.128485123171432e-07,
+      "loss": 0.1354,
+      "step": 3825
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.109124848687772e-07,
+      "loss": 0.1276,
+      "step": 3826
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0898237187165225e-07,
+      "loss": 0.1256,
+      "step": 3827
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0705817450395357e-07,
+      "loss": 0.1201,
+      "step": 3828
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051398939402583e-07,
+      "loss": 0.122,
+      "step": 3829
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0322753135152983e-07,
+      "loss": 0.1265,
+      "step": 3830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0132108790512207e-07,
+      "loss": 0.1325,
+      "step": 3831
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9942056476477186e-07,
+      "loss": 0.1321,
+      "step": 3832
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9752596309060575e-07,
+      "loss": 0.123,
+      "step": 3833
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9563728403913193e-07,
+      "loss": 0.1254,
+      "step": 3834
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9375452876324816e-07,
+      "loss": 0.1175,
+      "step": 3835
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9187769841223066e-07,
+      "loss": 0.1204,
+      "step": 3836
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9000679413174306e-07,
+      "loss": 0.1349,
+      "step": 3837
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.881418170638284e-07,
+      "loss": 0.1141,
+      "step": 3838
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8628276834691714e-07,
+      "loss": 0.1275,
+      "step": 3839
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.844296491158127e-07,
+      "loss": 0.1322,
+      "step": 3840
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.82582460501708e-07,
+      "loss": 0.118,
+      "step": 3841
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.807412036321677e-07,
+      "loss": 0.1236,
+      "step": 3842
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.789058796311406e-07,
+      "loss": 0.1306,
+      "step": 3843
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.770764896189515e-07,
+      "loss": 0.122,
+      "step": 3844
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7525303471230614e-07,
+      "loss": 0.1251,
+      "step": 3845
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7343551602428187e-07,
+      "loss": 0.1276,
+      "step": 3846
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7162393466433903e-07,
+      "loss": 0.1218,
+      "step": 3847
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6981829173830744e-07,
+      "loss": 0.1179,
+      "step": 3848
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6801858834839656e-07,
+      "loss": 0.1274,
+      "step": 3849
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6622482559318765e-07,
+      "loss": 0.115,
+      "step": 3850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.64437004567637e-07,
+      "loss": 0.1195,
+      "step": 3851
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.626551263630728e-07,
+      "loss": 0.13,
+      "step": 3852
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6087919206719716e-07,
+      "loss": 0.1224,
+      "step": 3853
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.591092027640818e-07,
+      "loss": 0.1296,
+      "step": 3854
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.573451595341714e-07,
+      "loss": 0.1208,
+      "step": 3855
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.555870634542801e-07,
+      "loss": 0.1179,
+      "step": 3856
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.538349155975917e-07,
+      "loss": 0.124,
+      "step": 3857
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5208871703365943e-07,
+      "loss": 0.1178,
+      "step": 3858
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.503484688284041e-07,
+      "loss": 0.1225,
+      "step": 3859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4861417204411707e-07,
+      "loss": 0.1246,
+      "step": 3860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4688582773945146e-07,
+      "loss": 0.1373,
+      "step": 3861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4516343696943226e-07,
+      "loss": 0.143,
+      "step": 3862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4344700078544724e-07,
+      "loss": 0.1274,
+      "step": 3863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.417365202352506e-07,
+      "loss": 0.1321,
+      "step": 3864
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.400319963629605e-07,
+      "loss": 0.1179,
+      "step": 3865
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3833343020905897e-07,
+      "loss": 0.1312,
+      "step": 3866
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3664082281039115e-07,
+      "loss": 0.1104,
+      "step": 3867
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3495417520016606e-07,
+      "loss": 0.1167,
+      "step": 3868
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3327348840795348e-07,
+      "loss": 0.1303,
+      "step": 3869
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3159876345968613e-07,
+      "loss": 0.1306,
+      "step": 3870
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2993000137765287e-07,
+      "loss": 0.1198,
+      "step": 3871
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2826720318050998e-07,
+      "loss": 0.1235,
+      "step": 3872
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.266103698832667e-07,
+      "loss": 0.1233,
+      "step": 3873
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2495950249729505e-07,
+      "loss": 0.1275,
+      "step": 3874
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.233146020303234e-07,
+      "loss": 0.1221,
+      "step": 3875
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2167566948643972e-07,
+      "loss": 0.1255,
+      "step": 3876
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2004270586608478e-07,
+      "loss": 0.128,
+      "step": 3877
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.184157121660624e-07,
+      "loss": 0.1156,
+      "step": 3878
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1679468937952475e-07,
+      "loss": 0.1196,
+      "step": 3879
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.15179638495987e-07,
+      "loss": 0.1217,
+      "step": 3880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1357056050131165e-07,
+      "loss": 0.1312,
+      "step": 3881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1196745637772076e-07,
+      "loss": 0.1204,
+      "step": 3882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.103703271037849e-07,
+      "loss": 0.127,
+      "step": 3883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0877917365443313e-07,
+      "loss": 0.1281,
+      "step": 3884
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0719399700094067e-07,
+      "loss": 0.1357,
+      "step": 3885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.056147981109402e-07,
+      "loss": 0.1198,
+      "step": 3886
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.040415779484106e-07,
+      "loss": 0.1246,
+      "step": 3887
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0247433747368372e-07,
+      "loss": 0.1326,
+      "step": 3888
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.009130776434409e-07,
+      "loss": 0.118,
+      "step": 3889
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.993577994107121e-07,
+      "loss": 0.1264,
+      "step": 3890
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.978085037248767e-07,
+      "loss": 0.1159,
+      "step": 3891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9626519153166156e-07,
+      "loss": 0.1245,
+      "step": 3892
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9472786377314311e-07,
+      "loss": 0.1138,
+      "step": 3893
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.931965213877407e-07,
+      "loss": 0.1203,
+      "step": 3894
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9167116531022435e-07,
+      "loss": 0.1089,
+      "step": 3895
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.901517964717059e-07,
+      "loss": 0.1231,
+      "step": 3896
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.886384157996457e-07,
+      "loss": 0.1226,
+      "step": 3897
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8713102421784924e-07,
+      "loss": 0.1204,
+      "step": 3898
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8562962264646156e-07,
+      "loss": 0.1209,
+      "step": 3899
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8413421200197512e-07,
+      "loss": 0.1206,
+      "step": 3900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.826447931972253e-07,
+      "loss": 0.1232,
+      "step": 3901
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.811613671413881e-07,
+      "loss": 0.1266,
+      "step": 3902
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.796839347399837e-07,
+      "loss": 0.1202,
+      "step": 3903
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7821249689487062e-07,
+      "loss": 0.1207,
+      "step": 3904
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7674705450425044e-07,
+      "loss": 0.124,
+      "step": 3905
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7528760846266423e-07,
+      "loss": 0.1278,
+      "step": 3906
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.738341596609927e-07,
+      "loss": 0.1218,
+      "step": 3907
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7238670898645615e-07,
+      "loss": 0.1165,
+      "step": 3908
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7094525732261336e-07,
+      "loss": 0.1194,
+      "step": 3909
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6950980554936047e-07,
+      "loss": 0.119,
+      "step": 3910
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6808035454293215e-07,
+      "loss": 0.1275,
+      "step": 3911
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6665690517589818e-07,
+      "loss": 0.119,
+      "step": 3912
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6523945831716792e-07,
+      "loss": 0.1307,
+      "step": 3913
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6382801483198374e-07,
+      "loss": 0.1208,
+      "step": 3914
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6242257558192642e-07,
+      "loss": 0.1325,
+      "step": 3915
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6102314142490638e-07,
+      "loss": 0.1246,
+      "step": 3916
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5962971321517583e-07,
+      "loss": 0.1194,
+      "step": 3917
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.582422918033133e-07,
+      "loss": 0.1307,
+      "step": 3918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5686087803623572e-07,
+      "loss": 0.1212,
+      "step": 3919
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5548547275719085e-07,
+      "loss": 0.1292,
+      "step": 3920
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5411607680575924e-07,
+      "loss": 0.1195,
+      "step": 3921
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.527526910178523e-07,
+      "loss": 0.1254,
+      "step": 3922
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5139531622571424e-07,
+      "loss": 0.122,
+      "step": 3923
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5004395325792008e-07,
+      "loss": 0.1237,
+      "step": 3924
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4869860293937112e-07,
+      "loss": 0.1211,
+      "step": 3925
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.473592660913037e-07,
+      "loss": 0.1219,
+      "step": 3926
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4602594353128164e-07,
+      "loss": 0.1256,
+      "step": 3927
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.44698636073195e-07,
+      "loss": 0.1226,
+      "step": 3928
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.433773445272657e-07,
+      "loss": 0.1326,
+      "step": 3929
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4206206970004076e-07,
+      "loss": 0.125,
+      "step": 3930
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4075281239439685e-07,
+      "loss": 0.1157,
+      "step": 3931
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.394495734095347e-07,
+      "loss": 0.1219,
+      "step": 3932
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3815235354098233e-07,
+      "loss": 0.1222,
+      "step": 3933
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3686115358059415e-07,
+      "loss": 0.1183,
+      "step": 3934
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3557597431655078e-07,
+      "loss": 0.1292,
+      "step": 3935
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.342968165333558e-07,
+      "loss": 0.1262,
+      "step": 3936
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3302368101183683e-07,
+      "loss": 0.1146,
+      "step": 3937
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3175656852914664e-07,
+      "loss": 0.1279,
+      "step": 3938
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.304954798587621e-07,
+      "loss": 0.1241,
+      "step": 3939
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2924041577048185e-07,
+      "loss": 0.1351,
+      "step": 3940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.279913770304253e-07,
+      "loss": 0.1176,
+      "step": 3941
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.26748364401037e-07,
+      "loss": 0.1244,
+      "step": 3942
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.255113786410822e-07,
+      "loss": 0.1356,
+      "step": 3943
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.242804205056458e-07,
+      "loss": 0.1357,
+      "step": 3944
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2305549074613342e-07,
+      "loss": 0.1264,
+      "step": 3945
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.218365901102736e-07,
+      "loss": 0.1217,
+      "step": 3946
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2062371934211004e-07,
+      "loss": 0.1159,
+      "step": 3947
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.194168791820105e-07,
+      "loss": 0.1217,
+      "step": 3948
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1821607036665794e-07,
+      "loss": 0.1152,
+      "step": 3949
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.170212936290549e-07,
+      "loss": 0.1287,
+      "step": 3950
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1583254969852132e-07,
+      "loss": 0.1495,
+      "step": 3951
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1464983930069562e-07,
+      "loss": 0.1322,
+      "step": 3952
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1347316315753255e-07,
+      "loss": 0.1196,
+      "step": 3953
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1230252198730307e-07,
+      "loss": 0.1345,
+      "step": 3954
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1113791650459449e-07,
+      "loss": 0.1233,
+      "step": 3955
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0997934742031147e-07,
+      "loss": 0.1248,
+      "step": 3956
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.088268154416694e-07,
+      "loss": 0.1302,
+      "step": 3957
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.076803212722044e-07,
+      "loss": 0.1325,
+      "step": 3958
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0653986561176221e-07,
+      "loss": 0.1227,
+      "step": 3959
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0540544915650597e-07,
+      "loss": 0.1176,
+      "step": 3960
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0427707259890841e-07,
+      "loss": 0.1203,
+      "step": 3961
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0315473662775855e-07,
+      "loss": 0.1286,
+      "step": 3962
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0203844192815726e-07,
+      "loss": 0.1109,
+      "step": 3963
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0092818918151948e-07,
+      "loss": 0.1237,
+      "step": 3964
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.98239790655664e-08,
+      "loss": 0.1211,
+      "step": 3965
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.872581225433664e-08,
+      "loss": 0.1299,
+      "step": 3966
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.76336894181773e-08,
+      "loss": 0.1271,
+      "step": 3967
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.654761122374623e-08,
+      "loss": 0.1303,
+      "step": 3968
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.546757833401088e-08,
+      "loss": 0.1371,
+      "step": 3969
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.439359140825055e-08,
+      "loss": 0.1129,
+      "step": 3970
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.332565110205193e-08,
+      "loss": 0.1274,
+      "step": 3971
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.226375806731136e-08,
+      "loss": 0.1335,
+      "step": 3972
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.120791295223364e-08,
+      "loss": 0.1226,
+      "step": 3973
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.01581164013321e-08,
+      "loss": 0.1211,
+      "step": 3974
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.911436905542747e-08,
+      "loss": 0.1189,
+      "step": 3975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.807667155164901e-08,
+      "loss": 0.1339,
+      "step": 3976
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.704502452343111e-08,
+      "loss": 0.1277,
+      "step": 3977
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.60194286005156e-08,
+      "loss": 0.1312,
+      "step": 3978
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.49998844089528e-08,
+      "loss": 0.1293,
+      "step": 3979
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.398639257109487e-08,
+      "loss": 0.1265,
+      "step": 3980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.29789537056025e-08,
+      "loss": 0.1226,
+      "step": 3981
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.197756842743931e-08,
+      "loss": 0.1089,
+      "step": 3982
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.098223734787635e-08,
+      "loss": 0.1292,
+      "step": 3983
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.999296107448761e-08,
+      "loss": 0.1235,
+      "step": 3984
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.900974021114893e-08,
+      "loss": 0.1299,
+      "step": 3985
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.803257535804465e-08,
+      "loss": 0.1301,
+      "step": 3986
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.706146711165873e-08,
+      "loss": 0.1292,
+      "step": 3987
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.609641606477703e-08,
+      "loss": 0.1212,
+      "step": 3988
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.513742280649272e-08,
+      "loss": 0.1133,
+      "step": 3989
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.418448792219646e-08,
+      "loss": 0.1283,
+      "step": 3990
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.323761199358404e-08,
+      "loss": 0.1316,
+      "step": 3991
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.229679559864866e-08,
+      "loss": 0.1143,
+      "step": 3992
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.136203931168983e-08,
+      "loss": 0.1313,
+      "step": 3993
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.043334370330335e-08,
+      "loss": 0.1168,
+      "step": 3994
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.951070934038795e-08,
+      "loss": 0.1249,
+      "step": 3995
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.859413678614201e-08,
+      "loss": 0.1221,
+      "step": 3996
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.768362660006356e-08,
+      "loss": 0.1197,
+      "step": 3997
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.677917933794908e-08,
+      "loss": 0.1228,
+      "step": 3998
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.588079555189586e-08,
+      "loss": 0.1205,
+      "step": 3999
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.498847579029744e-08,
+      "loss": 0.1257,
+      "step": 4000
+    },
+    {
+      "epoch": 2.89,
+      "eval_loss": 0.12249945104122162,
+      "eval_runtime": 715.8846,
+      "eval_samples_per_second": 69.844,
+      "eval_steps_per_second": 2.183,
+      "step": 4000
+    }
+  ],
+  "max_steps": 4146,
+  "num_train_epochs": 3,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}