diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35499 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 5913,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.756756756756757e-08,
+      "loss": 1.7898,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-07,
+      "loss": 1.8447,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0270270270270273e-07,
+      "loss": 1.7463,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 1.7783,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.378378378378379e-07,
+      "loss": 1.7857,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540546e-07,
+      "loss": 1.7453,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7297297297297305e-07,
+      "loss": 1.7872,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 1.6942,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.081081081081082e-07,
+      "loss": 1.7741,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.756756756756758e-07,
+      "loss": 1.7713,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.432432432432434e-07,
+      "loss": 1.8188,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 1.7388,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.783783783783785e-07,
+      "loss": 1.7138,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.459459459459461e-07,
+      "loss": 1.6154,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0135135135135136e-06,
+      "loss": 1.7073,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.7068,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.148648648648649e-06,
+      "loss": 1.5564,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2162162162162164e-06,
+      "loss": 1.5045,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.2837837837837838e-06,
+      "loss": 1.5459,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 1.5248,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.418918918918919e-06,
+      "loss": 1.4911,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4864864864864868e-06,
+      "loss": 1.4677,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.5540540540540541e-06,
+      "loss": 1.4029,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.3354,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6891891891891894e-06,
+      "loss": 1.2673,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.756756756756757e-06,
+      "loss": 1.1685,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8243243243243245e-06,
+      "loss": 1.1968,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 1.1691,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9594594594594595e-06,
+      "loss": 1.1353,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0270270270270273e-06,
+      "loss": 1.1432,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.0945945945945946e-06,
+      "loss": 1.0809,
+      "step": 31
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.0808,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.22972972972973e-06,
+      "loss": 1.0946,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.297297297297298e-06,
+      "loss": 0.9575,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.364864864864865e-06,
+      "loss": 1.0004,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 0.9515,
+      "step": 36
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5e-06,
+      "loss": 0.9952,
+      "step": 37
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.5675675675675675e-06,
+      "loss": 0.9819,
+      "step": 38
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.6351351351351353e-06,
+      "loss": 0.918,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.9437,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.7702702702702703e-06,
+      "loss": 0.9362,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.837837837837838e-06,
+      "loss": 0.921,
+      "step": 42
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9054054054054054e-06,
+      "loss": 0.8952,
+      "step": 43
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 0.9011,
+      "step": 44
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.040540540540541e-06,
+      "loss": 0.9196,
+      "step": 45
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1081081081081082e-06,
+      "loss": 0.8413,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.1756756756756755e-06,
+      "loss": 0.8664,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 0.8251,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.310810810810811e-06,
+      "loss": 0.7825,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.3783783783783788e-06,
+      "loss": 0.8964,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.445945945945946e-06,
+      "loss": 0.8164,
+      "step": 51
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 0.857,
+      "step": 52
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.5810810810810816e-06,
+      "loss": 0.8079,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.648648648648649e-06,
+      "loss": 0.8284,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.7162162162162162e-06,
+      "loss": 0.8172,
+      "step": 55
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.8114,
+      "step": 56
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.851351351351352e-06,
+      "loss": 0.6927,
+      "step": 57
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.918918918918919e-06,
+      "loss": 0.7675,
+      "step": 58
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.986486486486487e-06,
+      "loss": 0.8073,
+      "step": 59
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 0.8386,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.121621621621622e-06,
+      "loss": 0.824,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.189189189189189e-06,
+      "loss": 0.7604,
+      "step": 62
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.256756756756757e-06,
+      "loss": 0.7001,
+      "step": 63
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 0.7201,
+      "step": 64
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.391891891891892e-06,
+      "loss": 0.7453,
+      "step": 65
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.45945945945946e-06,
+      "loss": 0.741,
+      "step": 66
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.527027027027027e-06,
+      "loss": 0.7596,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 0.7203,
+      "step": 68
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.6621621621621625e-06,
+      "loss": 0.7006,
+      "step": 69
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.72972972972973e-06,
+      "loss": 0.7067,
+      "step": 70
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.797297297297297e-06,
+      "loss": 0.7625,
+      "step": 71
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.74,
+      "step": 72
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.932432432432433e-06,
+      "loss": 0.6893,
+      "step": 73
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5e-06,
+      "loss": 0.7566,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.067567567567568e-06,
+      "loss": 0.6891,
+      "step": 75
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 0.7002,
+      "step": 76
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.202702702702704e-06,
+      "loss": 0.7115,
+      "step": 77
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.2702702702702705e-06,
+      "loss": 0.7362,
+      "step": 78
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.337837837837838e-06,
+      "loss": 0.6205,
+      "step": 79
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.7121,
+      "step": 80
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.472972972972973e-06,
+      "loss": 0.7612,
+      "step": 81
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.540540540540541e-06,
+      "loss": 0.6909,
+      "step": 82
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.608108108108109e-06,
+      "loss": 0.7354,
+      "step": 83
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 0.7116,
+      "step": 84
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.743243243243244e-06,
+      "loss": 0.7095,
+      "step": 85
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.810810810810811e-06,
+      "loss": 0.6904,
+      "step": 86
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.8783783783783786e-06,
+      "loss": 0.6747,
+      "step": 87
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.6878,
+      "step": 88
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.013513513513514e-06,
+      "loss": 0.6787,
+      "step": 89
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.081081081081082e-06,
+      "loss": 0.648,
+      "step": 90
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.1486486486486495e-06,
+      "loss": 0.6925,
+      "step": 91
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 0.6525,
+      "step": 92
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.283783783783784e-06,
+      "loss": 0.6501,
+      "step": 93
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.351351351351351e-06,
+      "loss": 0.6105,
+      "step": 94
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.41891891891892e-06,
+      "loss": 0.6182,
+      "step": 95
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 0.659,
+      "step": 96
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.554054054054054e-06,
+      "loss": 0.6242,
+      "step": 97
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.621621621621622e-06,
+      "loss": 0.6388,
+      "step": 98
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.689189189189191e-06,
+      "loss": 0.6316,
+      "step": 99
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 0.6343,
+      "step": 100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.824324324324325e-06,
+      "loss": 0.6156,
+      "step": 101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.891891891891892e-06,
+      "loss": 0.6936,
+      "step": 102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.95945945945946e-06,
+      "loss": 0.6887,
+      "step": 103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 0.6216,
+      "step": 104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.0945945945945946e-06,
+      "loss": 0.6424,
+      "step": 105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.162162162162163e-06,
+      "loss": 0.6465,
+      "step": 106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.229729729729731e-06,
+      "loss": 0.7082,
+      "step": 107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 0.6706,
+      "step": 108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.3648648648648655e-06,
+      "loss": 0.5883,
+      "step": 109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.4324324324324324e-06,
+      "loss": 0.671,
+      "step": 110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.6206,
+      "step": 111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 0.6255,
+      "step": 112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.635135135135135e-06,
+      "loss": 0.6089,
+      "step": 113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.702702702702704e-06,
+      "loss": 0.7344,
+      "step": 114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.77027027027027e-06,
+      "loss": 0.5563,
+      "step": 115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 0.6589,
+      "step": 116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.905405405405406e-06,
+      "loss": 0.646,
+      "step": 117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.972972972972974e-06,
+      "loss": 0.5854,
+      "step": 118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.040540540540541e-06,
+      "loss": 0.6309,
+      "step": 119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.6569,
+      "step": 120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 8.175675675675677e-06,
+      "loss": 0.6585,
+      "step": 121
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.243243243243245e-06,
+      "loss": 0.5931,
+      "step": 122
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.31081081081081e-06,
+      "loss": 0.6001,
+      "step": 123
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 0.5866,
+      "step": 124
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.445945945945948e-06,
+      "loss": 0.683,
+      "step": 125
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.513513513513514e-06,
+      "loss": 0.703,
+      "step": 126
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.581081081081082e-06,
+      "loss": 0.6212,
+      "step": 127
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.623,
+      "step": 128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.716216216216217e-06,
+      "loss": 0.5536,
+      "step": 129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.783783783783785e-06,
+      "loss": 0.634,
+      "step": 130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.851351351351351e-06,
+      "loss": 0.6622,
+      "step": 131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 0.6623,
+      "step": 132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.986486486486488e-06,
+      "loss": 0.5932,
+      "step": 133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.054054054054054e-06,
+      "loss": 0.5911,
+      "step": 134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.121621621621622e-06,
+      "loss": 0.6833,
+      "step": 135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 0.6364,
+      "step": 136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.256756756756757e-06,
+      "loss": 0.6273,
+      "step": 137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.324324324324325e-06,
+      "loss": 0.5768,
+      "step": 138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391891891891893e-06,
+      "loss": 0.6211,
+      "step": 139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 0.6356,
+      "step": 140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.527027027027028e-06,
+      "loss": 0.5648,
+      "step": 141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.594594594594594e-06,
+      "loss": 0.5981,
+      "step": 142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.662162162162164e-06,
+      "loss": 0.6102,
+      "step": 143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.6231,
+      "step": 144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.797297297297298e-06,
+      "loss": 0.6897,
+      "step": 145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.864864864864865e-06,
+      "loss": 0.6056,
+      "step": 146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.932432432432433e-06,
+      "loss": 0.6466,
+      "step": 147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1e-05,
+      "loss": 0.6224,
+      "step": 148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0067567567567569e-05,
+      "loss": 0.6256,
+      "step": 149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0135135135135136e-05,
+      "loss": 0.5564,
+      "step": 150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0202702702702702e-05,
+      "loss": 0.5466,
+      "step": 151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 0.5578,
+      "step": 152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0337837837837838e-05,
+      "loss": 0.6381,
+      "step": 153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.0405405405405407e-05,
+      "loss": 0.6441,
+      "step": 154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0472972972972975e-05,
+      "loss": 0.6271,
+      "step": 155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 0.5801,
+      "step": 156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0608108108108109e-05,
+      "loss": 0.6178,
+      "step": 157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0675675675675677e-05,
+      "loss": 0.5268,
+      "step": 158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0743243243243244e-05,
+      "loss": 0.613,
+      "step": 159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.5878,
+      "step": 160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.087837837837838e-05,
+      "loss": 0.6356,
+      "step": 161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0945945945945946e-05,
+      "loss": 0.6726,
+      "step": 162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1013513513513514e-05,
+      "loss": 0.64,
+      "step": 163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 0.6254,
+      "step": 164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.114864864864865e-05,
+      "loss": 0.6436,
+      "step": 165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1216216216216219e-05,
+      "loss": 0.5707,
+      "step": 166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.1283783783783785e-05,
+      "loss": 0.6057,
+      "step": 167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 0.615,
+      "step": 168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.141891891891892e-05,
+      "loss": 0.5559,
+      "step": 169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1486486486486488e-05,
+      "loss": 0.6225,
+      "step": 170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1554054054054056e-05,
+      "loss": 0.6817,
+      "step": 171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 0.6579,
+      "step": 172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.168918918918919e-05,
+      "loss": 0.621,
+      "step": 173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.1756756756756757e-05,
+      "loss": 0.6153,
+      "step": 174
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1824324324324325e-05,
+      "loss": 0.6369,
+      "step": 175
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.5602,
+      "step": 176
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1959459459459462e-05,
+      "loss": 0.6182,
+      "step": 177
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2027027027027028e-05,
+      "loss": 0.5802,
+      "step": 178
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2094594594594596e-05,
+      "loss": 0.584,
+      "step": 179
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 0.5996,
+      "step": 180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2229729729729731e-05,
+      "loss": 0.6033,
+      "step": 181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2297297297297299e-05,
+      "loss": 0.5809,
+      "step": 182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2364864864864865e-05,
+      "loss": 0.5749,
+      "step": 183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.681,
+      "step": 184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.25e-05,
+      "loss": 0.5948,
+      "step": 185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2567567567567568e-05,
+      "loss": 0.5452,
+      "step": 186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.2635135135135138e-05,
+      "loss": 0.5256,
+      "step": 187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 0.5869,
+      "step": 188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2770270270270272e-05,
+      "loss": 0.5765,
+      "step": 189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.283783783783784e-05,
+      "loss": 0.6452,
+      "step": 190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2905405405405407e-05,
+      "loss": 0.6267,
+      "step": 191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.61,
+      "step": 192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.3040540540540541e-05,
+      "loss": 0.6087,
+      "step": 193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3108108108108109e-05,
+      "loss": 0.627,
+      "step": 194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3175675675675676e-05,
+      "loss": 0.6142,
+      "step": 195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 0.5681,
+      "step": 196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3310810810810812e-05,
+      "loss": 0.5719,
+      "step": 197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3378378378378381e-05,
+      "loss": 0.6442,
+      "step": 198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3445945945945946e-05,
+      "loss": 0.605,
+      "step": 199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.6069,
+      "step": 200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3581081081081083e-05,
+      "loss": 0.6583,
+      "step": 201
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.364864864864865e-05,
+      "loss": 0.6061,
+      "step": 202
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3716216216216218e-05,
+      "loss": 0.6476,
+      "step": 203
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 0.5404,
+      "step": 204
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.3851351351351352e-05,
+      "loss": 0.6033,
+      "step": 205
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.391891891891892e-05,
+      "loss": 0.6076,
+      "step": 206
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3986486486486488e-05,
+      "loss": 0.6162,
+      "step": 207
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.5755,
+      "step": 208
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4121621621621621e-05,
+      "loss": 0.5501,
+      "step": 209
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4189189189189189e-05,
+      "loss": 0.5638,
+      "step": 210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4256756756756759e-05,
+      "loss": 0.5164,
+      "step": 211
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 0.5869,
+      "step": 212
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4391891891891894e-05,
+      "loss": 0.5655,
+      "step": 213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4459459459459462e-05,
+      "loss": 0.6264,
+      "step": 214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4527027027027028e-05,
+      "loss": 0.595,
+      "step": 215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.5769,
+      "step": 216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4662162162162163e-05,
+      "loss": 0.5922,
+      "step": 217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4729729729729731e-05,
+      "loss": 0.5418,
+      "step": 218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4797297297297299e-05,
+      "loss": 0.5421,
+      "step": 219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 0.5374,
+      "step": 220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4932432432432433e-05,
+      "loss": 0.5465,
+      "step": 221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5934,
+      "step": 222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.506756756756757e-05,
+      "loss": 0.5955,
+      "step": 223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.5019,
+      "step": 224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5202702702702704e-05,
+      "loss": 0.5591,
+      "step": 225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527027027027027e-05,
+      "loss": 0.5576,
+      "step": 226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5337837837837837e-05,
+      "loss": 0.5685,
+      "step": 227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 0.548,
+      "step": 228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5472972972972976e-05,
+      "loss": 0.647,
+      "step": 229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.554054054054054e-05,
+      "loss": 0.637,
+      "step": 230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.560810810810811e-05,
+      "loss": 0.5854,
+      "step": 231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.578,
+      "step": 232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5743243243243244e-05,
+      "loss": 0.547,
+      "step": 233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.581081081081081e-05,
+      "loss": 0.5788,
+      "step": 234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.587837837837838e-05,
+      "loss": 0.5531,
+      "step": 235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.5773,
+      "step": 236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6013513513513515e-05,
+      "loss": 0.605,
+      "step": 237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6081081081081083e-05,
+      "loss": 0.5939,
+      "step": 238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.614864864864865e-05,
+      "loss": 0.5844,
+      "step": 239
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.5998,
+      "step": 240
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6283783783783786e-05,
+      "loss": 0.6386,
+      "step": 241
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6351351351351354e-05,
+      "loss": 0.5286,
+      "step": 242
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641891891891892e-05,
+      "loss": 0.5633,
+      "step": 243
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 0.5837,
+      "step": 244
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6554054054054057e-05,
+      "loss": 0.5342,
+      "step": 245
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662162162162162e-05,
+      "loss": 0.612,
+      "step": 246
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.668918918918919e-05,
+      "loss": 0.569,
+      "step": 247
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.5616,
+      "step": 248
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6824324324324324e-05,
+      "loss": 0.5335,
+      "step": 249
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6891891891891896e-05,
+      "loss": 0.5681,
+      "step": 250
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6959459459459463e-05,
+      "loss": 0.5847,
+      "step": 251
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 0.5606,
+      "step": 252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7094594594594595e-05,
+      "loss": 0.5943,
+      "step": 253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7162162162162163e-05,
+      "loss": 0.5158,
+      "step": 254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.722972972972973e-05,
+      "loss": 0.5451,
+      "step": 255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 0.5463,
+      "step": 256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7364864864864866e-05,
+      "loss": 0.5546,
+      "step": 257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7432432432432434e-05,
+      "loss": 0.5816,
+      "step": 258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.5527,
+      "step": 259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 0.5318,
+      "step": 260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7635135135135137e-05,
+      "loss": 0.6303,
+      "step": 261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7702702702702702e-05,
+      "loss": 0.5588,
+      "step": 262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7770270270270273e-05,
+      "loss": 0.6319,
+      "step": 263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.5217,
+      "step": 264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.790540540540541e-05,
+      "loss": 0.5663,
+      "step": 265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7972972972972976e-05,
+      "loss": 0.5309,
+      "step": 266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.804054054054054e-05,
+      "loss": 0.6477,
+      "step": 267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 0.5731,
+      "step": 268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8175675675675676e-05,
+      "loss": 0.5505,
+      "step": 269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8243243243243244e-05,
+      "loss": 0.5552,
+      "step": 270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.831081081081081e-05,
+      "loss": 0.6045,
+      "step": 271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.5157,
+      "step": 272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8445945945945947e-05,
+      "loss": 0.5161,
+      "step": 273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8513513513513515e-05,
+      "loss": 0.5815,
+      "step": 274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8581081081081082e-05,
+      "loss": 0.57,
+      "step": 275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 0.613,
+      "step": 276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8716216216216218e-05,
+      "loss": 0.5625,
+      "step": 277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8783783783783786e-05,
+      "loss": 0.5741,
+      "step": 278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8851351351351353e-05,
+      "loss": 0.5403,
+      "step": 279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.5408,
+      "step": 280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.898648648648649e-05,
+      "loss": 0.5726,
+      "step": 281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9054054054054057e-05,
+      "loss": 0.5448,
+      "step": 282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.912162162162162e-05,
+      "loss": 0.6339,
+      "step": 283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 0.5291,
+      "step": 284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.925675675675676e-05,
+      "loss": 0.5458,
+      "step": 285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9324324324324328e-05,
+      "loss": 0.6173,
+      "step": 286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9391891891891895e-05,
+      "loss": 0.5534,
+      "step": 287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.5505,
+      "step": 288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9527027027027027e-05,
+      "loss": 0.5114,
+      "step": 289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9594594594594595e-05,
+      "loss": 0.4992,
+      "step": 290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9662162162162163e-05,
+      "loss": 0.5733,
+      "step": 291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 0.4525,
+      "step": 292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.97972972972973e-05,
+      "loss": 0.559,
+      "step": 293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9864864864864866e-05,
+      "loss": 0.4898,
+      "step": 294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9932432432432434e-05,
+      "loss": 0.5812,
+      "step": 295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2e-05,
+      "loss": 0.5324,
+      "step": 296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9999999459936504e-05,
+      "loss": 0.5364,
+      "step": 297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.999999783974608e-05,
+      "loss": 0.5537,
+      "step": 298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9999995139428895e-05,
+      "loss": 0.5752,
+      "step": 299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.999999135898525e-05,
+      "loss": 0.5425,
+      "step": 300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.999998649841554e-05,
+      "loss": 0.5548,
+      "step": 301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.999998055772031e-05,
+      "loss": 0.5653,
+      "step": 302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.999997353690018e-05,
+      "loss": 0.5724,
+      "step": 303
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.999996543595592e-05,
+      "loss": 0.584,
+      "step": 304
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9999956254888407e-05,
+      "loss": 0.4942,
+      "step": 305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999945993698628e-05,
+      "loss": 0.5502,
+      "step": 306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999934652387694e-05,
+      "loss": 0.5341,
+      "step": 307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999922230956823e-05,
+      "loss": 0.5537,
+      "step": 308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999908729407366e-05,
+      "loss": 0.5598,
+      "step": 309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9999894147740776e-05,
+      "loss": 0.5026,
+      "step": 310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.999987848595863e-05,
+      "loss": 0.5777,
+      "step": 311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.999986174406262e-05,
+      "loss": 0.5641,
+      "step": 312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999843922054553e-05,
+      "loss": 0.5561,
+      "step": 313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999825019936353e-05,
+      "loss": 0.6003,
+      "step": 314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999805037710064e-05,
+      "loss": 0.6545,
+      "step": 315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999783975377847e-05,
+      "loss": 0.5752,
+      "step": 316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.999976183294197e-05,
+      "loss": 0.6076,
+      "step": 317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9999738610404825e-05,
+      "loss": 0.5824,
+      "step": 318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.999971430776893e-05,
+      "loss": 0.4935,
+      "step": 319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999688925036897e-05,
+      "loss": 0.5571,
+      "step": 320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999662462211477e-05,
+      "loss": 0.5686,
+      "step": 321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999634919295524e-05,
+      "loss": 0.5588,
+      "step": 322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999606296292017e-05,
+      "loss": 0.4885,
+      "step": 323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999576593204043e-05,
+      "loss": 0.6437,
+      "step": 324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9999545810034816e-05,
+      "loss": 0.5148,
+      "step": 325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999951394678765e-05,
+      "loss": 0.564,
+      "step": 326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999481003466003e-05,
+      "loss": 0.5533,
+      "step": 327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999944698007342e-05,
+      "loss": 0.5643,
+      "step": 328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999411876613578e-05,
+      "loss": 0.5243,
+      "step": 329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9999375693090274e-05,
+      "loss": 0.4773,
+      "step": 330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.999933842950741e-05,
+      "loss": 0.5497,
+      "step": 331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999300085869015e-05,
+      "loss": 0.5233,
+      "step": 332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999260662179232e-05,
+      "loss": 0.5529,
+      "step": 333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999220158442315e-05,
+      "loss": 0.5242,
+      "step": 334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999178574662642e-05,
+      "loss": 0.5603,
+      "step": 335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.99991359108447e-05,
+      "loss": 0.5458,
+      "step": 336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999092166993106e-05,
+      "loss": 0.5439,
+      "step": 337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9999047343112577e-05,
+      "loss": 0.5901,
+      "step": 338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9999001439207958e-05,
+      "loss": 0.6261,
+      "step": 339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998954455284204e-05,
+      "loss": 0.5701,
+      "step": 340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999890639134639e-05,
+      "loss": 0.4995,
+      "step": 341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998857247399718e-05,
+      "loss": 0.545,
+      "step": 342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9998807023449482e-05,
+      "loss": 0.5569,
+      "step": 343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.999875571950111e-05,
+      "loss": 0.5423,
+      "step": 344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998703335560153e-05,
+      "loss": 0.5318,
+      "step": 345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998649871632257e-05,
+      "loss": 0.5711,
+      "step": 346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998595327723205e-05,
+      "loss": 0.4777,
+      "step": 347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998539703838885e-05,
+      "loss": 0.5569,
+      "step": 348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998482999985304e-05,
+      "loss": 0.5296,
+      "step": 349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9998425216168588e-05,
+      "loss": 0.5469,
+      "step": 350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.999836635239498e-05,
+      "loss": 0.5815,
+      "step": 351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9998306408670837e-05,
+      "loss": 0.5465,
+      "step": 352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9998245385002635e-05,
+      "loss": 0.5109,
+      "step": 353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.999818328139696e-05,
+      "loss": 0.5329,
+      "step": 354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9998120097860524e-05,
+      "loss": 0.5531,
+      "step": 355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9998055834400153e-05,
+      "loss": 0.532,
+      "step": 356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9997990491022787e-05,
+      "loss": 0.5294,
+      "step": 357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9997924067735483e-05,
+      "loss": 0.5408,
+      "step": 358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997856564545416e-05,
+      "loss": 0.5341,
+      "step": 359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997787981459876e-05,
+      "loss": 0.5735,
+      "step": 360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997718318486275e-05,
+      "loss": 0.5402,
+      "step": 361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997647575632134e-05,
+      "loss": 0.5407,
+      "step": 362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997575752905094e-05,
+      "loss": 0.5631,
+      "step": 363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9997502850312915e-05,
+      "loss": 0.6028,
+      "step": 364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997428867863465e-05,
+      "loss": 0.5011,
+      "step": 365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997353805564742e-05,
+      "loss": 0.5205,
+      "step": 366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997277663424854e-05,
+      "loss": 0.5514,
+      "step": 367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997200441452022e-05,
+      "loss": 0.522,
+      "step": 368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9997122139654587e-05,
+      "loss": 0.6173,
+      "step": 369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999704275804101e-05,
+      "loss": 0.5363,
+      "step": 370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.999696229661986e-05,
+      "loss": 0.587,
+      "step": 371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.999688075539983e-05,
+      "loss": 0.6062,
+      "step": 372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9996798134389728e-05,
+      "loss": 0.5636,
+      "step": 373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9996714433598483e-05,
+      "loss": 0.5548,
+      "step": 374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9996629653035128e-05,
+      "loss": 0.6046,
+      "step": 375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.999654379270882e-05,
+      "loss": 0.6019,
+      "step": 376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.999645685262884e-05,
+      "loss": 0.509,
+      "step": 377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9996368832804572e-05,
+      "loss": 0.538,
+      "step": 378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.999627973324553e-05,
+      "loss": 0.499,
+      "step": 379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9996189553961333e-05,
+      "loss": 0.632,
+      "step": 380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9996098294961722e-05,
+      "loss": 0.5725,
+      "step": 381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9996005956256555e-05,
+      "loss": 0.595,
+      "step": 382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9995912537855803e-05,
+      "loss": 0.559,
+      "step": 383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9995818039769564e-05,
+      "loss": 0.5434,
+      "step": 384
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9995722462008035e-05,
+      "loss": 0.526,
+      "step": 385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9995625804581547e-05,
+      "loss": 0.495,
+      "step": 386
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9995528067500535e-05,
+      "loss": 0.5222,
+      "step": 387
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9995429250775564e-05,
+      "loss": 0.5484,
+      "step": 388
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9995329354417295e-05,
+      "loss": 0.5056,
+      "step": 389
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.999522837843653e-05,
+      "loss": 0.5506,
+      "step": 390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999512632284417e-05,
+      "loss": 0.5296,
+      "step": 391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999502318765124e-05,
+      "loss": 0.5602,
+      "step": 392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9994918972868877e-05,
+      "loss": 0.4763,
+      "step": 393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999481367850834e-05,
+      "loss": 0.4914,
+      "step": 394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9994707304581e-05,
+      "loss": 0.5578,
+      "step": 395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.999459985109835e-05,
+      "loss": 0.5634,
+      "step": 396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9994491318071993e-05,
+      "loss": 0.5367,
+      "step": 397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9994381705513658e-05,
+      "loss": 0.6322,
+      "step": 398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9994271013435176e-05,
+      "loss": 0.556,
+      "step": 399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9994159241848507e-05,
+      "loss": 0.5287,
+      "step": 400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9994046390765726e-05,
+      "loss": 0.5331,
+      "step": 401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.999393246019902e-05,
+      "loss": 0.5717,
+      "step": 402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9993817450160696e-05,
+      "loss": 0.4836,
+      "step": 403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9993701360663174e-05,
+      "loss": 0.5528,
+      "step": 404
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9993584191718997e-05,
+      "loss": 0.5089,
+      "step": 405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999346594334082e-05,
+      "loss": 0.4756,
+      "step": 406
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.999334661554141e-05,
+      "loss": 0.5641,
+      "step": 407
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9993226208333664e-05,
+      "loss": 0.5672,
+      "step": 408
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9993104721730582e-05,
+      "loss": 0.4974,
+      "step": 409
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9992982155745287e-05,
+      "loss": 0.5127,
+      "step": 410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999285851039102e-05,
+      "loss": 0.5224,
+      "step": 411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9992733785681134e-05,
+      "loss": 0.5516,
+      "step": 412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.99926079816291e-05,
+      "loss": 0.5701,
+      "step": 413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9992481098248507e-05,
+      "loss": 0.4873,
+      "step": 414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9992353135553064e-05,
+      "loss": 0.4651,
+      "step": 415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999222409355659e-05,
+      "loss": 0.4551,
+      "step": 416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.999209397227302e-05,
+      "loss": 0.5146,
+      "step": 417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991962771716414e-05,
+      "loss": 0.5517,
+      "step": 418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.999183049190094e-05,
+      "loss": 0.4813,
+      "step": 419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991697132840884e-05,
+      "loss": 0.5602,
+      "step": 420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991562694550655e-05,
+      "loss": 0.5144,
+      "step": 421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991427177044774e-05,
+      "loss": 0.5225,
+      "step": 422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9991290580337873e-05,
+      "loss": 0.5138,
+      "step": 423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999115290444471e-05,
+      "loss": 0.5812,
+      "step": 424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.999101414938016e-05,
+      "loss": 0.5741,
+      "step": 425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9990874315159202e-05,
+      "loss": 0.493,
+      "step": 426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9990733401796945e-05,
+      "loss": 0.5262,
+      "step": 427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9990591409308607e-05,
+      "loss": 0.518,
+      "step": 428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9990448337709528e-05,
+      "loss": 0.5356,
+      "step": 429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9990304187015163e-05,
+      "loss": 0.5106,
+      "step": 430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9990158957241074e-05,
+      "loss": 0.5178,
+      "step": 431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.999001264840296e-05,
+      "loss": 0.5292,
+      "step": 432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.998986526051661e-05,
+      "loss": 0.5329,
+      "step": 433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9989716793597948e-05,
+      "loss": 0.5605,
+      "step": 434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.998956724766302e-05,
+      "loss": 0.479,
+      "step": 435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9989416622727966e-05,
+      "loss": 0.5462,
+      "step": 436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9989264918809062e-05,
+      "loss": 0.54,
+      "step": 437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9989112135922697e-05,
+      "loss": 0.5143,
+      "step": 438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988958274085367e-05,
+      "loss": 0.5544,
+      "step": 439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988803333313694e-05,
+      "loss": 0.5476,
+      "step": 440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988647313624414e-05,
+      "loss": 0.5082,
+      "step": 441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988490215034377e-05,
+      "loss": 0.4654,
+      "step": 442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9988332037560555e-05,
+      "loss": 0.5191,
+      "step": 443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.998817278122003e-05,
+      "loss": 0.5056,
+      "step": 444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9988012446030005e-05,
+      "loss": 0.533,
+      "step": 445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987851032007798e-05,
+      "loss": 0.4872,
+      "step": 446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987688539170846e-05,
+      "loss": 0.5389,
+      "step": 447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.998752496753669e-05,
+      "loss": 0.5002,
+      "step": 448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987360317123015e-05,
+      "loss": 0.4608,
+      "step": 449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9987194587947592e-05,
+      "loss": 0.6426,
+      "step": 450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.998702778002833e-05,
+      "loss": 0.529,
+      "step": 451
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.998685989338324e-05,
+      "loss": 0.5164,
+      "step": 452
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986690928030455e-05,
+      "loss": 0.4688,
+      "step": 453
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986520883988233e-05,
+      "loss": 0.5327,
+      "step": 454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986349761274933e-05,
+      "loss": 0.4793,
+      "step": 455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9986177559909045e-05,
+      "loss": 0.5271,
+      "step": 456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.998600427990917e-05,
+      "loss": 0.4964,
+      "step": 457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9985829921294013e-05,
+      "loss": 0.5155,
+      "step": 458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.998565448408242e-05,
+      "loss": 0.5036,
+      "step": 459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.998547796829333e-05,
+      "loss": 0.5282,
+      "step": 460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.998530037394582e-05,
+      "loss": 0.5208,
+      "step": 461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9985121701059064e-05,
+      "loss": 0.4893,
+      "step": 462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9984941949652362e-05,
+      "loss": 0.5045,
+      "step": 463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9984761119745128e-05,
+      "loss": 0.5435,
+      "step": 464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.99845792113569e-05,
+      "loss": 0.5259,
+      "step": 465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9984396224507323e-05,
+      "loss": 0.4956,
+      "step": 466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.998421215921616e-05,
+      "loss": 0.5005,
+      "step": 467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9984027015503296e-05,
+      "loss": 0.5142,
+      "step": 468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9983840793388727e-05,
+      "loss": 0.4932,
+      "step": 469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.998365349289257e-05,
+      "loss": 0.4921,
+      "step": 470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.998346511403505e-05,
+      "loss": 0.469,
+      "step": 471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9983275656836517e-05,
+      "loss": 0.4887,
+      "step": 472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9983085121317436e-05,
+      "loss": 0.5055,
+      "step": 473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.998289350749839e-05,
+      "loss": 0.4951,
+      "step": 474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9982700815400073e-05,
+      "loss": 0.5219,
+      "step": 475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9982507045043294e-05,
+      "loss": 0.505,
+      "step": 476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9982312196448987e-05,
+      "loss": 0.511,
+      "step": 477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9982116269638197e-05,
+      "loss": 0.4584,
+      "step": 478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9981919264632088e-05,
+      "loss": 0.5666,
+      "step": 479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9981721181451936e-05,
+      "loss": 0.5325,
+      "step": 480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9981522020119142e-05,
+      "loss": 0.5318,
+      "step": 481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9981321780655212e-05,
+      "loss": 0.5532,
+      "step": 482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9981120463081778e-05,
+      "loss": 0.448,
+      "step": 483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980918067420583e-05,
+      "loss": 0.5097,
+      "step": 484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980714593693492e-05,
+      "loss": 0.5536,
+      "step": 485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980510041922476e-05,
+      "loss": 0.4453,
+      "step": 486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980304412129634e-05,
+      "loss": 0.4689,
+      "step": 487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9980097704337176e-05,
+      "loss": 0.5346,
+      "step": 488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.997988991856743e-05,
+      "loss": 0.5295,
+      "step": 489
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9979681054842838e-05,
+      "loss": 0.5239,
+      "step": 490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9979471113185958e-05,
+      "loss": 0.4734,
+      "step": 491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.997926009361947e-05,
+      "loss": 0.6027,
+      "step": 492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9979047996166164e-05,
+      "loss": 0.5668,
+      "step": 493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9978834820848952e-05,
+      "loss": 0.5193,
+      "step": 494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9978620567690856e-05,
+      "loss": 0.4646,
+      "step": 495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9978405236715025e-05,
+      "loss": 0.5085,
+      "step": 496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9978188827944706e-05,
+      "loss": 0.569,
+      "step": 497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9977971341403285e-05,
+      "loss": 0.5084,
+      "step": 498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.997775277711425e-05,
+      "loss": 0.5062,
+      "step": 499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9977533135101206e-05,
+      "loss": 0.4807,
+      "step": 500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9977312415387877e-05,
+      "loss": 0.4759,
+      "step": 501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9977090617998106e-05,
+      "loss": 0.4872,
+      "step": 502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.997686774295585e-05,
+      "loss": 0.4819,
+      "step": 503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.997664379028518e-05,
+      "loss": 0.51,
+      "step": 504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.997641876001029e-05,
+      "loss": 0.5114,
+      "step": 505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9976192652155483e-05,
+      "loss": 0.5463,
+      "step": 506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.997596546674518e-05,
+      "loss": 0.5175,
+      "step": 507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9975737203803924e-05,
+      "loss": 0.5078,
+      "step": 508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9975507863356368e-05,
+      "loss": 0.4822,
+      "step": 509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.997527744542728e-05,
+      "loss": 0.475,
+      "step": 510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9975045950041556e-05,
+      "loss": 0.4978,
+      "step": 511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9974813377224195e-05,
+      "loss": 0.5204,
+      "step": 512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.997457972700032e-05,
+      "loss": 0.5216,
+      "step": 513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9974344999395167e-05,
+      "loss": 0.5167,
+      "step": 514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.997410919443409e-05,
+      "loss": 0.4756,
+      "step": 515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9973872312142556e-05,
+      "loss": 0.5203,
+      "step": 516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9973634352546156e-05,
+      "loss": 0.4888,
+      "step": 517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9973395315670593e-05,
+      "loss": 0.512,
+      "step": 518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9973155201541682e-05,
+      "loss": 0.5114,
+      "step": 519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.997291401018536e-05,
+      "loss": 0.5066,
+      "step": 520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.997267174162768e-05,
+      "loss": 0.4773,
+      "step": 521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9972428395894806e-05,
+      "loss": 0.5345,
+      "step": 522
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9972183973013027e-05,
+      "loss": 0.5019,
+      "step": 523
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971938473008743e-05,
+      "loss": 0.5211,
+      "step": 524
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997169189590847e-05,
+      "loss": 0.4664,
+      "step": 525
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997144424173884e-05,
+      "loss": 0.4829,
+      "step": 526
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9971195510526607e-05,
+      "loss": 0.4404,
+      "step": 527
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997094570229863e-05,
+      "loss": 0.454,
+      "step": 528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.99706948170819e-05,
+      "loss": 0.5397,
+      "step": 529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9970442854903512e-05,
+      "loss": 0.497,
+      "step": 530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9970189815790683e-05,
+      "loss": 0.5275,
+      "step": 531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9969935699770742e-05,
+      "loss": 0.5494,
+      "step": 532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9969680506871138e-05,
+      "loss": 0.4456,
+      "step": 533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9969424237119428e-05,
+      "loss": 0.4981,
+      "step": 534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9969166890543304e-05,
+      "loss": 0.4984,
+      "step": 535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9968908467170556e-05,
+      "loss": 0.506,
+      "step": 536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9968648967029098e-05,
+      "loss": 0.4817,
+      "step": 537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.996838839014696e-05,
+      "loss": 0.4676,
+      "step": 538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.996812673655229e-05,
+      "loss": 0.4809,
+      "step": 539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9967864006273343e-05,
+      "loss": 0.4833,
+      "step": 540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9967600199338503e-05,
+      "loss": 0.5524,
+      "step": 541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9967335315776265e-05,
+      "loss": 0.478,
+      "step": 542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9967069355615235e-05,
+      "loss": 0.5395,
+      "step": 543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9966802318884142e-05,
+      "loss": 0.4801,
+      "step": 544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9966534205611833e-05,
+      "loss": 0.4229,
+      "step": 545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.996626501582726e-05,
+      "loss": 0.5238,
+      "step": 546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.996599474955951e-05,
+      "loss": 0.504,
+      "step": 547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.996572340683777e-05,
+      "loss": 0.4839,
+      "step": 548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.996545098769134e-05,
+      "loss": 0.528,
+      "step": 549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9965177492149656e-05,
+      "loss": 0.4331,
+      "step": 550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9964902920242254e-05,
+      "loss": 0.5174,
+      "step": 551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9964627271998792e-05,
+      "loss": 0.5231,
+      "step": 552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9964350547449047e-05,
+      "loss": 0.498,
+      "step": 553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9964072746622905e-05,
+      "loss": 0.4588,
+      "step": 554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.996379386955037e-05,
+      "loss": 0.4823,
+      "step": 555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.996351391626157e-05,
+      "loss": 0.5078,
+      "step": 556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9963232886786738e-05,
+      "loss": 0.4459,
+      "step": 557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.996295078115623e-05,
+      "loss": 0.4544,
+      "step": 558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.996266759940052e-05,
+      "loss": 0.4501,
+      "step": 559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9962383341550194e-05,
+      "loss": 0.5005,
+      "step": 560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9962098007635956e-05,
+      "loss": 0.461,
+      "step": 561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9961811597688622e-05,
+      "loss": 0.5439,
+      "step": 562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.996152411173913e-05,
+      "loss": 0.4777,
+      "step": 563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9961235549818532e-05,
+      "loss": 0.4871,
+      "step": 564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9960945911958e-05,
+      "loss": 0.4674,
+      "step": 565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9960655198188813e-05,
+      "loss": 0.5191,
+      "step": 566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9960363408542376e-05,
+      "loss": 0.4425,
+      "step": 567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9960070543050204e-05,
+      "loss": 0.4746,
+      "step": 568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.995977660174393e-05,
+      "loss": 0.4342,
+      "step": 569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9959481584655306e-05,
+      "loss": 0.4424,
+      "step": 570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9959185491816192e-05,
+      "loss": 0.4858,
+      "step": 571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9958888323258574e-05,
+      "loss": 0.4419,
+      "step": 572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9958590079014552e-05,
+      "loss": 0.4712,
+      "step": 573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9958290759116332e-05,
+      "loss": 0.5155,
+      "step": 574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9957990363596252e-05,
+      "loss": 0.5149,
+      "step": 575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.995768889248676e-05,
+      "loss": 0.5083,
+      "step": 576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.995738634582041e-05,
+      "loss": 0.488,
+      "step": 577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.995708272362989e-05,
+      "loss": 0.5283,
+      "step": 578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.995677802594799e-05,
+      "loss": 0.5182,
+      "step": 579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9956472252807616e-05,
+      "loss": 0.5213,
+      "step": 580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9956165404241807e-05,
+      "loss": 0.4751,
+      "step": 581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9955857480283702e-05,
+      "loss": 0.5448,
+      "step": 582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9955548480966558e-05,
+      "loss": 0.5201,
+      "step": 583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9955238406323754e-05,
+      "loss": 0.4848,
+      "step": 584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9954927256388777e-05,
+      "loss": 0.5157,
+      "step": 585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9954615031195242e-05,
+      "loss": 0.5097,
+      "step": 586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9954301730776868e-05,
+      "loss": 0.5076,
+      "step": 587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9953987355167498e-05,
+      "loss": 0.4053,
+      "step": 588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9953671904401092e-05,
+      "loss": 0.4562,
+      "step": 589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.995335537851171e-05,
+      "loss": 0.4517,
+      "step": 590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9953037777533557e-05,
+      "loss": 0.5223,
+      "step": 591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.995271910150093e-05,
+      "loss": 0.4349,
+      "step": 592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9952399350448247e-05,
+      "loss": 0.5338,
+      "step": 593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.995207852441005e-05,
+      "loss": 0.4626,
+      "step": 594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.995175662342099e-05,
+      "loss": 0.4989,
+      "step": 595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9951433647515838e-05,
+      "loss": 0.5725,
+      "step": 596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9951109596729477e-05,
+      "loss": 0.5279,
+      "step": 597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9950784471096915e-05,
+      "loss": 0.4842,
+      "step": 598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9950458270653264e-05,
+      "loss": 0.4859,
+      "step": 599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9950130995433756e-05,
+      "loss": 0.4805,
+      "step": 600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9949802645473748e-05,
+      "loss": 0.5637,
+      "step": 601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9949473220808698e-05,
+      "loss": 0.4539,
+      "step": 602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9949142721474194e-05,
+      "loss": 0.4437,
+      "step": 603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.994881114750593e-05,
+      "loss": 0.4455,
+      "step": 604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9948478498939723e-05,
+      "loss": 0.4681,
+      "step": 605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9948144775811503e-05,
+      "loss": 0.4991,
+      "step": 606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9947809978157314e-05,
+      "loss": 0.5093,
+      "step": 607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9947474106013326e-05,
+      "loss": 0.4953,
+      "step": 608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9947137159415804e-05,
+      "loss": 0.4688,
+      "step": 609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9946799138401152e-05,
+      "loss": 0.4722,
+      "step": 610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9946460043005882e-05,
+      "loss": 0.4275,
+      "step": 611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9946119873266615e-05,
+      "loss": 0.4725,
+      "step": 612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9945778629220096e-05,
+      "loss": 0.4759,
+      "step": 613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9945436310903185e-05,
+      "loss": 0.4554,
+      "step": 614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.994509291835285e-05,
+      "loss": 0.5022,
+      "step": 615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9944748451606194e-05,
+      "loss": 0.4634,
+      "step": 616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9944402910700416e-05,
+      "loss": 0.4838,
+      "step": 617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.994405629567284e-05,
+      "loss": 0.4924,
+      "step": 618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9943708606560902e-05,
+      "loss": 0.5056,
+      "step": 619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9943359843402158e-05,
+      "loss": 0.4944,
+      "step": 620
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9943010006234284e-05,
+      "loss": 0.4533,
+      "step": 621
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9942659095095057e-05,
+      "loss": 0.4258,
+      "step": 622
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.994230711002239e-05,
+      "loss": 0.4535,
+      "step": 623
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.99419540510543e-05,
+      "loss": 0.4735,
+      "step": 624
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.994159991822892e-05,
+      "loss": 0.4646,
+      "step": 625
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.9941244711584497e-05,
+      "loss": 0.4357,
+      "step": 626
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.99408884311594e-05,
+      "loss": 0.5119,
+      "step": 627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9940531076992115e-05,
+      "loss": 0.4699,
+      "step": 628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.994017264912124e-05,
+      "loss": 0.4674,
+      "step": 629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9939813147585487e-05,
+      "loss": 0.4504,
+      "step": 630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.993945257242369e-05,
+      "loss": 0.4949,
+      "step": 631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9939090923674792e-05,
+      "loss": 0.4741,
+      "step": 632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.993872820137786e-05,
+      "loss": 0.4617,
+      "step": 633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.993836440557207e-05,
+      "loss": 0.5,
+      "step": 634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9937999536296717e-05,
+      "loss": 0.4718,
+      "step": 635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9937633593591213e-05,
+      "loss": 0.4785,
+      "step": 636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.993726657749508e-05,
+      "loss": 0.4643,
+      "step": 637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9936898488047963e-05,
+      "loss": 0.5087,
+      "step": 638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.993652932528962e-05,
+      "loss": 0.5765,
+      "step": 639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.993615908925993e-05,
+      "loss": 0.4623,
+      "step": 640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9935787779998876e-05,
+      "loss": 0.3893,
+      "step": 641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.993541539754657e-05,
+      "loss": 0.4297,
+      "step": 642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9935041941943226e-05,
+      "loss": 0.48,
+      "step": 643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9934667413229195e-05,
+      "loss": 0.5143,
+      "step": 644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9934291811444914e-05,
+      "loss": 0.452,
+      "step": 645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.993391513663097e-05,
+      "loss": 0.4591,
+      "step": 646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9933537388828034e-05,
+      "loss": 0.458,
+      "step": 647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.993315856807692e-05,
+      "loss": 0.4172,
+      "step": 648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9932778674418535e-05,
+      "loss": 0.5741,
+      "step": 649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.993239770789392e-05,
+      "loss": 0.4814,
+      "step": 650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.993201566854422e-05,
+      "loss": 0.4782,
+      "step": 651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.99316325564107e-05,
+      "loss": 0.5025,
+      "step": 652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.993124837153474e-05,
+      "loss": 0.5182,
+      "step": 653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9930863113957842e-05,
+      "loss": 0.4925,
+      "step": 654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9930476783721615e-05,
+      "loss": 0.4521,
+      "step": 655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.993008938086779e-05,
+      "loss": 0.4577,
+      "step": 656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9929700905438207e-05,
+      "loss": 0.4634,
+      "step": 657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9929311357474832e-05,
+      "loss": 0.3937,
+      "step": 658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9928920737019735e-05,
+      "loss": 0.3847,
+      "step": 659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9928529044115112e-05,
+      "loss": 0.3304,
+      "step": 660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.992813627880327e-05,
+      "loss": 0.3831,
+      "step": 661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9927742441126635e-05,
+      "loss": 0.3811,
+      "step": 662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.992734753112774e-05,
+      "loss": 0.3928,
+      "step": 663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.992695154884925e-05,
+      "loss": 0.3811,
+      "step": 664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9926554494333924e-05,
+      "loss": 0.3404,
+      "step": 665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9926156367624658e-05,
+      "loss": 0.423,
+      "step": 666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.992575716876445e-05,
+      "loss": 0.3876,
+      "step": 667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9925356897796425e-05,
+      "loss": 0.3528,
+      "step": 668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.992495555476381e-05,
+      "loss": 0.3263,
+      "step": 669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9924553139709957e-05,
+      "loss": 0.3937,
+      "step": 670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9924149652678336e-05,
+      "loss": 0.3974,
+      "step": 671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9923745093712524e-05,
+      "loss": 0.3734,
+      "step": 672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.992333946285622e-05,
+      "loss": 0.4078,
+      "step": 673
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9922932760153238e-05,
+      "loss": 0.3784,
+      "step": 674
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9922524985647507e-05,
+      "loss": 0.3601,
+      "step": 675
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9922116139383074e-05,
+      "loss": 0.327,
+      "step": 676
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9921706221404095e-05,
+      "loss": 0.3754,
+      "step": 677
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9921295231754843e-05,
+      "loss": 0.4144,
+      "step": 678
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9920883170479724e-05,
+      "loss": 0.3632,
+      "step": 679
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9920470037623233e-05,
+      "loss": 0.3764,
+      "step": 680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9920055833229996e-05,
+      "loss": 0.3681,
+      "step": 681
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9919640557344756e-05,
+      "loss": 0.322,
+      "step": 682
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.991922421001237e-05,
+      "loss": 0.3405,
+      "step": 683
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.99188067912778e-05,
+      "loss": 0.3383,
+      "step": 684
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.991838830118614e-05,
+      "loss": 0.3704,
+      "step": 685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.991796873978259e-05,
+      "loss": 0.3143,
+      "step": 686
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9917548107112468e-05,
+      "loss": 0.3948,
+      "step": 687
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9917126403221206e-05,
+      "loss": 0.3362,
+      "step": 688
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.991670362815436e-05,
+      "loss": 0.3377,
+      "step": 689
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.991627978195758e-05,
+      "loss": 0.3504,
+      "step": 690
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9915854864676665e-05,
+      "loss": 0.3596,
+      "step": 691
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99154288763575e-05,
+      "loss": 0.3504,
+      "step": 692
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.99150018170461e-05,
+      "loss": 0.3337,
+      "step": 693
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9914573686788595e-05,
+      "loss": 0.411,
+      "step": 694
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9914144485631226e-05,
+      "loss": 0.3678,
+      "step": 695
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9913714213620354e-05,
+      "loss": 0.3562,
+      "step": 696
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9913282870802452e-05,
+      "loss": 0.3444,
+      "step": 697
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.991285045722411e-05,
+      "loss": 0.4012,
+      "step": 698
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9912416972932035e-05,
+      "loss": 0.4125,
+      "step": 699
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9911982417973052e-05,
+      "loss": 0.3374,
+      "step": 700
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9911546792394094e-05,
+      "loss": 0.3609,
+      "step": 701
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9911110096242215e-05,
+      "loss": 0.3921,
+      "step": 702
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9910672329564584e-05,
+      "loss": 0.3307,
+      "step": 703
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9910233492408487e-05,
+      "loss": 0.3628,
+      "step": 704
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9909793584821323e-05,
+      "loss": 0.3833,
+      "step": 705
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9909352606850607e-05,
+      "loss": 0.3913,
+      "step": 706
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9908910558543974e-05,
+      "loss": 0.365,
+      "step": 707
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990846743994916e-05,
+      "loss": 0.3332,
+      "step": 708
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990802325111404e-05,
+      "loss": 0.3684,
+      "step": 709
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990757799208659e-05,
+      "loss": 0.3655,
+      "step": 710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9907131662914892e-05,
+      "loss": 0.3824,
+      "step": 711
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.990668426364717e-05,
+      "loss": 0.3839,
+      "step": 712
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.990623579433174e-05,
+      "loss": 0.3461,
+      "step": 713
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9905786255017046e-05,
+      "loss": 0.3748,
+      "step": 714
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9905335645751644e-05,
+      "loss": 0.367,
+      "step": 715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9904883966584204e-05,
+      "loss": 0.4079,
+      "step": 716
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9904431217563513e-05,
+      "loss": 0.3833,
+      "step": 717
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.990397739873847e-05,
+      "loss": 0.4113,
+      "step": 718
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9903522510158105e-05,
+      "loss": 0.3433,
+      "step": 719
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.990306655187154e-05,
+      "loss": 0.4084,
+      "step": 720
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9902609523928028e-05,
+      "loss": 0.3607,
+      "step": 721
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9902151426376934e-05,
+      "loss": 0.3614,
+      "step": 722
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.990169225926774e-05,
+      "loss": 0.3772,
+      "step": 723
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.990123202265004e-05,
+      "loss": 0.4327,
+      "step": 724
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9900770716573545e-05,
+      "loss": 0.369,
+      "step": 725
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9900308341088084e-05,
+      "loss": 0.3369,
+      "step": 726
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9899844896243596e-05,
+      "loss": 0.4339,
+      "step": 727
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9899380382090142e-05,
+      "loss": 0.3668,
+      "step": 728
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9898914798677898e-05,
+      "loss": 0.3609,
+      "step": 729
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9898448146057147e-05,
+      "loss": 0.3521,
+      "step": 730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9897980424278295e-05,
+      "loss": 0.3137,
+      "step": 731
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9897511633391865e-05,
+      "loss": 0.3491,
+      "step": 732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.989704177344849e-05,
+      "loss": 0.3578,
+      "step": 733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9896570844498916e-05,
+      "loss": 0.3788,
+      "step": 734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.989609884659402e-05,
+      "loss": 0.3782,
+      "step": 735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9895625779784774e-05,
+      "loss": 0.309,
+      "step": 736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9895151644122283e-05,
+      "loss": 0.3802,
+      "step": 737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9894676439657752e-05,
+      "loss": 0.3363,
+      "step": 738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9894200166442513e-05,
+      "loss": 0.3665,
+      "step": 739
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9893722824528016e-05,
+      "loss": 0.3755,
+      "step": 740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9893244413965806e-05,
+      "loss": 0.3856,
+      "step": 741
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9892764934807572e-05,
+      "loss": 0.37,
+      "step": 742
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.989228438710509e-05,
+      "loss": 0.3528,
+      "step": 743
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9891802770910276e-05,
+      "loss": 0.3553,
+      "step": 744
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9891320086275146e-05,
+      "loss": 0.3781,
+      "step": 745
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9890836333251836e-05,
+      "loss": 0.3286,
+      "step": 746
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.98903515118926e-05,
+      "loss": 0.371,
+      "step": 747
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9889865622249802e-05,
+      "loss": 0.3607,
+      "step": 748
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9889378664375926e-05,
+      "loss": 0.3831,
+      "step": 749
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.988889063832357e-05,
+      "loss": 0.3216,
+      "step": 750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9888401544145445e-05,
+      "loss": 0.4095,
+      "step": 751
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.988791138189438e-05,
+      "loss": 0.4143,
+      "step": 752
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.988742015162332e-05,
+      "loss": 0.3929,
+      "step": 753
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.988692785338532e-05,
+      "loss": 0.3767,
+      "step": 754
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.988643448723356e-05,
+      "loss": 0.3718,
+      "step": 755
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.988594005322133e-05,
+      "loss": 0.3798,
+      "step": 756
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.988544455140203e-05,
+      "loss": 0.387,
+      "step": 757
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9884947981829186e-05,
+      "loss": 0.3431,
+      "step": 758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.988445034455643e-05,
+      "loss": 0.3675,
+      "step": 759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9883951639637514e-05,
+      "loss": 0.3635,
+      "step": 760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9883451867126303e-05,
+      "loss": 0.3444,
+      "step": 761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.988295102707678e-05,
+      "loss": 0.3443,
+      "step": 762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9882449119543043e-05,
+      "loss": 0.4138,
+      "step": 763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9881946144579304e-05,
+      "loss": 0.3811,
+      "step": 764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9881442102239893e-05,
+      "loss": 0.4023,
+      "step": 765
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.988093699257925e-05,
+      "loss": 0.3895,
+      "step": 766
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.988043081565193e-05,
+      "loss": 0.4182,
+      "step": 767
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9879923571512613e-05,
+      "loss": 0.4063,
+      "step": 768
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.987941526021609e-05,
+      "loss": 0.4,
+      "step": 769
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9878905881817254e-05,
+      "loss": 0.3279,
+      "step": 770
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.987839543637113e-05,
+      "loss": 0.3447,
+      "step": 771
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9877883923932858e-05,
+      "loss": 0.3593,
+      "step": 772
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.987737134455768e-05,
+      "loss": 0.3901,
+      "step": 773
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9876857698300963e-05,
+      "loss": 0.3661,
+      "step": 774
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.987634298521819e-05,
+      "loss": 0.3626,
+      "step": 775
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.987582720536495e-05,
+      "loss": 0.3609,
+      "step": 776
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.987531035879697e-05,
+      "loss": 0.3999,
+      "step": 777
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9874792445570053e-05,
+      "loss": 0.3627,
+      "step": 778
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9874273465740158e-05,
+      "loss": 0.3656,
+      "step": 779
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9873753419363336e-05,
+      "loss": 0.3696,
+      "step": 780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9873232306495757e-05,
+      "loss": 0.3655,
+      "step": 781
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9872710127193707e-05,
+      "loss": 0.395,
+      "step": 782
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.987218688151359e-05,
+      "loss": 0.3716,
+      "step": 783
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9871662569511924e-05,
+      "loss": 0.3527,
+      "step": 784
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9871137191245343e-05,
+      "loss": 0.3375,
+      "step": 785
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9870610746770585e-05,
+      "loss": 0.367,
+      "step": 786
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9870083236144527e-05,
+      "loss": 0.3957,
+      "step": 787
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9869554659424133e-05,
+      "loss": 0.3973,
+      "step": 788
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9869025016666506e-05,
+      "loss": 0.3596,
+      "step": 789
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9868494307928848e-05,
+      "loss": 0.3784,
+      "step": 790
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.9867962533268487e-05,
+      "loss": 0.3386,
+      "step": 791
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.986742969274286e-05,
+      "loss": 0.3382,
+      "step": 792
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9866895786409516e-05,
+      "loss": 0.3061,
+      "step": 793
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9866360814326133e-05,
+      "loss": 0.3743,
+      "step": 794
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9865824776550483e-05,
+      "loss": 0.3811,
+      "step": 795
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9865287673140473e-05,
+      "loss": 0.3745,
+      "step": 796
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9864749504154116e-05,
+      "loss": 0.3487,
+      "step": 797
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.986421026964954e-05,
+      "loss": 0.3699,
+      "step": 798
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9863669969684988e-05,
+      "loss": 0.3864,
+      "step": 799
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9863128604318824e-05,
+      "loss": 0.3983,
+      "step": 800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9862586173609516e-05,
+      "loss": 0.4144,
+      "step": 801
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9862042677615657e-05,
+      "loss": 0.3676,
+      "step": 802
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.986149811639595e-05,
+      "loss": 0.3523,
+      "step": 803
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9860952490009214e-05,
+      "loss": 0.3776,
+      "step": 804
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.986040579851439e-05,
+      "loss": 0.3801,
+      "step": 805
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9859858041970518e-05,
+      "loss": 0.4028,
+      "step": 806
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.985930922043677e-05,
+      "loss": 0.3824,
+      "step": 807
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.985875933397242e-05,
+      "loss": 0.3768,
+      "step": 808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.985820838263687e-05,
+      "loss": 0.3527,
+      "step": 809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.985765636648962e-05,
+      "loss": 0.3989,
+      "step": 810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9857103285590302e-05,
+      "loss": 0.3364,
+      "step": 811
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9856549139998655e-05,
+      "loss": 0.4144,
+      "step": 812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9855993929774532e-05,
+      "loss": 0.3628,
+      "step": 813
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9855437654977904e-05,
+      "loss": 0.4016,
+      "step": 814
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9854880315668855e-05,
+      "loss": 0.3359,
+      "step": 815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.985432191190758e-05,
+      "loss": 0.4135,
+      "step": 816
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9853762443754404e-05,
+      "loss": 0.3398,
+      "step": 817
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9853201911269752e-05,
+      "loss": 0.368,
+      "step": 818
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.985264031451416e-05,
+      "loss": 0.351,
+      "step": 819
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9852077653548304e-05,
+      "loss": 0.4253,
+      "step": 820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9851513928432946e-05,
+      "loss": 0.3908,
+      "step": 821
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.985094913922898e-05,
+      "loss": 0.4189,
+      "step": 822
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.985038328599741e-05,
+      "loss": 0.3882,
+      "step": 823
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9849816368799356e-05,
+      "loss": 0.3618,
+      "step": 824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.984924838769605e-05,
+      "loss": 0.3465,
+      "step": 825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9848679342748844e-05,
+      "loss": 0.3673,
+      "step": 826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.98481092340192e-05,
+      "loss": 0.4223,
+      "step": 827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9847538061568697e-05,
+      "loss": 0.3561,
+      "step": 828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.984696582545903e-05,
+      "loss": 0.3893,
+      "step": 829
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9846392525752007e-05,
+      "loss": 0.3822,
+      "step": 830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9845818162509554e-05,
+      "loss": 0.3326,
+      "step": 831
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9845242735793705e-05,
+      "loss": 0.3739,
+      "step": 832
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9844666245666616e-05,
+      "loss": 0.3688,
+      "step": 833
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9844088692190553e-05,
+      "loss": 0.3514,
+      "step": 834
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9843510075427902e-05,
+      "loss": 0.4103,
+      "step": 835
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9842930395441165e-05,
+      "loss": 0.4014,
+      "step": 836
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9842349652292946e-05,
+      "loss": 0.3625,
+      "step": 837
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9841767846045978e-05,
+      "loss": 0.3378,
+      "step": 838
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9841184976763103e-05,
+      "loss": 0.3603,
+      "step": 839
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9840601044507272e-05,
+      "loss": 0.364,
+      "step": 840
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9840016049341567e-05,
+      "loss": 0.3707,
+      "step": 841
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9839429991329168e-05,
+      "loss": 0.4078,
+      "step": 842
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.983884287053338e-05,
+      "loss": 0.342,
+      "step": 843
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.983825468701762e-05,
+      "loss": 0.3773,
+      "step": 844
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.983766544084542e-05,
+      "loss": 0.3797,
+      "step": 845
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9837075132080418e-05,
+      "loss": 0.3508,
+      "step": 846
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.983648376078638e-05,
+      "loss": 0.4656,
+      "step": 847
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.983589132702719e-05,
+      "loss": 0.38,
+      "step": 848
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9835297830866827e-05,
+      "loss": 0.3499,
+      "step": 849
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9834703272369397e-05,
+      "loss": 0.4034,
+      "step": 850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9834107651599125e-05,
+      "loss": 0.3438,
+      "step": 851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9833510968620346e-05,
+      "loss": 0.3733,
+      "step": 852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9832913223497504e-05,
+      "loss": 0.4085,
+      "step": 853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9832314416295163e-05,
+      "loss": 0.422,
+      "step": 854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9831714547078008e-05,
+      "loss": 0.3159,
+      "step": 855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.983111361591083e-05,
+      "loss": 0.3546,
+      "step": 856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9830511622858535e-05,
+      "loss": 0.35,
+      "step": 857
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9829908567986145e-05,
+      "loss": 0.413,
+      "step": 858
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9829304451358803e-05,
+      "loss": 0.3809,
+      "step": 859
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9828699273041754e-05,
+      "loss": 0.3652,
+      "step": 860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9828093033100372e-05,
+      "loss": 0.3734,
+      "step": 861
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9827485731600132e-05,
+      "loss": 0.394,
+      "step": 862
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.982687736860664e-05,
+      "loss": 0.3503,
+      "step": 863
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9826267944185597e-05,
+      "loss": 0.3762,
+      "step": 864
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.982565745840283e-05,
+      "loss": 0.4014,
+      "step": 865
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9825045911324287e-05,
+      "loss": 0.35,
+      "step": 866
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9824433303016015e-05,
+      "loss": 0.3671,
+      "step": 867
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9823819633544185e-05,
+      "loss": 0.3702,
+      "step": 868
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.982320490297508e-05,
+      "loss": 0.3423,
+      "step": 869
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.9822589111375104e-05,
+      "loss": 0.4135,
+      "step": 870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9821972258810766e-05,
+      "loss": 0.3261,
+      "step": 871
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9821354345348693e-05,
+      "loss": 0.4031,
+      "step": 872
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.982073537105563e-05,
+      "loss": 0.374,
+      "step": 873
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9820115335998435e-05,
+      "loss": 0.3479,
+      "step": 874
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9819494240244076e-05,
+      "loss": 0.3705,
+      "step": 875
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.981887208385964e-05,
+      "loss": 0.3988,
+      "step": 876
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9818248866912334e-05,
+      "loss": 0.4078,
+      "step": 877
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9817624589469466e-05,
+      "loss": 0.3687,
+      "step": 878
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.981699925159847e-05,
+      "loss": 0.3428,
+      "step": 879
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.981637285336689e-05,
+      "loss": 0.3312,
+      "step": 880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.981574539484238e-05,
+      "loss": 0.3616,
+      "step": 881
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9815116876092716e-05,
+      "loss": 0.3732,
+      "step": 882
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9814487297185788e-05,
+      "loss": 0.3529,
+      "step": 883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.98138566581896e-05,
+      "loss": 0.3763,
+      "step": 884
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9813224959172267e-05,
+      "loss": 0.3183,
+      "step": 885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.981259220020202e-05,
+      "loss": 0.3429,
+      "step": 886
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9811958381347204e-05,
+      "loss": 0.3993,
+      "step": 887
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.981132350267628e-05,
+      "loss": 0.3496,
+      "step": 888
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9810687564257826e-05,
+      "loss": 0.4235,
+      "step": 889
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.981005056616053e-05,
+      "loss": 0.3702,
+      "step": 890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9809412508453194e-05,
+      "loss": 0.3693,
+      "step": 891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9808773391204737e-05,
+      "loss": 0.3743,
+      "step": 892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.980813321448419e-05,
+      "loss": 0.3582,
+      "step": 893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9807491978360708e-05,
+      "loss": 0.3866,
+      "step": 894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9806849682903538e-05,
+      "loss": 0.3686,
+      "step": 895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9806206328182075e-05,
+      "loss": 0.4136,
+      "step": 896
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9805561914265793e-05,
+      "loss": 0.3268,
+      "step": 897
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9804916441224305e-05,
+      "loss": 0.3463,
+      "step": 898
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9804269909127328e-05,
+      "loss": 0.3531,
+      "step": 899
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9803622318044698e-05,
+      "loss": 0.3686,
+      "step": 900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9802973668046364e-05,
+      "loss": 0.3485,
+      "step": 901
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.980232395920238e-05,
+      "loss": 0.3788,
+      "step": 902
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9801673191582935e-05,
+      "loss": 0.3553,
+      "step": 903
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.980102136525831e-05,
+      "loss": 0.4301,
+      "step": 904
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9800368480298916e-05,
+      "loss": 0.3701,
+      "step": 905
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.979971453677527e-05,
+      "loss": 0.3514,
+      "step": 906
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9799059534758008e-05,
+      "loss": 0.4222,
+      "step": 907
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.979840347431788e-05,
+      "loss": 0.3948,
+      "step": 908
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9797746355525743e-05,
+      "loss": 0.3674,
+      "step": 909
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.979708817845258e-05,
+      "loss": 0.3932,
+      "step": 910
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9796428943169478e-05,
+      "loss": 0.3635,
+      "step": 911
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9795768649747648e-05,
+      "loss": 0.3819,
+      "step": 912
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9795107298258406e-05,
+      "loss": 0.3264,
+      "step": 913
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.979444488877319e-05,
+      "loss": 0.3682,
+      "step": 914
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9793781421363544e-05,
+      "loss": 0.3206,
+      "step": 915
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9793116896101138e-05,
+      "loss": 0.3446,
+      "step": 916
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.979245131305774e-05,
+      "loss": 0.3165,
+      "step": 917
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9791784672305248e-05,
+      "loss": 0.4072,
+      "step": 918
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9791116973915662e-05,
+      "loss": 0.4469,
+      "step": 919
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.979044821796111e-05,
+      "loss": 0.3944,
+      "step": 920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.978977840451382e-05,
+      "loss": 0.3325,
+      "step": 921
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9789107533646138e-05,
+      "loss": 0.367,
+      "step": 922
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9788435605430535e-05,
+      "loss": 0.3887,
+      "step": 923
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9787762619939582e-05,
+      "loss": 0.3901,
+      "step": 924
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9787088577245973e-05,
+      "loss": 0.344,
+      "step": 925
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9786413477422507e-05,
+      "loss": 0.3591,
+      "step": 926
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9785737320542113e-05,
+      "loss": 0.3984,
+      "step": 927
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9785060106677818e-05,
+      "loss": 0.2998,
+      "step": 928
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.978438183590277e-05,
+      "loss": 0.3717,
+      "step": 929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.978370250829023e-05,
+      "loss": 0.3327,
+      "step": 930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.978302212391358e-05,
+      "loss": 0.3842,
+      "step": 931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.97823406828463e-05,
+      "loss": 0.3795,
+      "step": 932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9781658185162008e-05,
+      "loss": 0.3684,
+      "step": 933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.978097463093441e-05,
+      "loss": 0.3648,
+      "step": 934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9780290020237345e-05,
+      "loss": 0.3515,
+      "step": 935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9779604353144757e-05,
+      "loss": 0.3724,
+      "step": 936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9778917629730708e-05,
+      "loss": 0.3879,
+      "step": 937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9778229850069373e-05,
+      "loss": 0.3447,
+      "step": 938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.977754101423504e-05,
+      "loss": 0.3395,
+      "step": 939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.977685112230211e-05,
+      "loss": 0.3488,
+      "step": 940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9776160174345106e-05,
+      "loss": 0.3755,
+      "step": 941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.9775468170438655e-05,
+      "loss": 0.3809,
+      "step": 942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9774775110657503e-05,
+      "loss": 0.3788,
+      "step": 943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9774080995076508e-05,
+      "loss": 0.3598,
+      "step": 944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9773385823770646e-05,
+      "loss": 0.4157,
+      "step": 945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9772689596815e-05,
+      "loss": 0.3502,
+      "step": 946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.977199231428478e-05,
+      "loss": 0.3561,
+      "step": 947
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9771293976255293e-05,
+      "loss": 0.3949,
+      "step": 948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.977059458280197e-05,
+      "loss": 0.3581,
+      "step": 949
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9769894134000354e-05,
+      "loss": 0.3899,
+      "step": 950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9769192629926106e-05,
+      "loss": 0.3873,
+      "step": 951
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9768490070654994e-05,
+      "loss": 0.3328,
+      "step": 952
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9767786456262902e-05,
+      "loss": 0.38,
+      "step": 953
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9767081786825836e-05,
+      "loss": 0.3326,
+      "step": 954
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9766376062419903e-05,
+      "loss": 0.3789,
+      "step": 955
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.976566928312133e-05,
+      "loss": 0.3471,
+      "step": 956
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9764961449006464e-05,
+      "loss": 0.3427,
+      "step": 957
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9764252560151753e-05,
+      "loss": 0.3231,
+      "step": 958
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.976354261663377e-05,
+      "loss": 0.3543,
+      "step": 959
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9762831618529203e-05,
+      "loss": 0.346,
+      "step": 960
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9762119565914835e-05,
+      "loss": 0.3621,
+      "step": 961
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.976140645886759e-05,
+      "loss": 0.3791,
+      "step": 962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.976069229746448e-05,
+      "loss": 0.3952,
+      "step": 963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.975997708178266e-05,
+      "loss": 0.3344,
+      "step": 964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.975926081189937e-05,
+      "loss": 0.4108,
+      "step": 965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9758543487891975e-05,
+      "loss": 0.3784,
+      "step": 966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9757825109837965e-05,
+      "loss": 0.3578,
+      "step": 967
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9757105677814926e-05,
+      "loss": 0.3634,
+      "step": 968
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9756385191900568e-05,
+      "loss": 0.3657,
+      "step": 969
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9755663652172716e-05,
+      "loss": 0.356,
+      "step": 970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9754941058709302e-05,
+      "loss": 0.3984,
+      "step": 971
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9754217411588374e-05,
+      "loss": 0.3689,
+      "step": 972
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9753492710888096e-05,
+      "loss": 0.3603,
+      "step": 973
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9752766956686748e-05,
+      "loss": 0.3642,
+      "step": 974
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9752040149062717e-05,
+      "loss": 0.3865,
+      "step": 975
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9751312288094507e-05,
+      "loss": 0.3393,
+      "step": 976
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.975058337386074e-05,
+      "loss": 0.3887,
+      "step": 977
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9749853406440147e-05,
+      "loss": 0.3779,
+      "step": 978
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.974912238591157e-05,
+      "loss": 0.371,
+      "step": 979
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9748390312353972e-05,
+      "loss": 0.3786,
+      "step": 980
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9747657185846428e-05,
+      "loss": 0.3576,
+      "step": 981
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.974692300646812e-05,
+      "loss": 0.3827,
+      "step": 982
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9746187774298352e-05,
+      "loss": 0.3551,
+      "step": 983
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9745451489416537e-05,
+      "loss": 0.3959,
+      "step": 984
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9744714151902206e-05,
+      "loss": 0.3464,
+      "step": 985
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9743975761834996e-05,
+      "loss": 0.3799,
+      "step": 986
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9743236319294666e-05,
+      "loss": 0.3505,
+      "step": 987
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9742495824361083e-05,
+      "loss": 0.4055,
+      "step": 988
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.974175427711423e-05,
+      "loss": 0.3465,
+      "step": 989
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9741011677634208e-05,
+      "loss": 0.3749,
+      "step": 990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9740268026001225e-05,
+      "loss": 0.3485,
+      "step": 991
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.97395233222956e-05,
+      "loss": 0.3611,
+      "step": 992
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9738777566597772e-05,
+      "loss": 0.3647,
+      "step": 993
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.97380307589883e-05,
+      "loss": 0.3622,
+      "step": 994
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9737282899547837e-05,
+      "loss": 0.3863,
+      "step": 995
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9736533988357173e-05,
+      "loss": 0.3706,
+      "step": 996
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.973578402549719e-05,
+      "loss": 0.355,
+      "step": 997
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9735033011048898e-05,
+      "loss": 0.3658,
+      "step": 998
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9734280945093417e-05,
+      "loss": 0.334,
+      "step": 999
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9733527827711977e-05,
+      "loss": 0.3953,
+      "step": 1000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9732773658985925e-05,
+      "loss": 0.3514,
+      "step": 1001
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9732018438996722e-05,
+      "loss": 0.3638,
+      "step": 1002
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.973126216782594e-05,
+      "loss": 0.3621,
+      "step": 1003
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9730504845555266e-05,
+      "loss": 0.344,
+      "step": 1004
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9729746472266504e-05,
+      "loss": 0.3468,
+      "step": 1005
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9728987048041562e-05,
+      "loss": 0.3729,
+      "step": 1006
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9728226572962474e-05,
+      "loss": 0.3689,
+      "step": 1007
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9727465047111378e-05,
+      "loss": 0.393,
+      "step": 1008
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9726702470570524e-05,
+      "loss": 0.3368,
+      "step": 1009
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9725938843422284e-05,
+      "loss": 0.3449,
+      "step": 1010
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9725174165749143e-05,
+      "loss": 0.4259,
+      "step": 1011
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.972440843763369e-05,
+      "loss": 0.3642,
+      "step": 1012
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.972364165915863e-05,
+      "loss": 0.3842,
+      "step": 1013
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.97228738304068e-05,
+      "loss": 0.407,
+      "step": 1014
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.9722104951461118e-05,
+      "loss": 0.4024,
+      "step": 1015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9721335022404644e-05,
+      "loss": 0.3633,
+      "step": 1016
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9720564043320535e-05,
+      "loss": 0.3574,
+      "step": 1017
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.971979201429207e-05,
+      "loss": 0.37,
+      "step": 1018
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9719018935402632e-05,
+      "loss": 0.3862,
+      "step": 1019
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.971824480673573e-05,
+      "loss": 0.3387,
+      "step": 1020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9717469628374977e-05,
+      "loss": 0.3504,
+      "step": 1021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.97166934004041e-05,
+      "loss": 0.3712,
+      "step": 1022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9715916122906946e-05,
+      "loss": 0.3635,
+      "step": 1023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9715137795967467e-05,
+      "loss": 0.4021,
+      "step": 1024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9714358419669735e-05,
+      "loss": 0.3499,
+      "step": 1025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.971357799409793e-05,
+      "loss": 0.3139,
+      "step": 1026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.971279651933635e-05,
+      "loss": 0.382,
+      "step": 1027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.97120139954694e-05,
+      "loss": 0.3544,
+      "step": 1028
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9711230422581607e-05,
+      "loss": 0.3753,
+      "step": 1029
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9710445800757606e-05,
+      "loss": 0.39,
+      "step": 1030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9709660130082146e-05,
+      "loss": 0.4125,
+      "step": 1031
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9708873410640092e-05,
+      "loss": 0.3612,
+      "step": 1032
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.970808564251641e-05,
+      "loss": 0.3841,
+      "step": 1033
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.97072968257962e-05,
+      "loss": 0.3614,
+      "step": 1034
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.970650696056466e-05,
+      "loss": 0.3625,
+      "step": 1035
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9705716046907103e-05,
+      "loss": 0.3886,
+      "step": 1036
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9704924084908962e-05,
+      "loss": 0.375,
+      "step": 1037
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9704131074655778e-05,
+      "loss": 0.3905,
+      "step": 1038
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9703337016233203e-05,
+      "loss": 0.3977,
+      "step": 1039
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9702541909727008e-05,
+      "loss": 0.4145,
+      "step": 1040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9701745755223074e-05,
+      "loss": 0.3365,
+      "step": 1041
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9700948552807394e-05,
+      "loss": 0.3473,
+      "step": 1042
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9700150302566082e-05,
+      "loss": 0.371,
+      "step": 1043
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9699351004585354e-05,
+      "loss": 0.3622,
+      "step": 1044
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9698550658951543e-05,
+      "loss": 0.3969,
+      "step": 1045
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9697749265751102e-05,
+      "loss": 0.3686,
+      "step": 1046
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9696946825070586e-05,
+      "loss": 0.3221,
+      "step": 1047
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9696143336996676e-05,
+      "loss": 0.3222,
+      "step": 1048
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.969533880161615e-05,
+      "loss": 0.3422,
+      "step": 1049
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9694533219015913e-05,
+      "loss": 0.3542,
+      "step": 1050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9693726589282978e-05,
+      "loss": 0.3684,
+      "step": 1051
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.969291891250447e-05,
+      "loss": 0.4076,
+      "step": 1052
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.969211018876763e-05,
+      "loss": 0.3678,
+      "step": 1053
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9691300418159813e-05,
+      "loss": 0.3593,
+      "step": 1054
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9690489600768476e-05,
+      "loss": 0.362,
+      "step": 1055
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9689677736681203e-05,
+      "loss": 0.354,
+      "step": 1056
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9688864825985688e-05,
+      "loss": 0.3165,
+      "step": 1057
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968805086876973e-05,
+      "loss": 0.3484,
+      "step": 1058
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968723586512125e-05,
+      "loss": 0.3261,
+      "step": 1059
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.968641981512828e-05,
+      "loss": 0.3442,
+      "step": 1060
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9685602718878962e-05,
+      "loss": 0.4248,
+      "step": 1061
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.968478457646155e-05,
+      "loss": 0.3501,
+      "step": 1062
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.968396538796442e-05,
+      "loss": 0.3763,
+      "step": 1063
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9683145153476052e-05,
+      "loss": 0.3685,
+      "step": 1064
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.968232387308504e-05,
+      "loss": 0.374,
+      "step": 1065
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9681501546880094e-05,
+      "loss": 0.3515,
+      "step": 1066
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9680678174950033e-05,
+      "loss": 0.3809,
+      "step": 1067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.96798537573838e-05,
+      "loss": 0.3718,
+      "step": 1068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9679028294270433e-05,
+      "loss": 0.331,
+      "step": 1069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9678201785699095e-05,
+      "loss": 0.3431,
+      "step": 1070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.967737423175906e-05,
+      "loss": 0.3801,
+      "step": 1071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.967654563253972e-05,
+      "loss": 0.357,
+      "step": 1072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9675715988130567e-05,
+      "loss": 0.3928,
+      "step": 1073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9674885298621214e-05,
+      "loss": 0.3667,
+      "step": 1074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9674053564101388e-05,
+      "loss": 0.378,
+      "step": 1075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9673220784660925e-05,
+      "loss": 0.3346,
+      "step": 1076
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.967238696038978e-05,
+      "loss": 0.3939,
+      "step": 1077
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9671552091378015e-05,
+      "loss": 0.3511,
+      "step": 1078
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.96707161777158e-05,
+      "loss": 0.3733,
+      "step": 1079
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9669879219493432e-05,
+      "loss": 0.3595,
+      "step": 1080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966904121680131e-05,
+      "loss": 0.3345,
+      "step": 1081
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966820216972995e-05,
+      "loss": 0.3796,
+      "step": 1082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966736207836998e-05,
+      "loss": 0.3522,
+      "step": 1083
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966652094281214e-05,
+      "loss": 0.3884,
+      "step": 1084
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966567876314728e-05,
+      "loss": 0.3657,
+      "step": 1085
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966483553946637e-05,
+      "loss": 0.3185,
+      "step": 1086
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.966399127186049e-05,
+      "loss": 0.3831,
+      "step": 1087
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9663145960420828e-05,
+      "loss": 0.3717,
+      "step": 1088
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9662299605238692e-05,
+      "loss": 0.3737,
+      "step": 1089
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9661452206405497e-05,
+      "loss": 0.4205,
+      "step": 1090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9660603764012772e-05,
+      "loss": 0.3983,
+      "step": 1091
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9659754278152157e-05,
+      "loss": 0.3786,
+      "step": 1092
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9658903748915416e-05,
+      "loss": 0.3754,
+      "step": 1093
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9658052176394407e-05,
+      "loss": 0.3338,
+      "step": 1094
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.965719956068112e-05,
+      "loss": 0.3849,
+      "step": 1095
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.965634590186764e-05,
+      "loss": 0.3911,
+      "step": 1096
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.965549120004618e-05,
+      "loss": 0.3135,
+      "step": 1097
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9654635455309055e-05,
+      "loss": 0.3346,
+      "step": 1098
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9653778667748695e-05,
+      "loss": 0.3457,
+      "step": 1099
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9652920837457644e-05,
+      "loss": 0.3732,
+      "step": 1100
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9652061964528565e-05,
+      "loss": 0.4146,
+      "step": 1101
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.965120204905422e-05,
+      "loss": 0.3563,
+      "step": 1102
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.965034109112749e-05,
+      "loss": 0.3668,
+      "step": 1103
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9649479090841378e-05,
+      "loss": 0.401,
+      "step": 1104
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9648616048288983e-05,
+      "loss": 0.3364,
+      "step": 1105
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.9647751963563526e-05,
+      "loss": 0.3693,
+      "step": 1106
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.964688683675834e-05,
+      "loss": 0.3281,
+      "step": 1107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.964602066796687e-05,
+      "loss": 0.3844,
+      "step": 1108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9645153457282674e-05,
+      "loss": 0.3725,
+      "step": 1109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9644285204799424e-05,
+      "loss": 0.3553,
+      "step": 1110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.964341591061089e-05,
+      "loss": 0.356,
+      "step": 1111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.964254557481098e-05,
+      "loss": 0.4114,
+      "step": 1112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.96416741974937e-05,
+      "loss": 0.3675,
+      "step": 1113
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.964080177875317e-05,
+      "loss": 0.3663,
+      "step": 1114
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9639928318683615e-05,
+      "loss": 0.3459,
+      "step": 1115
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9639053817379382e-05,
+      "loss": 0.3324,
+      "step": 1116
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9638178274934936e-05,
+      "loss": 0.4127,
+      "step": 1117
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.963730169144484e-05,
+      "loss": 0.3411,
+      "step": 1118
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9636424067003772e-05,
+      "loss": 0.3441,
+      "step": 1119
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.9635545401706536e-05,
+      "loss": 0.358,
+      "step": 1120
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9634665695648035e-05,
+      "loss": 0.3697,
+      "step": 1121
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9633784948923287e-05,
+      "loss": 0.405,
+      "step": 1122
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9632903161627427e-05,
+      "loss": 0.3586,
+      "step": 1123
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9632020333855694e-05,
+      "loss": 0.3686,
+      "step": 1124
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.963113646570345e-05,
+      "loss": 0.337,
+      "step": 1125
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.963025155726616e-05,
+      "loss": 0.3787,
+      "step": 1126
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962936560863941e-05,
+      "loss": 0.3679,
+      "step": 1127
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962847861991889e-05,
+      "loss": 0.4044,
+      "step": 1128
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.9627590591200406e-05,
+      "loss": 0.372,
+      "step": 1129
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962670152257988e-05,
+      "loss": 0.3454,
+      "step": 1130
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962581141415334e-05,
+      "loss": 0.3593,
+      "step": 1131
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.962492026601693e-05,
+      "loss": 0.3766,
+      "step": 1132
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.96240280782669e-05,
+      "loss": 0.4256,
+      "step": 1133
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.962313485099962e-05,
+      "loss": 0.3773,
+      "step": 1134
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9622240584311577e-05,
+      "loss": 0.3155,
+      "step": 1135
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9621345278299357e-05,
+      "loss": 0.3864,
+      "step": 1136
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9620448933059667e-05,
+      "loss": 0.3532,
+      "step": 1137
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.961955154868932e-05,
+      "loss": 0.3674,
+      "step": 1138
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9618653125285247e-05,
+      "loss": 0.3852,
+      "step": 1139
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.961775366294449e-05,
+      "loss": 0.342,
+      "step": 1140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.96168531617642e-05,
+      "loss": 0.3892,
+      "step": 1141
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9615951621841648e-05,
+      "loss": 0.3874,
+      "step": 1142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9615049043274207e-05,
+      "loss": 0.3723,
+      "step": 1143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.9614145426159364e-05,
+      "loss": 0.3551,
+      "step": 1144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.961324077059473e-05,
+      "loss": 0.3632,
+      "step": 1145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.961233507667801e-05,
+      "loss": 0.39,
+      "step": 1146
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9611428344507036e-05,
+      "loss": 0.3288,
+      "step": 1147
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9610520574179748e-05,
+      "loss": 0.3435,
+      "step": 1148
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.960961176579419e-05,
+      "loss": 0.3816,
+      "step": 1149
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9608701919448532e-05,
+      "loss": 0.4018,
+      "step": 1150
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9607791035241044e-05,
+      "loss": 0.3424,
+      "step": 1151
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9606879113270114e-05,
+      "loss": 0.3494,
+      "step": 1152
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.9605966153634245e-05,
+      "loss": 0.3392,
+      "step": 1153
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9605052156432042e-05,
+      "loss": 0.3996,
+      "step": 1154
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9604137121762235e-05,
+      "loss": 0.3476,
+      "step": 1155
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9603221049723653e-05,
+      "loss": 0.3959,
+      "step": 1156
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9602303940415252e-05,
+      "loss": 0.3549,
+      "step": 1157
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9601385793936078e-05,
+      "loss": 0.3453,
+      "step": 1158
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9600466610385315e-05,
+      "loss": 0.3155,
+      "step": 1159
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.959954638986224e-05,
+      "loss": 0.3791,
+      "step": 1160
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.959862513246625e-05,
+      "loss": 0.3874,
+      "step": 1161
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9597702838296855e-05,
+      "loss": 0.3822,
+      "step": 1162
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.959677950745367e-05,
+      "loss": 0.373,
+      "step": 1163
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9595855140036427e-05,
+      "loss": 0.3253,
+      "step": 1164
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.3217,
+      "step": 1165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.9594003295879265e-05,
+      "loss": 0.3265,
+      "step": 1166
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9593075819339363e-05,
+      "loss": 0.3825,
+      "step": 1167
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.959214730662545e-05,
+      "loss": 0.3392,
+      "step": 1168
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.959121775783782e-05,
+      "loss": 0.3888,
+      "step": 1169
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.959028717307687e-05,
+      "loss": 0.3895,
+      "step": 1170
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.958935555244312e-05,
+      "loss": 0.3701,
+      "step": 1171
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9588422896037195e-05,
+      "loss": 0.3631,
+      "step": 1172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9587489203959837e-05,
+      "loss": 0.3992,
+      "step": 1173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.958655447631189e-05,
+      "loss": 0.3542,
+      "step": 1174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.958561871319432e-05,
+      "loss": 0.3984,
+      "step": 1175
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.95846819147082e-05,
+      "loss": 0.3934,
+      "step": 1176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9583744080954717e-05,
+      "loss": 0.3899,
+      "step": 1177
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9582805212035175e-05,
+      "loss": 0.3307,
+      "step": 1178
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.9581865308050972e-05,
+      "loss": 0.3835,
+      "step": 1179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.958092436910364e-05,
+      "loss": 0.3569,
+      "step": 1180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9579982395294806e-05,
+      "loss": 0.3102,
+      "step": 1181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9579039386726214e-05,
+      "loss": 0.3508,
+      "step": 1182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.957809534349973e-05,
+      "loss": 0.3782,
+      "step": 1183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.957715026571731e-05,
+      "loss": 0.3271,
+      "step": 1184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.957620415348105e-05,
+      "loss": 0.3428,
+      "step": 1185
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9575257006893125e-05,
+      "loss": 0.3264,
+      "step": 1186
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9574308826055853e-05,
+      "loss": 0.3475,
+      "step": 1187
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9573359611071638e-05,
+      "loss": 0.3302,
+      "step": 1188
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9572409362043017e-05,
+      "loss": 0.3698,
+      "step": 1189
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9571458079072622e-05,
+      "loss": 0.3406,
+      "step": 1190
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.957050576226321e-05,
+      "loss": 0.3252,
+      "step": 1191
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.956955241171764e-05,
+      "loss": 0.3351,
+      "step": 1192
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9568598027538883e-05,
+      "loss": 0.3554,
+      "step": 1193
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9567642609830027e-05,
+      "loss": 0.362,
+      "step": 1194
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9566686158694274e-05,
+      "loss": 0.345,
+      "step": 1195
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9565728674234924e-05,
+      "loss": 0.3557,
+      "step": 1196
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9564770156555403e-05,
+      "loss": 0.3434,
+      "step": 1197
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9563810605759242e-05,
+      "loss": 0.4019,
+      "step": 1198
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9562850021950085e-05,
+      "loss": 0.3753,
+      "step": 1199
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.956188840523169e-05,
+      "loss": 0.3745,
+      "step": 1200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9560925755707918e-05,
+      "loss": 0.3574,
+      "step": 1201
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955996207348275e-05,
+      "loss": 0.3638,
+      "step": 1202
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9558997358660278e-05,
+      "loss": 0.3504,
+      "step": 1203
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9558031611344698e-05,
+      "loss": 0.3756,
+      "step": 1204
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.955706483164033e-05,
+      "loss": 0.3698,
+      "step": 1205
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9556097019651594e-05,
+      "loss": 0.347,
+      "step": 1206
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9555128175483026e-05,
+      "loss": 0.3512,
+      "step": 1207
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.955415829923928e-05,
+      "loss": 0.3481,
+      "step": 1208
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9553187391025102e-05,
+      "loss": 0.3255,
+      "step": 1209
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9552215450945372e-05,
+      "loss": 0.3699,
+      "step": 1210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.955124247910507e-05,
+      "loss": 0.3645,
+      "step": 1211
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9550268475609292e-05,
+      "loss": 0.3659,
+      "step": 1212
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9549293440563236e-05,
+      "loss": 0.3795,
+      "step": 1213
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9548317374072227e-05,
+      "loss": 0.3907,
+      "step": 1214
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.954734027624168e-05,
+      "loss": 0.369,
+      "step": 1215
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9546362147177147e-05,
+      "loss": 0.3141,
+      "step": 1216
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.954538298698427e-05,
+      "loss": 0.3695,
+      "step": 1217
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9544402795768817e-05,
+      "loss": 0.3377,
+      "step": 1218
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9543421573636656e-05,
+      "loss": 0.3733,
+      "step": 1219
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.954243932069377e-05,
+      "loss": 0.3656,
+      "step": 1220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9541456037046262e-05,
+      "loss": 0.4423,
+      "step": 1221
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9540471722800334e-05,
+      "loss": 0.3972,
+      "step": 1222
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9539486378062307e-05,
+      "loss": 0.3932,
+      "step": 1223
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9538500002938607e-05,
+      "loss": 0.3609,
+      "step": 1224
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9537512597535778e-05,
+      "loss": 0.4089,
+      "step": 1225
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.953652416196047e-05,
+      "loss": 0.342,
+      "step": 1226
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.953553469631945e-05,
+      "loss": 0.3212,
+      "step": 1227
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9534544200719593e-05,
+      "loss": 0.3738,
+      "step": 1228
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.953355267526788e-05,
+      "loss": 0.3753,
+      "step": 1229
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9532560120071414e-05,
+      "loss": 0.3857,
+      "step": 1230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9531566535237397e-05,
+      "loss": 0.3987,
+      "step": 1231
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9530571920873156e-05,
+      "loss": 0.3245,
+      "step": 1232
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9529576277086118e-05,
+      "loss": 0.3625,
+      "step": 1233
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9528579603983827e-05,
+      "loss": 0.3725,
+      "step": 1234
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9527581901673932e-05,
+      "loss": 0.357,
+      "step": 1235
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9526583170264207e-05,
+      "loss": 0.3558,
+      "step": 1236
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9525583409862518e-05,
+      "loss": 0.3783,
+      "step": 1237
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9524582620576856e-05,
+      "loss": 0.3561,
+      "step": 1238
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9523580802515317e-05,
+      "loss": 0.3646,
+      "step": 1239
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.952257795578611e-05,
+      "loss": 0.3539,
+      "step": 1240
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.952157408049756e-05,
+      "loss": 0.4017,
+      "step": 1241
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9520569176758096e-05,
+      "loss": 0.3472,
+      "step": 1242
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9519563244676255e-05,
+      "loss": 0.3582,
+      "step": 1243
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9518556284360696e-05,
+      "loss": 0.3917,
+      "step": 1244
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9517548295920185e-05,
+      "loss": 0.352,
+      "step": 1245
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9516539279463595e-05,
+      "loss": 0.3585,
+      "step": 1246
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.951552923509991e-05,
+      "loss": 0.3114,
+      "step": 1247
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9514518162938226e-05,
+      "loss": 0.3576,
+      "step": 1248
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9513506063087758e-05,
+      "loss": 0.3295,
+      "step": 1249
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9512492935657826e-05,
+      "loss": 0.3463,
+      "step": 1250
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9511478780757858e-05,
+      "loss": 0.3609,
+      "step": 1251
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9510463598497395e-05,
+      "loss": 0.2955,
+      "step": 1252
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9509447388986086e-05,
+      "loss": 0.3766,
+      "step": 1253
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9508430152333702e-05,
+      "loss": 0.3758,
+      "step": 1254
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9507411888650114e-05,
+      "loss": 0.3433,
+      "step": 1255
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9506392598045307e-05,
+      "loss": 0.3326,
+      "step": 1256
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.950537228062938e-05,
+      "loss": 0.3342,
+      "step": 1257
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9504350936512534e-05,
+      "loss": 0.3344,
+      "step": 1258
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9503328565805093e-05,
+      "loss": 0.3433,
+      "step": 1259
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.950230516861748e-05,
+      "loss": 0.3861,
+      "step": 1260
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9501280745060244e-05,
+      "loss": 0.3692,
+      "step": 1261
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9500255295244034e-05,
+      "loss": 0.318,
+      "step": 1262
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9499228819279602e-05,
+      "loss": 0.369,
+      "step": 1263
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.949820131727783e-05,
+      "loss": 0.3856,
+      "step": 1264
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9497172789349696e-05,
+      "loss": 0.3591,
+      "step": 1265
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9496143235606296e-05,
+      "loss": 0.3831,
+      "step": 1266
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9495112656158838e-05,
+      "loss": 0.3902,
+      "step": 1267
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9494081051118633e-05,
+      "loss": 0.3289,
+      "step": 1268
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9493048420597106e-05,
+      "loss": 0.3355,
+      "step": 1269
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9492014764705804e-05,
+      "loss": 0.3288,
+      "step": 1270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9490980083556364e-05,
+      "loss": 0.3846,
+      "step": 1271
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.948994437726055e-05,
+      "loss": 0.4037,
+      "step": 1272
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.948890764593023e-05,
+      "loss": 0.3311,
+      "step": 1273
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9487869889677386e-05,
+      "loss": 0.3904,
+      "step": 1274
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9486831108614106e-05,
+      "loss": 0.3373,
+      "step": 1275
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9485791302852593e-05,
+      "loss": 0.3328,
+      "step": 1276
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.948475047250516e-05,
+      "loss": 0.3453,
+      "step": 1277
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9483708617684228e-05,
+      "loss": 0.3313,
+      "step": 1278
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9482665738502334e-05,
+      "loss": 0.397,
+      "step": 1279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9481621835072118e-05,
+      "loss": 0.3734,
+      "step": 1280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9480576907506338e-05,
+      "loss": 0.3308,
+      "step": 1281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9479530955917858e-05,
+      "loss": 0.3543,
+      "step": 1282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9478483980419654e-05,
+      "loss": 0.3836,
+      "step": 1283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9477435981124812e-05,
+      "loss": 0.3437,
+      "step": 1284
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.947638695814653e-05,
+      "loss": 0.3606,
+      "step": 1285
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9475336911598117e-05,
+      "loss": 0.367,
+      "step": 1286
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9474285841592993e-05,
+      "loss": 0.323,
+      "step": 1287
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.947323374824468e-05,
+      "loss": 0.3354,
+      "step": 1288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9472180631666823e-05,
+      "loss": 0.3739,
+      "step": 1289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.947112649197317e-05,
+      "loss": 0.3635,
+      "step": 1290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.947007132927758e-05,
+      "loss": 0.3435,
+      "step": 1291
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.946901514369403e-05,
+      "loss": 0.341,
+      "step": 1292
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9467957935336596e-05,
+      "loss": 0.3665,
+      "step": 1293
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9466899704319472e-05,
+      "loss": 0.3678,
+      "step": 1294
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.946584045075696e-05,
+      "loss": 0.3619,
+      "step": 1295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9464780174763472e-05,
+      "loss": 0.3606,
+      "step": 1296
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.946371887645353e-05,
+      "loss": 0.3664,
+      "step": 1297
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9462656555941773e-05,
+      "loss": 0.3336,
+      "step": 1298
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9461593213342938e-05,
+      "loss": 0.3896,
+      "step": 1299
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9460528848771883e-05,
+      "loss": 0.3584,
+      "step": 1300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9459463462343577e-05,
+      "loss": 0.3876,
+      "step": 1301
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.945839705417309e-05,
+      "loss": 0.3304,
+      "step": 1302
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.945732962437561e-05,
+      "loss": 0.3289,
+      "step": 1303
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.945626117306643e-05,
+      "loss": 0.3625,
+      "step": 1304
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.945519170036096e-05,
+      "loss": 0.3492,
+      "step": 1305
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9454121206374714e-05,
+      "loss": 0.3485,
+      "step": 1306
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.945304969122332e-05,
+      "loss": 0.3859,
+      "step": 1307
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9451977155022513e-05,
+      "loss": 0.3921,
+      "step": 1308
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9450903597888145e-05,
+      "loss": 0.3685,
+      "step": 1309
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.944982901993617e-05,
+      "loss": 0.3532,
+      "step": 1310
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.944875342128266e-05,
+      "loss": 0.3092,
+      "step": 1311
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.944767680204379e-05,
+      "loss": 0.3518,
+      "step": 1312
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9446599162335845e-05,
+      "loss": 0.338,
+      "step": 1313
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9445520502275234e-05,
+      "loss": 0.3408,
+      "step": 1314
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9444440821978457e-05,
+      "loss": 0.2365,
+      "step": 1315
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9443360121562138e-05,
+      "loss": 0.2188,
+      "step": 1316
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9442278401143e-05,
+      "loss": 0.2213,
+      "step": 1317
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9441195660837896e-05,
+      "loss": 0.2252,
+      "step": 1318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9440111900763758e-05,
+      "loss": 0.2435,
+      "step": 1319
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.943902712103766e-05,
+      "loss": 0.2268,
+      "step": 1320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9437941321776766e-05,
+      "loss": 0.2443,
+      "step": 1321
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9436854503098357e-05,
+      "loss": 0.2375,
+      "step": 1322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9435766665119823e-05,
+      "loss": 0.2421,
+      "step": 1323
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9434677807958663e-05,
+      "loss": 0.2067,
+      "step": 1324
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.943358793173249e-05,
+      "loss": 0.2205,
+      "step": 1325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9432497036559022e-05,
+      "loss": 0.2051,
+      "step": 1326
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.943140512255609e-05,
+      "loss": 0.2635,
+      "step": 1327
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9430312189841638e-05,
+      "loss": 0.2167,
+      "step": 1328
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.942921823853371e-05,
+      "loss": 0.2415,
+      "step": 1329
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9428123268750477e-05,
+      "loss": 0.2609,
+      "step": 1330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9427027280610197e-05,
+      "loss": 0.2454,
+      "step": 1331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9425930274231262e-05,
+      "loss": 0.2188,
+      "step": 1332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9424832249732155e-05,
+      "loss": 0.2393,
+      "step": 1333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.942373320723148e-05,
+      "loss": 0.2421,
+      "step": 1334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9422633146847943e-05,
+      "loss": 0.2238,
+      "step": 1335
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9421532068700374e-05,
+      "loss": 0.271,
+      "step": 1336
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9420429972907695e-05,
+      "loss": 0.2674,
+      "step": 1337
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9419326859588947e-05,
+      "loss": 0.2289,
+      "step": 1338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9418222728863286e-05,
+      "loss": 0.2298,
+      "step": 1339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9417117580849964e-05,
+      "loss": 0.307,
+      "step": 1340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.941601141566836e-05,
+      "loss": 0.2106,
+      "step": 1341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9414904233437947e-05,
+      "loss": 0.2362,
+      "step": 1342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9413796034278315e-05,
+      "loss": 0.2416,
+      "step": 1343
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.941268681830917e-05,
+      "loss": 0.212,
+      "step": 1344
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9411576585650313e-05,
+      "loss": 0.2114,
+      "step": 1345
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9410465336421667e-05,
+      "loss": 0.2174,
+      "step": 1346
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9409353070743262e-05,
+      "loss": 0.2164,
+      "step": 1347
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9408239788735235e-05,
+      "loss": 0.2015,
+      "step": 1348
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.940712549051784e-05,
+      "loss": 0.2333,
+      "step": 1349
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9406010176211426e-05,
+      "loss": 0.2478,
+      "step": 1350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.940489384593647e-05,
+      "loss": 0.2429,
+      "step": 1351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9403776499813546e-05,
+      "loss": 0.2434,
+      "step": 1352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.940265813796334e-05,
+      "loss": 0.2537,
+      "step": 1353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9401538760506646e-05,
+      "loss": 0.2114,
+      "step": 1354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.940041836756438e-05,
+      "loss": 0.2624,
+      "step": 1355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9399296959257557e-05,
+      "loss": 0.247,
+      "step": 1356
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9398174535707298e-05,
+      "loss": 0.221,
+      "step": 1357
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.939705109703484e-05,
+      "loss": 0.2528,
+      "step": 1358
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9395926643361533e-05,
+      "loss": 0.2412,
+      "step": 1359
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.939480117480883e-05,
+      "loss": 0.2638,
+      "step": 1360
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9393674691498294e-05,
+      "loss": 0.2406,
+      "step": 1361
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9392547193551602e-05,
+      "loss": 0.2375,
+      "step": 1362
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9391418681090538e-05,
+      "loss": 0.2416,
+      "step": 1363
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.939028915423699e-05,
+      "loss": 0.214,
+      "step": 1364
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9389158613112973e-05,
+      "loss": 0.2519,
+      "step": 1365
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.938802705784059e-05,
+      "loss": 0.2542,
+      "step": 1366
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.9386894488542064e-05,
+      "loss": 0.2507,
+      "step": 1367
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.938576090533973e-05,
+      "loss": 0.2617,
+      "step": 1368
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.938462630835603e-05,
+      "loss": 0.2291,
+      "step": 1369
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9383490697713515e-05,
+      "loss": 0.2076,
+      "step": 1370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.938235407353484e-05,
+      "loss": 0.2453,
+      "step": 1371
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9381216435942783e-05,
+      "loss": 0.2484,
+      "step": 1372
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9380077785060213e-05,
+      "loss": 0.2479,
+      "step": 1373
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9378938121010128e-05,
+      "loss": 0.2106,
+      "step": 1374
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9377797443915623e-05,
+      "loss": 0.2414,
+      "step": 1375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9376655753899903e-05,
+      "loss": 0.2424,
+      "step": 1376
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9375513051086292e-05,
+      "loss": 0.2115,
+      "step": 1377
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9374369335598212e-05,
+      "loss": 0.2372,
+      "step": 1378
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9373224607559195e-05,
+      "loss": 0.242,
+      "step": 1379
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.937207886709289e-05,
+      "loss": 0.2292,
+      "step": 1380
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.937093211432305e-05,
+      "loss": 0.2391,
+      "step": 1381
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.9369784349373544e-05,
+      "loss": 0.2505,
+      "step": 1382
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.936863557236834e-05,
+      "loss": 0.2317,
+      "step": 1383
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.936748578343152e-05,
+      "loss": 0.2194,
+      "step": 1384
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.936633498268728e-05,
+      "loss": 0.2563,
+      "step": 1385
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9365183170259917e-05,
+      "loss": 0.2417,
+      "step": 1386
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9364030346273847e-05,
+      "loss": 0.2269,
+      "step": 1387
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.936287651085358e-05,
+      "loss": 0.2482,
+      "step": 1388
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.9361721664123757e-05,
+      "loss": 0.2422,
+      "step": 1389
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9360565806209103e-05,
+      "loss": 0.2359,
+      "step": 1390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.935940893723448e-05,
+      "loss": 0.2507,
+      "step": 1391
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.935825105732483e-05,
+      "loss": 0.2175,
+      "step": 1392
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.935709216660523e-05,
+      "loss": 0.2328,
+      "step": 1393
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.935593226520085e-05,
+      "loss": 0.2457,
+      "step": 1394
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9354771353236968e-05,
+      "loss": 0.2209,
+      "step": 1395
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9353609430838987e-05,
+      "loss": 0.2326,
+      "step": 1396
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9352446498132413e-05,
+      "loss": 0.2216,
+      "step": 1397
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9351282555242843e-05,
+      "loss": 0.2226,
+      "step": 1398
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.9350117602296007e-05,
+      "loss": 0.2242,
+      "step": 1399
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.934895163941773e-05,
+      "loss": 0.2148,
+      "step": 1400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.934778466673396e-05,
+      "loss": 0.2277,
+      "step": 1401
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.934661668437073e-05,
+      "loss": 0.2327,
+      "step": 1402
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9345447692454213e-05,
+      "loss": 0.2377,
+      "step": 1403
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9344277691110666e-05,
+      "loss": 0.2281,
+      "step": 1404
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.934310668046646e-05,
+      "loss": 0.2303,
+      "step": 1405
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9341934660648092e-05,
+      "loss": 0.2297,
+      "step": 1406
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9340761631782142e-05,
+      "loss": 0.2369,
+      "step": 1407
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.933958759399532e-05,
+      "loss": 0.2355,
+      "step": 1408
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9338412547414433e-05,
+      "loss": 0.2233,
+      "step": 1409
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9337236492166404e-05,
+      "loss": 0.2407,
+      "step": 1410
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9336059428378264e-05,
+      "loss": 0.2471,
+      "step": 1411
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.933488135617714e-05,
+      "loss": 0.2324,
+      "step": 1412
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9333702275690288e-05,
+      "loss": 0.2231,
+      "step": 1413
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9332522187045063e-05,
+      "loss": 0.2344,
+      "step": 1414
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.9331341090368927e-05,
+      "loss": 0.2315,
+      "step": 1415
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9330158985789455e-05,
+      "loss": 0.2734,
+      "step": 1416
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.932897587343433e-05,
+      "loss": 0.2116,
+      "step": 1417
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.932779175343134e-05,
+      "loss": 0.2604,
+      "step": 1418
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9326606625908385e-05,
+      "loss": 0.2063,
+      "step": 1419
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.932542049099348e-05,
+      "loss": 0.2379,
+      "step": 1420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9324233348814738e-05,
+      "loss": 0.2588,
+      "step": 1421
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9323045199500387e-05,
+      "loss": 0.2454,
+      "step": 1422
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9321856043178756e-05,
+      "loss": 0.2212,
+      "step": 1423
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.93206658799783e-05,
+      "loss": 0.2187,
+      "step": 1424
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.9319474710027563e-05,
+      "loss": 0.2102,
+      "step": 1425
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.931828253345521e-05,
+      "loss": 0.2352,
+      "step": 1426
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.931708935039001e-05,
+      "loss": 0.2472,
+      "step": 1427
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.931589516096084e-05,
+      "loss": 0.2089,
+      "step": 1428
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9314699965296694e-05,
+      "loss": 0.2232,
+      "step": 1429
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.931350376352666e-05,
+      "loss": 0.2167,
+      "step": 1430
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9312306555779955e-05,
+      "loss": 0.2279,
+      "step": 1431
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9311108342185882e-05,
+      "loss": 0.2572,
+      "step": 1432
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9309909122873863e-05,
+      "loss": 0.234,
+      "step": 1433
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9308708897973433e-05,
+      "loss": 0.2128,
+      "step": 1434
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.9307507667614232e-05,
+      "loss": 0.2731,
+      "step": 1435
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9306305431926005e-05,
+      "loss": 0.2538,
+      "step": 1436
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9305102191038613e-05,
+      "loss": 0.2502,
+      "step": 1437
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.930389794508202e-05,
+      "loss": 0.2245,
+      "step": 1438
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9302692694186295e-05,
+      "loss": 0.2373,
+      "step": 1439
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9301486438481628e-05,
+      "loss": 0.2309,
+      "step": 1440
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9300279178098304e-05,
+      "loss": 0.2476,
+      "step": 1441
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.9299070913166723e-05,
+      "loss": 0.2303,
+      "step": 1442
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9297861643817397e-05,
+      "loss": 0.2488,
+      "step": 1443
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.929665137018094e-05,
+      "loss": 0.2392,
+      "step": 1444
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9295440092388077e-05,
+      "loss": 0.2179,
+      "step": 1445
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.929422781056964e-05,
+      "loss": 0.223,
+      "step": 1446
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.929301452485657e-05,
+      "loss": 0.2592,
+      "step": 1447
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.9291800235379923e-05,
+      "loss": 0.2268,
+      "step": 1448
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9290584942270853e-05,
+      "loss": 0.2404,
+      "step": 1449
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9289368645660626e-05,
+      "loss": 0.2281,
+      "step": 1450
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9288151345680623e-05,
+      "loss": 0.2537,
+      "step": 1451
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9286933042462324e-05,
+      "loss": 0.2608,
+      "step": 1452
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.928571373613732e-05,
+      "loss": 0.2492,
+      "step": 1453
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9284493426837313e-05,
+      "loss": 0.234,
+      "step": 1454
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9283272114694115e-05,
+      "loss": 0.2534,
+      "step": 1455
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9282049799839636e-05,
+      "loss": 0.2351,
+      "step": 1456
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.928082648240591e-05,
+      "loss": 0.2378,
+      "step": 1457
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9279602162525063e-05,
+      "loss": 0.2361,
+      "step": 1458
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9278376840329345e-05,
+      "loss": 0.2251,
+      "step": 1459
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.92771505159511e-05,
+      "loss": 0.2394,
+      "step": 1460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.9275923189522788e-05,
+      "loss": 0.2311,
+      "step": 1461
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.927469486117698e-05,
+      "loss": 0.234,
+      "step": 1462
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9273465531046343e-05,
+      "loss": 0.2435,
+      "step": 1463
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9272235199263668e-05,
+      "loss": 0.2539,
+      "step": 1464
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.927100386596184e-05,
+      "loss": 0.2418,
+      "step": 1465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9269771531273864e-05,
+      "loss": 0.2262,
+      "step": 1466
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.926853819533285e-05,
+      "loss": 0.2298,
+      "step": 1467
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9267303858272003e-05,
+      "loss": 0.2435,
+      "step": 1468
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9266068520224655e-05,
+      "loss": 0.2544,
+      "step": 1469
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.926483218132424e-05,
+      "loss": 0.2608,
+      "step": 1470
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9263594841704297e-05,
+      "loss": 0.2359,
+      "step": 1471
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.926235650149847e-05,
+      "loss": 0.2455,
+      "step": 1472
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9261117160840516e-05,
+      "loss": 0.2362,
+      "step": 1473
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.9259876819864304e-05,
+      "loss": 0.1991,
+      "step": 1474
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9258635478703805e-05,
+      "loss": 0.2468,
+      "step": 1475
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.92573931374931e-05,
+      "loss": 0.2396,
+      "step": 1476
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925614979636637e-05,
+      "loss": 0.221,
+      "step": 1477
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9254905455457925e-05,
+      "loss": 0.2269,
+      "step": 1478
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9253660114902163e-05,
+      "loss": 0.2132,
+      "step": 1479
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.925241377483359e-05,
+      "loss": 0.2198,
+      "step": 1480
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.9251166435386837e-05,
+      "loss": 0.2283,
+      "step": 1481
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.924991809669663e-05,
+      "loss": 0.2411,
+      "step": 1482
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.92486687588978e-05,
+      "loss": 0.263,
+      "step": 1483
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.92474184221253e-05,
+      "loss": 0.2231,
+      "step": 1484
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9246167086514175e-05,
+      "loss": 0.2515,
+      "step": 1485
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9244914752199587e-05,
+      "loss": 0.2203,
+      "step": 1486
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9243661419316806e-05,
+      "loss": 0.2205,
+      "step": 1487
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.9242407088001207e-05,
+      "loss": 0.258,
+      "step": 1488
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9241151758388273e-05,
+      "loss": 0.2618,
+      "step": 1489
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9239895430613595e-05,
+      "loss": 0.2609,
+      "step": 1490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.923863810481287e-05,
+      "loss": 0.2288,
+      "step": 1491
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.923737978112191e-05,
+      "loss": 0.2427,
+      "step": 1492
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.923612045967663e-05,
+      "loss": 0.233,
+      "step": 1493
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.923486014061305e-05,
+      "loss": 0.2132,
+      "step": 1494
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9233598824067302e-05,
+      "loss": 0.2414,
+      "step": 1495
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9232336510175626e-05,
+      "loss": 0.2657,
+      "step": 1496
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9231073199074358e-05,
+      "loss": 0.2308,
+      "step": 1497
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9229808890899962e-05,
+      "loss": 0.2606,
+      "step": 1498
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9228543585789e-05,
+      "loss": 0.2389,
+      "step": 1499
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9227277283878137e-05,
+      "loss": 0.242,
+      "step": 1500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.9226009985304147e-05,
+      "loss": 0.2318,
+      "step": 1501
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.922474169020392e-05,
+      "loss": 0.2371,
+      "step": 1502
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9223472398714444e-05,
+      "loss": 0.2233,
+      "step": 1503
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.922220211097282e-05,
+      "loss": 0.2385,
+      "step": 1504
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.922093082711626e-05,
+      "loss": 0.2284,
+      "step": 1505
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.921965854728207e-05,
+      "loss": 0.2418,
+      "step": 1506
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.9218385271607675e-05,
+      "loss": 0.2517,
+      "step": 1507
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.921711100023061e-05,
+      "loss": 0.2433,
+      "step": 1508
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9215835733288503e-05,
+      "loss": 0.2413,
+      "step": 1509
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9214559470919112e-05,
+      "loss": 0.2297,
+      "step": 1510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.921328221326028e-05,
+      "loss": 0.2504,
+      "step": 1511
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9212003960449968e-05,
+      "loss": 0.247,
+      "step": 1512
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9210724712626245e-05,
+      "loss": 0.2487,
+      "step": 1513
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.920944446992729e-05,
+      "loss": 0.2743,
+      "step": 1514
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9208163232491376e-05,
+      "loss": 0.2429,
+      "step": 1515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9206881000456902e-05,
+      "loss": 0.2521,
+      "step": 1516
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.920559777396236e-05,
+      "loss": 0.2699,
+      "step": 1517
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.920431355314636e-05,
+      "loss": 0.2265,
+      "step": 1518
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9203028338147603e-05,
+      "loss": 0.2317,
+      "step": 1519
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.9201742129104922e-05,
+      "loss": 0.2229,
+      "step": 1520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.920045492615724e-05,
+      "loss": 0.2587,
+      "step": 1521
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919916672944359e-05,
+      "loss": 0.2455,
+      "step": 1522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9197877539103106e-05,
+      "loss": 0.2433,
+      "step": 1523
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919658735527505e-05,
+      "loss": 0.2507,
+      "step": 1524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919529617809877e-05,
+      "loss": 0.2488,
+      "step": 1525
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.919400400771373e-05,
+      "loss": 0.2344,
+      "step": 1526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.9192710844259505e-05,
+      "loss": 0.2613,
+      "step": 1527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.919141668787577e-05,
+      "loss": 0.2697,
+      "step": 1528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.919012153870231e-05,
+      "loss": 0.2446,
+      "step": 1529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9188825396879022e-05,
+      "loss": 0.2526,
+      "step": 1530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9187528262545897e-05,
+      "loss": 0.2499,
+      "step": 1531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9186230135843053e-05,
+      "loss": 0.2448,
+      "step": 1532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9184931016910692e-05,
+      "loss": 0.2471,
+      "step": 1533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.9183630905889146e-05,
+      "loss": 0.2197,
+      "step": 1534
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.918232980291884e-05,
+      "loss": 0.2339,
+      "step": 1535
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.918102770814031e-05,
+      "loss": 0.2596,
+      "step": 1536
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.917972462169419e-05,
+      "loss": 0.2269,
+      "step": 1537
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9178420543721244e-05,
+      "loss": 0.2703,
+      "step": 1538
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.917711547436232e-05,
+      "loss": 0.2693,
+      "step": 1539
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9175809413758388e-05,
+      "loss": 0.2478,
+      "step": 1540
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9174502362050514e-05,
+      "loss": 0.2308,
+      "step": 1541
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9173194319379876e-05,
+      "loss": 0.263,
+      "step": 1542
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9171885285887763e-05,
+      "loss": 0.2451,
+      "step": 1543
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9170575261715566e-05,
+      "loss": 0.2579,
+      "step": 1544
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.9169264247004783e-05,
+      "loss": 0.2411,
+      "step": 1545
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.916795224189702e-05,
+      "loss": 0.2581,
+      "step": 1546
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.916663924653399e-05,
+      "loss": 0.2623,
+      "step": 1547
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9165325261057515e-05,
+      "loss": 0.2483,
+      "step": 1548
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9164010285609524e-05,
+      "loss": 0.2421,
+      "step": 1549
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9162694320332047e-05,
+      "loss": 0.2399,
+      "step": 1550
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9161377365367224e-05,
+      "loss": 0.22,
+      "step": 1551
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9160059420857306e-05,
+      "loss": 0.2465,
+      "step": 1552
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.915874048694465e-05,
+      "loss": 0.2584,
+      "step": 1553
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9157420563771715e-05,
+      "loss": 0.2445,
+      "step": 1554
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9156099651481065e-05,
+      "loss": 0.2242,
+      "step": 1555
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.915477775021538e-05,
+      "loss": 0.2405,
+      "step": 1556
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.915345486011744e-05,
+      "loss": 0.215,
+      "step": 1557
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.915213098133014e-05,
+      "loss": 0.2516,
+      "step": 1558
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9150806113996472e-05,
+      "loss": 0.2393,
+      "step": 1559
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.9149480258259535e-05,
+      "loss": 0.2537,
+      "step": 1560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.914815341426254e-05,
+      "loss": 0.2376,
+      "step": 1561
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9146825582148803e-05,
+      "loss": 0.2551,
+      "step": 1562
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9145496762061747e-05,
+      "loss": 0.2502,
+      "step": 1563
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9144166954144908e-05,
+      "loss": 0.2495,
+      "step": 1564
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9142836158541912e-05,
+      "loss": 0.2275,
+      "step": 1565
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.9141504375396505e-05,
+      "loss": 0.2355,
+      "step": 1566
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.914017160485254e-05,
+      "loss": 0.2162,
+      "step": 1567
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.913883784705397e-05,
+      "loss": 0.2614,
+      "step": 1568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9137503102144858e-05,
+      "loss": 0.2492,
+      "step": 1569
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9136167370269375e-05,
+      "loss": 0.2571,
+      "step": 1570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.913483065157179e-05,
+      "loss": 0.2561,
+      "step": 1571
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9133492946196498e-05,
+      "loss": 0.2581,
+      "step": 1572
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9132154254287982e-05,
+      "loss": 0.2764,
+      "step": 1573
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9130814575990835e-05,
+      "loss": 0.2498,
+      "step": 1574
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9129473911449762e-05,
+      "loss": 0.2424,
+      "step": 1575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.912813226080957e-05,
+      "loss": 0.2662,
+      "step": 1576
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.912678962421518e-05,
+      "loss": 0.2613,
+      "step": 1577
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9125446001811604e-05,
+      "loss": 0.2864,
+      "step": 1578
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.912410139374398e-05,
+      "loss": 0.2638,
+      "step": 1579
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9122755800157537e-05,
+      "loss": 0.2462,
+      "step": 1580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9121409221197617e-05,
+      "loss": 0.2649,
+      "step": 1581
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9120061657009667e-05,
+      "loss": 0.2354,
+      "step": 1582
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9118713107739246e-05,
+      "loss": 0.2565,
+      "step": 1583
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.911736357353201e-05,
+      "loss": 0.2399,
+      "step": 1584
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9116013054533723e-05,
+      "loss": 0.2572,
+      "step": 1585
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9114661550890264e-05,
+      "loss": 0.2561,
+      "step": 1586
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.911330906274761e-05,
+      "loss": 0.2683,
+      "step": 1587
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9111955590251853e-05,
+      "loss": 0.2749,
+      "step": 1588
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9110601133549173e-05,
+      "loss": 0.2556,
+      "step": 1589
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9109245692785876e-05,
+      "loss": 0.2657,
+      "step": 1590
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9107889268108368e-05,
+      "loss": 0.2577,
+      "step": 1591
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9106531859663156e-05,
+      "loss": 0.2547,
+      "step": 1592
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9105173467596862e-05,
+      "loss": 0.2559,
+      "step": 1593
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9103814092056205e-05,
+      "loss": 0.2399,
+      "step": 1594
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9102453733188016e-05,
+      "loss": 0.2416,
+      "step": 1595
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9101092391139233e-05,
+      "loss": 0.2436,
+      "step": 1596
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.90997300660569e-05,
+      "loss": 0.2154,
+      "step": 1597
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9098366758088156e-05,
+      "loss": 0.2664,
+      "step": 1598
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9097002467380267e-05,
+      "loss": 0.2353,
+      "step": 1599
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9095637194080587e-05,
+      "loss": 0.2422,
+      "step": 1600
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9094270938336583e-05,
+      "loss": 0.2403,
+      "step": 1601
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9092903700295832e-05,
+      "loss": 0.2239,
+      "step": 1602
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.909153548010601e-05,
+      "loss": 0.2593,
+      "step": 1603
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.90901662779149e-05,
+      "loss": 0.2607,
+      "step": 1604
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.90887960938704e-05,
+      "loss": 0.2212,
+      "step": 1605
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.90874249281205e-05,
+      "loss": 0.2526,
+      "step": 1606
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.908605278081331e-05,
+      "loss": 0.2396,
+      "step": 1607
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9084679652097035e-05,
+      "loss": 0.2894,
+      "step": 1608
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.908330554211999e-05,
+      "loss": 0.2297,
+      "step": 1609
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9081930451030597e-05,
+      "loss": 0.2249,
+      "step": 1610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9080554378977386e-05,
+      "loss": 0.2473,
+      "step": 1611
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9079177326108985e-05,
+      "loss": 0.2199,
+      "step": 1612
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.907779929257414e-05,
+      "loss": 0.2303,
+      "step": 1613
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9076420278521693e-05,
+      "loss": 0.2331,
+      "step": 1614
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.907504028410059e-05,
+      "loss": 0.24,
+      "step": 1615
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9073659309459897e-05,
+      "loss": 0.2639,
+      "step": 1616
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.907227735474877e-05,
+      "loss": 0.245,
+      "step": 1617
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9070894420116478e-05,
+      "loss": 0.2728,
+      "step": 1618
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.90695105057124e-05,
+      "loss": 0.2379,
+      "step": 1619
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.906812561168601e-05,
+      "loss": 0.2351,
+      "step": 1620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9066739738186904e-05,
+      "loss": 0.2553,
+      "step": 1621
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9065352885364763e-05,
+      "loss": 0.2346,
+      "step": 1622
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.906396505336939e-05,
+      "loss": 0.2529,
+      "step": 1623
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.906257624235069e-05,
+      "loss": 0.2511,
+      "step": 1624
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.906118645245867e-05,
+      "loss": 0.2281,
+      "step": 1625
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9059795683843444e-05,
+      "loss": 0.2413,
+      "step": 1626
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9058403936655235e-05,
+      "loss": 0.2493,
+      "step": 1627
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9057011211044365e-05,
+      "loss": 0.2822,
+      "step": 1628
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9055617507161273e-05,
+      "loss": 0.2618,
+      "step": 1629
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.905422282515649e-05,
+      "loss": 0.2415,
+      "step": 1630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9052827165180666e-05,
+      "loss": 0.2393,
+      "step": 1631
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9051430527384543e-05,
+      "loss": 0.2327,
+      "step": 1632
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.905003291191898e-05,
+      "loss": 0.2296,
+      "step": 1633
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9048634318934937e-05,
+      "loss": 0.2415,
+      "step": 1634
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.904723474858348e-05,
+      "loss": 0.2565,
+      "step": 1635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9045834201015775e-05,
+      "loss": 0.2438,
+      "step": 1636
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9044432676383108e-05,
+      "loss": 0.2644,
+      "step": 1637
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9043030174836852e-05,
+      "loss": 0.2232,
+      "step": 1638
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9041626696528503e-05,
+      "loss": 0.2362,
+      "step": 1639
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.904022224160965e-05,
+      "loss": 0.2655,
+      "step": 1640
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9038816810231993e-05,
+      "loss": 0.2463,
+      "step": 1641
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9037410402547342e-05,
+      "loss": 0.2625,
+      "step": 1642
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9036003018707597e-05,
+      "loss": 0.257,
+      "step": 1643
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9034594658864774e-05,
+      "loss": 0.2413,
+      "step": 1644
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9033185323171003e-05,
+      "loss": 0.2739,
+      "step": 1645
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9031775011778503e-05,
+      "loss": 0.2223,
+      "step": 1646
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9030363724839604e-05,
+      "loss": 0.2515,
+      "step": 1647
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.902895146250675e-05,
+      "loss": 0.2498,
+      "step": 1648
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.902753822493248e-05,
+      "loss": 0.2456,
+      "step": 1649
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.902612401226944e-05,
+      "loss": 0.2392,
+      "step": 1650
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9024708824670383e-05,
+      "loss": 0.2468,
+      "step": 1651
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.902329266228817e-05,
+      "loss": 0.2684,
+      "step": 1652
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.902187552527576e-05,
+      "loss": 0.2614,
+      "step": 1653
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9020457413786228e-05,
+      "loss": 0.2485,
+      "step": 1654
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9019038327972742e-05,
+      "loss": 0.2429,
+      "step": 1655
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9017618267988585e-05,
+      "loss": 0.2301,
+      "step": 1656
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9016197233987137e-05,
+      "loss": 0.2381,
+      "step": 1657
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9014775226121894e-05,
+      "loss": 0.267,
+      "step": 1658
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9013352244546447e-05,
+      "loss": 0.2446,
+      "step": 1659
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9011928289414502e-05,
+      "loss": 0.2512,
+      "step": 1660
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9010503360879853e-05,
+      "loss": 0.2689,
+      "step": 1661
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.900907745909642e-05,
+      "loss": 0.2185,
+      "step": 1662
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9007650584218215e-05,
+      "loss": 0.262,
+      "step": 1663
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9006222736399356e-05,
+      "loss": 0.2243,
+      "step": 1664
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9004793915794075e-05,
+      "loss": 0.2405,
+      "step": 1665
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9003364122556696e-05,
+      "loss": 0.2689,
+      "step": 1666
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.900193335684166e-05,
+      "loss": 0.2302,
+      "step": 1667
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9000501618803502e-05,
+      "loss": 0.2665,
+      "step": 1668
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.899906890859688e-05,
+      "loss": 0.2724,
+      "step": 1669
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.899763522637653e-05,
+      "loss": 0.2571,
+      "step": 1670
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.899620057229732e-05,
+      "loss": 0.2623,
+      "step": 1671
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.89947649465142e-05,
+      "loss": 0.2753,
+      "step": 1672
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8993328349182244e-05,
+      "loss": 0.2321,
+      "step": 1673
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8991890780456618e-05,
+      "loss": 0.258,
+      "step": 1674
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8990452240492602e-05,
+      "loss": 0.2468,
+      "step": 1675
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8989012729445574e-05,
+      "loss": 0.2345,
+      "step": 1676
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8987572247471017e-05,
+      "loss": 0.2703,
+      "step": 1677
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.898613079472453e-05,
+      "loss": 0.2341,
+      "step": 1678
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8984688371361794e-05,
+      "loss": 0.2685,
+      "step": 1679
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8983244977538625e-05,
+      "loss": 0.2726,
+      "step": 1680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8981800613410918e-05,
+      "loss": 0.2521,
+      "step": 1681
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8980355279134682e-05,
+      "loss": 0.2514,
+      "step": 1682
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8978908974866036e-05,
+      "loss": 0.2655,
+      "step": 1683
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.89774617007612e-05,
+      "loss": 0.2442,
+      "step": 1684
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8976013456976492e-05,
+      "loss": 0.2554,
+      "step": 1685
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8974564243668348e-05,
+      "loss": 0.2321,
+      "step": 1686
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8973114060993295e-05,
+      "loss": 0.2509,
+      "step": 1687
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8971662909107973e-05,
+      "loss": 0.2401,
+      "step": 1688
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.897021078816913e-05,
+      "loss": 0.2485,
+      "step": 1689
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8968757698333606e-05,
+      "loss": 0.2771,
+      "step": 1690
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8967303639758356e-05,
+      "loss": 0.2419,
+      "step": 1691
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8965848612600437e-05,
+      "loss": 0.2287,
+      "step": 1692
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8964392617017013e-05,
+      "loss": 0.2422,
+      "step": 1693
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8962935653165347e-05,
+      "loss": 0.215,
+      "step": 1694
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8961477721202806e-05,
+      "loss": 0.262,
+      "step": 1695
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8960018821286873e-05,
+      "loss": 0.2342,
+      "step": 1696
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8958558953575123e-05,
+      "loss": 0.261,
+      "step": 1697
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8957098118225243e-05,
+      "loss": 0.2628,
+      "step": 1698
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.895563631539502e-05,
+      "loss": 0.2236,
+      "step": 1699
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8954173545242347e-05,
+      "loss": 0.2577,
+      "step": 1700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.895270980792522e-05,
+      "loss": 0.2367,
+      "step": 1701
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8951245103601744e-05,
+      "loss": 0.307,
+      "step": 1702
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8949779432430125e-05,
+      "loss": 0.2409,
+      "step": 1703
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8948312794568674e-05,
+      "loss": 0.2296,
+      "step": 1704
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.894684519017581e-05,
+      "loss": 0.2281,
+      "step": 1705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.894537661941005e-05,
+      "loss": 0.2512,
+      "step": 1706
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8943907082430013e-05,
+      "loss": 0.2598,
+      "step": 1707
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.894243657939443e-05,
+      "loss": 0.2763,
+      "step": 1708
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8940965110462144e-05,
+      "loss": 0.2636,
+      "step": 1709
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.893949267579208e-05,
+      "loss": 0.2489,
+      "step": 1710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8938019275543286e-05,
+      "loss": 0.2614,
+      "step": 1711
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8936544909874905e-05,
+      "loss": 0.276,
+      "step": 1712
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8935069578946192e-05,
+      "loss": 0.2262,
+      "step": 1713
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8933593282916494e-05,
+      "loss": 0.2406,
+      "step": 1714
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8932116021945277e-05,
+      "loss": 0.2436,
+      "step": 1715
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.89306377961921e-05,
+      "loss": 0.2694,
+      "step": 1716
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.892915860581663e-05,
+      "loss": 0.2395,
+      "step": 1717
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.892767845097864e-05,
+      "loss": 0.25,
+      "step": 1718
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8926197331838002e-05,
+      "loss": 0.2651,
+      "step": 1719
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8924715248554703e-05,
+      "loss": 0.243,
+      "step": 1720
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8923232201288818e-05,
+      "loss": 0.2345,
+      "step": 1721
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.892174819020054e-05,
+      "loss": 0.2547,
+      "step": 1722
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.892026321545016e-05,
+      "loss": 0.244,
+      "step": 1723
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8918777277198077e-05,
+      "loss": 0.2548,
+      "step": 1724
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8917290375604785e-05,
+      "loss": 0.2356,
+      "step": 1725
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8915802510830895e-05,
+      "loss": 0.2533,
+      "step": 1726
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.891431368303711e-05,
+      "loss": 0.2287,
+      "step": 1727
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8912823892384246e-05,
+      "loss": 0.2493,
+      "step": 1728
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8911333139033218e-05,
+      "loss": 0.2546,
+      "step": 1729
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8909841423145045e-05,
+      "loss": 0.2773,
+      "step": 1730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.890834874488085e-05,
+      "loss": 0.2712,
+      "step": 1731
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8906855104401866e-05,
+      "loss": 0.2509,
+      "step": 1732
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8905360501869424e-05,
+      "loss": 0.2738,
+      "step": 1733
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8903864937444955e-05,
+      "loss": 0.2116,
+      "step": 1734
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8902368411290007e-05,
+      "loss": 0.2275,
+      "step": 1735
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8900870923566215e-05,
+      "loss": 0.2649,
+      "step": 1736
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8899372474435335e-05,
+      "loss": 0.2241,
+      "step": 1737
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.889787306405921e-05,
+      "loss": 0.2461,
+      "step": 1738
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8896372692599804e-05,
+      "loss": 0.2528,
+      "step": 1739
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.889487136021917e-05,
+      "loss": 0.246,
+      "step": 1740
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8893369067079476e-05,
+      "loss": 0.2597,
+      "step": 1741
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8891865813342983e-05,
+      "loss": 0.2129,
+      "step": 1742
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8890361599172063e-05,
+      "loss": 0.2387,
+      "step": 1743
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8888856424729195e-05,
+      "loss": 0.2649,
+      "step": 1744
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8887350290176955e-05,
+      "loss": 0.2457,
+      "step": 1745
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.888584319567802e-05,
+      "loss": 0.2557,
+      "step": 1746
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8884335141395177e-05,
+      "loss": 0.25,
+      "step": 1747
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.888282612749132e-05,
+      "loss": 0.2689,
+      "step": 1748
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8881316154129438e-05,
+      "loss": 0.2486,
+      "step": 1749
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8879805221472626e-05,
+      "loss": 0.256,
+      "step": 1750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8878293329684084e-05,
+      "loss": 0.2629,
+      "step": 1751
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.887678047892712e-05,
+      "loss": 0.2718,
+      "step": 1752
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.887526666936514e-05,
+      "loss": 0.2523,
+      "step": 1753
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8873751901161647e-05,
+      "loss": 0.2501,
+      "step": 1754
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8872236174480266e-05,
+      "loss": 0.2529,
+      "step": 1755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8870719489484708e-05,
+      "loss": 0.2484,
+      "step": 1756
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8869201846338795e-05,
+      "loss": 0.2532,
+      "step": 1757
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8867683245206452e-05,
+      "loss": 0.2309,
+      "step": 1758
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.886616368625171e-05,
+      "loss": 0.2606,
+      "step": 1759
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8864643169638697e-05,
+      "loss": 0.24,
+      "step": 1760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8863121695531648e-05,
+      "loss": 0.2661,
+      "step": 1761
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8861599264094908e-05,
+      "loss": 0.2827,
+      "step": 1762
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.886007587549291e-05,
+      "loss": 0.269,
+      "step": 1763
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.8858551529890207e-05,
+      "loss": 0.2481,
+      "step": 1764
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8857026227451437e-05,
+      "loss": 0.2465,
+      "step": 1765
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8855499968341364e-05,
+      "loss": 0.2449,
+      "step": 1766
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8853972752724837e-05,
+      "loss": 0.2578,
+      "step": 1767
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.885244458076682e-05,
+      "loss": 0.2467,
+      "step": 1768
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.8850915452632368e-05,
+      "loss": 0.2548,
+      "step": 1769
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.884938536848665e-05,
+      "loss": 0.2429,
+      "step": 1770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8847854328494932e-05,
+      "loss": 0.2261,
+      "step": 1771
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8846322332822592e-05,
+      "loss": 0.2512,
+      "step": 1772
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.88447893816351e-05,
+      "loss": 0.2452,
+      "step": 1773
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.884325547509803e-05,
+      "loss": 0.2632,
+      "step": 1774
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.884172061337707e-05,
+      "loss": 0.2671,
+      "step": 1775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8840184796638003e-05,
+      "loss": 0.2347,
+      "step": 1776
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.8838648025046718e-05,
+      "loss": 0.2504,
+      "step": 1777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8837110298769202e-05,
+      "loss": 0.2414,
+      "step": 1778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8835571617971556e-05,
+      "loss": 0.2337,
+      "step": 1779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8834031982819967e-05,
+      "loss": 0.2379,
+      "step": 1780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.883249139348074e-05,
+      "loss": 0.2494,
+      "step": 1781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.8830949850120284e-05,
+      "loss": 0.254,
+      "step": 1782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.88294073529051e-05,
+      "loss": 0.2792,
+      "step": 1783
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8827863902001793e-05,
+      "loss": 0.2831,
+      "step": 1784
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8826319497577082e-05,
+      "loss": 0.234,
+      "step": 1785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.882477413979778e-05,
+      "loss": 0.2531,
+      "step": 1786
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8823227828830802e-05,
+      "loss": 0.2238,
+      "step": 1787
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8821680564843174e-05,
+      "loss": 0.2447,
+      "step": 1788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.882013234800202e-05,
+      "loss": 0.2494,
+      "step": 1789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.881858317847456e-05,
+      "loss": 0.2646,
+      "step": 1790
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.881703305642814e-05,
+      "loss": 0.2298,
+      "step": 1791
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8815481982030176e-05,
+      "loss": 0.2579,
+      "step": 1792
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.881392995544821e-05,
+      "loss": 0.2501,
+      "step": 1793
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.881237697684988e-05,
+      "loss": 0.233,
+      "step": 1794
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.881082304640293e-05,
+      "loss": 0.2361,
+      "step": 1795
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.88092681642752e-05,
+      "loss": 0.2518,
+      "step": 1796
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8807712330634645e-05,
+      "loss": 0.2601,
+      "step": 1797
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8806155545649304e-05,
+      "loss": 0.2402,
+      "step": 1798
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8804597809487336e-05,
+      "loss": 0.2714,
+      "step": 1799
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8803039122317e-05,
+      "loss": 0.245,
+      "step": 1800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.880147948430664e-05,
+      "loss": 0.2502,
+      "step": 1801
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8799918895624732e-05,
+      "loss": 0.2493,
+      "step": 1802
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.879835735643983e-05,
+      "loss": 0.2296,
+      "step": 1803
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8796794866920606e-05,
+      "loss": 0.2436,
+      "step": 1804
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8795231427235823e-05,
+      "loss": 0.2358,
+      "step": 1805
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8793667037554357e-05,
+      "loss": 0.2746,
+      "step": 1806
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.879210169804518e-05,
+      "loss": 0.2772,
+      "step": 1807
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8790535408877367e-05,
+      "loss": 0.2507,
+      "step": 1808
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.8788968170220102e-05,
+      "loss": 0.2681,
+      "step": 1809
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.878739998224266e-05,
+      "loss": 0.2377,
+      "step": 1810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8785830845114432e-05,
+      "loss": 0.2545,
+      "step": 1811
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.87842607590049e-05,
+      "loss": 0.2616,
+      "step": 1812
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8782689724083654e-05,
+      "loss": 0.2458,
+      "step": 1813
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8781117740520386e-05,
+      "loss": 0.2258,
+      "step": 1814
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.8779544808484894e-05,
+      "loss": 0.2599,
+      "step": 1815
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.877797092814707e-05,
+      "loss": 0.2559,
+      "step": 1816
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8776396099676914e-05,
+      "loss": 0.242,
+      "step": 1817
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8774820323244527e-05,
+      "loss": 0.2286,
+      "step": 1818
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.877324359902011e-05,
+      "loss": 0.258,
+      "step": 1819
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.877166592717398e-05,
+      "loss": 0.2404,
+      "step": 1820
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8770087307876537e-05,
+      "loss": 0.2414,
+      "step": 1821
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.876850774129829e-05,
+      "loss": 0.2553,
+      "step": 1822
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.8766927227609857e-05,
+      "loss": 0.2787,
+      "step": 1823
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8765345766981955e-05,
+      "loss": 0.2419,
+      "step": 1824
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8763763359585395e-05,
+      "loss": 0.2583,
+      "step": 1825
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.8762180005591105e-05,
+      "loss": 0.2378,
+      "step": 1826
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.87605957051701e-05,
+      "loss": 0.2522,
+      "step": 1827
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.875901045849351e-05,
+      "loss": 0.2639,
+      "step": 1828
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.875742426573256e-05,
+      "loss": 0.2519,
+      "step": 1829
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.875583712705858e-05,
+      "loss": 0.2771,
+      "step": 1830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8754249042642998e-05,
+      "loss": 0.2348,
+      "step": 1831
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8752660012657347e-05,
+      "loss": 0.2375,
+      "step": 1832
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.875107003727327e-05,
+      "loss": 0.2298,
+      "step": 1833
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8749479116662495e-05,
+      "loss": 0.2523,
+      "step": 1834
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8747887250996867e-05,
+      "loss": 0.2529,
+      "step": 1835
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.8746294440448324e-05,
+      "loss": 0.2464,
+      "step": 1836
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8744700685188916e-05,
+      "loss": 0.2358,
+      "step": 1837
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8743105985390782e-05,
+      "loss": 0.2355,
+      "step": 1838
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8741510341226177e-05,
+      "loss": 0.2477,
+      "step": 1839
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8739913752867446e-05,
+      "loss": 0.2484,
+      "step": 1840
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8738316220487037e-05,
+      "loss": 0.2808,
+      "step": 1841
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.8736717744257513e-05,
+      "loss": 0.2564,
+      "step": 1842
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8735118324351523e-05,
+      "loss": 0.2637,
+      "step": 1843
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.873351796094183e-05,
+      "loss": 0.2549,
+      "step": 1844
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8731916654201285e-05,
+      "loss": 0.2746,
+      "step": 1845
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8730314404302857e-05,
+      "loss": 0.2408,
+      "step": 1846
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.872871121141961e-05,
+      "loss": 0.2411,
+      "step": 1847
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8727107075724703e-05,
+      "loss": 0.2539,
+      "step": 1848
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8725501997391404e-05,
+      "loss": 0.2613,
+      "step": 1849
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.872389597659309e-05,
+      "loss": 0.2471,
+      "step": 1850
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.872228901350322e-05,
+      "loss": 0.2505,
+      "step": 1851
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8720681108295375e-05,
+      "loss": 0.2597,
+      "step": 1852
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8719072261143227e-05,
+      "loss": 0.2672,
+      "step": 1853
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.871746247222055e-05,
+      "loss": 0.2409,
+      "step": 1854
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8715851741701226e-05,
+      "loss": 0.2672,
+      "step": 1855
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.871424006975923e-05,
+      "loss": 0.2528,
+      "step": 1856
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8712627456568643e-05,
+      "loss": 0.2798,
+      "step": 1857
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.871101390230365e-05,
+      "loss": 0.2577,
+      "step": 1858
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.870939940713853e-05,
+      "loss": 0.2466,
+      "step": 1859
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.870778397124768e-05,
+      "loss": 0.2632,
+      "step": 1860
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8706167594805577e-05,
+      "loss": 0.2519,
+      "step": 1861
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8704550277986815e-05,
+      "loss": 0.2647,
+      "step": 1862
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8702932020966086e-05,
+      "loss": 0.2535,
+      "step": 1863
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.870131282391818e-05,
+      "loss": 0.2556,
+      "step": 1864
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8699692687017986e-05,
+      "loss": 0.2455,
+      "step": 1865
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8698071610440508e-05,
+      "loss": 0.2712,
+      "step": 1866
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.869644959436084e-05,
+      "loss": 0.2364,
+      "step": 1867
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.869482663895418e-05,
+      "loss": 0.2799,
+      "step": 1868
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.869320274439583e-05,
+      "loss": 0.255,
+      "step": 1869
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8691577910861182e-05,
+      "loss": 0.2412,
+      "step": 1870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8689952138525748e-05,
+      "loss": 0.2373,
+      "step": 1871
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.868832542756513e-05,
+      "loss": 0.2649,
+      "step": 1872
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8686697778155034e-05,
+      "loss": 0.2501,
+      "step": 1873
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.8685069190471262e-05,
+      "loss": 0.26,
+      "step": 1874
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.868343966468973e-05,
+      "loss": 0.2507,
+      "step": 1875
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.868180920098644e-05,
+      "loss": 0.2607,
+      "step": 1876
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8680177799537508e-05,
+      "loss": 0.2749,
+      "step": 1877
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8678545460519142e-05,
+      "loss": 0.2628,
+      "step": 1878
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.867691218410766e-05,
+      "loss": 0.2378,
+      "step": 1879
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8675277970479472e-05,
+      "loss": 0.2638,
+      "step": 1880
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.8673642819811096e-05,
+      "loss": 0.2606,
+      "step": 1881
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.867200673227915e-05,
+      "loss": 0.2629,
+      "step": 1882
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8670369708060352e-05,
+      "loss": 0.2586,
+      "step": 1883
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.866873174733152e-05,
+      "loss": 0.2275,
+      "step": 1884
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8667092850269573e-05,
+      "loss": 0.2468,
+      "step": 1885
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8665453017051536e-05,
+      "loss": 0.2585,
+      "step": 1886
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.866381224785453e-05,
+      "loss": 0.263,
+      "step": 1887
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.8662170542855783e-05,
+      "loss": 0.2612,
+      "step": 1888
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.866052790223261e-05,
+      "loss": 0.2595,
+      "step": 1889
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.865888432616245e-05,
+      "loss": 0.2457,
+      "step": 1890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8657239814822817e-05,
+      "loss": 0.2606,
+      "step": 1891
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.865559436839135e-05,
+      "loss": 0.2692,
+      "step": 1892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8653947987045775e-05,
+      "loss": 0.2489,
+      "step": 1893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.865230067096392e-05,
+      "loss": 0.257,
+      "step": 1894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.8650652420323715e-05,
+      "loss": 0.2606,
+      "step": 1895
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8649003235303197e-05,
+      "loss": 0.2531,
+      "step": 1896
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8647353116080494e-05,
+      "loss": 0.2489,
+      "step": 1897
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8645702062833845e-05,
+      "loss": 0.2734,
+      "step": 1898
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8644050075741574e-05,
+      "loss": 0.2723,
+      "step": 1899
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.864239715498213e-05,
+      "loss": 0.2653,
+      "step": 1900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8640743300734044e-05,
+      "loss": 0.2621,
+      "step": 1901
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8639088513175948e-05,
+      "loss": 0.2596,
+      "step": 1902
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.863743279248659e-05,
+      "loss": 0.2392,
+      "step": 1903
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.86357761388448e-05,
+      "loss": 0.2653,
+      "step": 1904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8634118552429522e-05,
+      "loss": 0.2795,
+      "step": 1905
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8632460033419796e-05,
+      "loss": 0.2636,
+      "step": 1906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8630800581994767e-05,
+      "loss": 0.257,
+      "step": 1907
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.8629140198333666e-05,
+      "loss": 0.2621,
+      "step": 1908
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8627478882615847e-05,
+      "loss": 0.2536,
+      "step": 1909
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8625816635020744e-05,
+      "loss": 0.2686,
+      "step": 1910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.862415345572791e-05,
+      "loss": 0.2258,
+      "step": 1911
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.862248934491698e-05,
+      "loss": 0.26,
+      "step": 1912
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8620824302767707e-05,
+      "loss": 0.2631,
+      "step": 1913
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.861915832945993e-05,
+      "loss": 0.2416,
+      "step": 1914
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.8617491425173603e-05,
+      "loss": 0.2556,
+      "step": 1915
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8615823590088767e-05,
+      "loss": 0.2364,
+      "step": 1916
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.861415482438557e-05,
+      "loss": 0.2526,
+      "step": 1917
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.861248512824426e-05,
+      "loss": 0.2277,
+      "step": 1918
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.8610814501845188e-05,
+      "loss": 0.2562,
+      "step": 1919
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.86091429453688e-05,
+      "loss": 0.2521,
+      "step": 1920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.860747045899565e-05,
+      "loss": 0.246,
+      "step": 1921
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.860579704290638e-05,
+      "loss": 0.2423,
+      "step": 1922
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8604122697281748e-05,
+      "loss": 0.2493,
+      "step": 1923
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.86024474223026e-05,
+      "loss": 0.2485,
+      "step": 1924
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8600771218149884e-05,
+      "loss": 0.2416,
+      "step": 1925
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.859909408500466e-05,
+      "loss": 0.2851,
+      "step": 1926
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.859741602304807e-05,
+      "loss": 0.2676,
+      "step": 1927
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.8595737032461376e-05,
+      "loss": 0.2785,
+      "step": 1928
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8594057113425922e-05,
+      "loss": 0.278,
+      "step": 1929
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8592376266123166e-05,
+      "loss": 0.2578,
+      "step": 1930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8590694490734655e-05,
+      "loss": 0.2565,
+      "step": 1931
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.858901178744205e-05,
+      "loss": 0.2308,
+      "step": 1932
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8587328156427098e-05,
+      "loss": 0.2661,
+      "step": 1933
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8585643597871658e-05,
+      "loss": 0.283,
+      "step": 1934
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8583958111957676e-05,
+      "loss": 0.2657,
+      "step": 1935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.858227169886721e-05,
+      "loss": 0.2748,
+      "step": 1936
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.858058435878242e-05,
+      "loss": 0.2387,
+      "step": 1937
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.857889609188555e-05,
+      "loss": 0.2441,
+      "step": 1938
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8577206898358958e-05,
+      "loss": 0.2452,
+      "step": 1939
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.85755167783851e-05,
+      "loss": 0.2717,
+      "step": 1940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.857382573214653e-05,
+      "loss": 0.2501,
+      "step": 1941
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8572133759825903e-05,
+      "loss": 0.2528,
+      "step": 1942
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8570440861605973e-05,
+      "loss": 0.2664,
+      "step": 1943
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.856874703766959e-05,
+      "loss": 0.2415,
+      "step": 1944
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8567052288199714e-05,
+      "loss": 0.2616,
+      "step": 1945
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.85653566133794e-05,
+      "loss": 0.2599,
+      "step": 1946
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8563660013391798e-05,
+      "loss": 0.2711,
+      "step": 1947
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8561962488420165e-05,
+      "loss": 0.2525,
+      "step": 1948
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8560264038647858e-05,
+      "loss": 0.261,
+      "step": 1949
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.855856466425832e-05,
+      "loss": 0.2679,
+      "step": 1950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.855686436543512e-05,
+      "loss": 0.2474,
+      "step": 1951
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8555163142361905e-05,
+      "loss": 0.2472,
+      "step": 1952
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8553460995222427e-05,
+      "loss": 0.233,
+      "step": 1953
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8551757924200542e-05,
+      "loss": 0.2543,
+      "step": 1954
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8550053929480202e-05,
+      "loss": 0.2252,
+      "step": 1955
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854834901124546e-05,
+      "loss": 0.2277,
+      "step": 1956
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854664316968047e-05,
+      "loss": 0.2487,
+      "step": 1957
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8544936404969482e-05,
+      "loss": 0.2758,
+      "step": 1958
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8543228717296853e-05,
+      "loss": 0.2706,
+      "step": 1959
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854152010684703e-05,
+      "loss": 0.2319,
+      "step": 1960
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8539810573804566e-05,
+      "loss": 0.2446,
+      "step": 1961
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8538100118354116e-05,
+      "loss": 0.2676,
+      "step": 1962
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8536388740680427e-05,
+      "loss": 0.2642,
+      "step": 1963
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.853467644096835e-05,
+      "loss": 0.2365,
+      "step": 1964
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8532963219402834e-05,
+      "loss": 0.273,
+      "step": 1965
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.853124907616893e-05,
+      "loss": 0.3027,
+      "step": 1966
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.852953401145179e-05,
+      "loss": 0.2816,
+      "step": 1967
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8527818025436662e-05,
+      "loss": 0.2477,
+      "step": 1968
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8526101118308885e-05,
+      "loss": 0.2489,
+      "step": 1969
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8524383290253924e-05,
+      "loss": 0.2537,
+      "step": 1970
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.852266454145731e-05,
+      "loss": 0.2708,
+      "step": 1971
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8520944872104697e-05,
+      "loss": 0.1754,
+      "step": 1972
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.851922428238183e-05,
+      "loss": 0.1591,
+      "step": 1973
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.851750277247456e-05,
+      "loss": 0.1635,
+      "step": 1974
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.851578034256882e-05,
+      "loss": 0.1552,
+      "step": 1975
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8514056992850664e-05,
+      "loss": 0.1747,
+      "step": 1976
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8512332723506233e-05,
+      "loss": 0.1737,
+      "step": 1977
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.851060753472177e-05,
+      "loss": 0.1506,
+      "step": 1978
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8508881426683618e-05,
+      "loss": 0.1465,
+      "step": 1979
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8507154399578217e-05,
+      "loss": 0.1678,
+      "step": 1980
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8505426453592105e-05,
+      "loss": 0.1522,
+      "step": 1981
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8503697588911927e-05,
+      "loss": 0.1357,
+      "step": 1982
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8501967805724424e-05,
+      "loss": 0.1654,
+      "step": 1983
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.850023710421643e-05,
+      "loss": 0.152,
+      "step": 1984
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8498505484574884e-05,
+      "loss": 0.1447,
+      "step": 1985
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8496772946986822e-05,
+      "loss": 0.153,
+      "step": 1986
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8495039491639385e-05,
+      "loss": 0.1411,
+      "step": 1987
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.84933051187198e-05,
+      "loss": 0.1588,
+      "step": 1988
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.849156982841541e-05,
+      "loss": 0.1497,
+      "step": 1989
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8489833620913644e-05,
+      "loss": 0.1546,
+      "step": 1990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8488096496402028e-05,
+      "loss": 0.1456,
+      "step": 1991
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.8486358455068205e-05,
+      "loss": 0.149,
+      "step": 1992
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.84846194970999e-05,
+      "loss": 0.137,
+      "step": 1993
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8482879622684943e-05,
+      "loss": 0.1487,
+      "step": 1994
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8481138832011264e-05,
+      "loss": 0.141,
+      "step": 1995
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8479397125266886e-05,
+      "loss": 0.152,
+      "step": 1996
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8477654502639944e-05,
+      "loss": 0.1635,
+      "step": 1997
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8475910964318653e-05,
+      "loss": 0.1656,
+      "step": 1998
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8474166510491345e-05,
+      "loss": 0.1393,
+      "step": 1999
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.8472421141346438e-05,
+      "loss": 0.1497,
+      "step": 2000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.847067485707246e-05,
+      "loss": 0.1576,
+      "step": 2001
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8468927657858024e-05,
+      "loss": 0.1428,
+      "step": 2002
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8467179543891857e-05,
+      "loss": 0.1468,
+      "step": 2003
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8465430515362777e-05,
+      "loss": 0.1621,
+      "step": 2004
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8463680572459695e-05,
+      "loss": 0.1405,
+      "step": 2005
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8461929715371633e-05,
+      "loss": 0.1622,
+      "step": 2006
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8460177944287702e-05,
+      "loss": 0.1787,
+      "step": 2007
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8458425259397116e-05,
+      "loss": 0.1355,
+      "step": 2008
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.845667166088919e-05,
+      "loss": 0.1616,
+      "step": 2009
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8454917148953334e-05,
+      "loss": 0.1649,
+      "step": 2010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8453161723779057e-05,
+      "loss": 0.1549,
+      "step": 2011
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.8451405385555965e-05,
+      "loss": 0.1391,
+      "step": 2012
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.844964813447377e-05,
+      "loss": 0.1502,
+      "step": 2013
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.844788997072227e-05,
+      "loss": 0.1378,
+      "step": 2014
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8446130894491378e-05,
+      "loss": 0.1605,
+      "step": 2015
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8444370905971093e-05,
+      "loss": 0.1691,
+      "step": 2016
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8442610005351513e-05,
+      "loss": 0.1534,
+      "step": 2017
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.844084819282284e-05,
+      "loss": 0.1333,
+      "step": 2018
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.843908546857537e-05,
+      "loss": 0.1567,
+      "step": 2019
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8437321832799502e-05,
+      "loss": 0.1537,
+      "step": 2020
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.843555728568573e-05,
+      "loss": 0.1578,
+      "step": 2021
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.843379182742465e-05,
+      "loss": 0.1316,
+      "step": 2022
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8432025458206953e-05,
+      "loss": 0.1599,
+      "step": 2023
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8430258178223427e-05,
+      "loss": 0.1609,
+      "step": 2024
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8428489987664958e-05,
+      "loss": 0.1447,
+      "step": 2025
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.8426720886722545e-05,
+      "loss": 0.161,
+      "step": 2026
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.842495087558726e-05,
+      "loss": 0.1404,
+      "step": 2027
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8423179954450294e-05,
+      "loss": 0.1534,
+      "step": 2028
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.842140812350292e-05,
+      "loss": 0.1629,
+      "step": 2029
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8419635382936533e-05,
+      "loss": 0.1474,
+      "step": 2030
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8417861732942602e-05,
+      "loss": 0.1701,
+      "step": 2031
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.8416087173712704e-05,
+      "loss": 0.1756,
+      "step": 2032
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.841431170543852e-05,
+      "loss": 0.1442,
+      "step": 2033
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.1574,
+      "step": 2034
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.841075804252446e-05,
+      "loss": 0.1675,
+      "step": 2035
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8408979848268435e-05,
+      "loss": 0.1607,
+      "step": 2036
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8407200745735795e-05,
+      "loss": 0.1776,
+      "step": 2037
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8405420735118715e-05,
+      "loss": 0.1588,
+      "step": 2038
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.8403639816609456e-05,
+      "loss": 0.1532,
+      "step": 2039
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.840185799040038e-05,
+      "loss": 0.1472,
+      "step": 2040
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.840007525668394e-05,
+      "loss": 0.1459,
+      "step": 2041
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8398291615652703e-05,
+      "loss": 0.1458,
+      "step": 2042
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8396507067499322e-05,
+      "loss": 0.1667,
+      "step": 2043
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.839472161241655e-05,
+      "loss": 0.1581,
+      "step": 2044
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.839293525059724e-05,
+      "loss": 0.1529,
+      "step": 2045
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.839114798223434e-05,
+      "loss": 0.1546,
+      "step": 2046
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8389359807520898e-05,
+      "loss": 0.1324,
+      "step": 2047
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8387570726650065e-05,
+      "loss": 0.1726,
+      "step": 2048
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8385780739815074e-05,
+      "loss": 0.1475,
+      "step": 2049
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8383989847209275e-05,
+      "loss": 0.1519,
+      "step": 2050
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8382198049026104e-05,
+      "loss": 0.1458,
+      "step": 2051
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.83804053454591e-05,
+      "loss": 0.1552,
+      "step": 2052
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8378611736701892e-05,
+      "loss": 0.1609,
+      "step": 2053
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.837681722294822e-05,
+      "loss": 0.1594,
+      "step": 2054
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8375021804391906e-05,
+      "loss": 0.1455,
+      "step": 2055
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8373225481226886e-05,
+      "loss": 0.1595,
+      "step": 2056
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8371428253647182e-05,
+      "loss": 0.1618,
+      "step": 2057
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.836963012184692e-05,
+      "loss": 0.1559,
+      "step": 2058
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8367831086020314e-05,
+      "loss": 0.1801,
+      "step": 2059
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8366031146361694e-05,
+      "loss": 0.1677,
+      "step": 2060
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8364230303065466e-05,
+      "loss": 0.177,
+      "step": 2061
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.836242855632615e-05,
+      "loss": 0.1548,
+      "step": 2062
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8360625906338356e-05,
+      "loss": 0.1407,
+      "step": 2063
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8358822353296786e-05,
+      "loss": 0.1708,
+      "step": 2064
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.835701789739626e-05,
+      "loss": 0.1743,
+      "step": 2065
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8355212538831674e-05,
+      "loss": 0.15,
+      "step": 2066
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.835340627779803e-05,
+      "loss": 0.1549,
+      "step": 2067
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8351599114490428e-05,
+      "loss": 0.136,
+      "step": 2068
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8349791049104066e-05,
+      "loss": 0.1556,
+      "step": 2069
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8347982081834233e-05,
+      "loss": 0.1831,
+      "step": 2070
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.834617221287633e-05,
+      "loss": 0.1604,
+      "step": 2071
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8344361442425834e-05,
+      "loss": 0.1549,
+      "step": 2072
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.834254977067834e-05,
+      "loss": 0.1772,
+      "step": 2073
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.834073719782953e-05,
+      "loss": 0.1822,
+      "step": 2074
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8338923724075183e-05,
+      "loss": 0.1651,
+      "step": 2075
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8337109349611178e-05,
+      "loss": 0.166,
+      "step": 2076
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.833529407463349e-05,
+      "loss": 0.1866,
+      "step": 2077
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.833347789933819e-05,
+      "loss": 0.1639,
+      "step": 2078
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.833166082392145e-05,
+      "loss": 0.1509,
+      "step": 2079
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.832984284857954e-05,
+      "loss": 0.1588,
+      "step": 2080
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.832802397350882e-05,
+      "loss": 0.1597,
+      "step": 2081
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8326204198905757e-05,
+      "loss": 0.1448,
+      "step": 2082
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.83243835249669e-05,
+      "loss": 0.158,
+      "step": 2083
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8322561951888914e-05,
+      "loss": 0.1617,
+      "step": 2084
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.832073947986855e-05,
+      "loss": 0.1572,
+      "step": 2085
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8318916109102657e-05,
+      "loss": 0.1678,
+      "step": 2086
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8317091839788183e-05,
+      "loss": 0.1538,
+      "step": 2087
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.831526667212217e-05,
+      "loss": 0.1635,
+      "step": 2088
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.831344060630176e-05,
+      "loss": 0.1671,
+      "step": 2089
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8311613642524193e-05,
+      "loss": 0.1305,
+      "step": 2090
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8309785780986808e-05,
+      "loss": 0.1625,
+      "step": 2091
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8307957021887027e-05,
+      "loss": 0.1729,
+      "step": 2092
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8306127365422385e-05,
+      "loss": 0.1605,
+      "step": 2093
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.830429681179051e-05,
+      "loss": 0.1597,
+      "step": 2094
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.830246536118912e-05,
+      "loss": 0.1609,
+      "step": 2095
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.830063301381604e-05,
+      "loss": 0.1417,
+      "step": 2096
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8298799769869187e-05,
+      "loss": 0.1458,
+      "step": 2097
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.829696562954657e-05,
+      "loss": 0.1823,
+      "step": 2098
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8295130593046305e-05,
+      "loss": 0.1857,
+      "step": 2099
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8293294660566593e-05,
+      "loss": 0.1469,
+      "step": 2100
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.829145783230574e-05,
+      "loss": 0.1578,
+      "step": 2101
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.828962010846215e-05,
+      "loss": 0.1708,
+      "step": 2102
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.828778148923432e-05,
+      "loss": 0.16,
+      "step": 2103
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.828594197482084e-05,
+      "loss": 0.168,
+      "step": 2104
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.82841015654204e-05,
+      "loss": 0.1532,
+      "step": 2105
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8282260261231798e-05,
+      "loss": 0.1563,
+      "step": 2106
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8280418062453912e-05,
+      "loss": 0.1386,
+      "step": 2107
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8278574969285718e-05,
+      "loss": 0.175,
+      "step": 2108
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8276730981926298e-05,
+      "loss": 0.1642,
+      "step": 2109
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8274886100574827e-05,
+      "loss": 0.1914,
+      "step": 2110
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8273040325430575e-05,
+      "loss": 0.1671,
+      "step": 2111
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8271193656692907e-05,
+      "loss": 0.1758,
+      "step": 2112
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.826934609456129e-05,
+      "loss": 0.1607,
+      "step": 2113
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.826749763923528e-05,
+      "loss": 0.1751,
+      "step": 2114
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8265648290914538e-05,
+      "loss": 0.1523,
+      "step": 2115
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8263798049798813e-05,
+      "loss": 0.1582,
+      "step": 2116
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8261946916087957e-05,
+      "loss": 0.1473,
+      "step": 2117
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.8260094889981917e-05,
+      "loss": 0.1603,
+      "step": 2118
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8258241971680733e-05,
+      "loss": 0.1574,
+      "step": 2119
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8256388161384548e-05,
+      "loss": 0.163,
+      "step": 2120
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.825453345929359e-05,
+      "loss": 0.1691,
+      "step": 2121
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8252677865608196e-05,
+      "loss": 0.1534,
+      "step": 2122
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8250821380528788e-05,
+      "loss": 0.1534,
+      "step": 2123
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.82489640042559e-05,
+      "loss": 0.1708,
+      "step": 2124
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.824710573699014e-05,
+      "loss": 0.1618,
+      "step": 2125
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8245246578932236e-05,
+      "loss": 0.1617,
+      "step": 2126
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.824338653028299e-05,
+      "loss": 0.1537,
+      "step": 2127
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8241525591243315e-05,
+      "loss": 0.154,
+      "step": 2128
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8239663762014224e-05,
+      "loss": 0.1654,
+      "step": 2129
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8237801042796807e-05,
+      "loss": 0.1572,
+      "step": 2130
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8235937433792265e-05,
+      "loss": 0.1664,
+      "step": 2131
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8234072935201893e-05,
+      "loss": 0.1709,
+      "step": 2132
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.823220754722708e-05,
+      "loss": 0.1594,
+      "step": 2133
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8230341270069312e-05,
+      "loss": 0.1626,
+      "step": 2134
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8228474103930166e-05,
+      "loss": 0.1762,
+      "step": 2135
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8226606049011328e-05,
+      "loss": 0.1416,
+      "step": 2136
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8224737105514565e-05,
+      "loss": 0.1593,
+      "step": 2137
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8222867273641747e-05,
+      "loss": 0.1618,
+      "step": 2138
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8220996553594842e-05,
+      "loss": 0.1771,
+      "step": 2139
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8219124945575912e-05,
+      "loss": 0.1562,
+      "step": 2140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8217252449787114e-05,
+      "loss": 0.1607,
+      "step": 2141
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8215379066430698e-05,
+      "loss": 0.1692,
+      "step": 2142
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8213504795709017e-05,
+      "loss": 0.1616,
+      "step": 2143
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8211629637824516e-05,
+      "loss": 0.1517,
+      "step": 2144
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.8209753592979734e-05,
+      "loss": 0.1538,
+      "step": 2145
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8207876661377305e-05,
+      "loss": 0.1526,
+      "step": 2146
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8205998843219974e-05,
+      "loss": 0.16,
+      "step": 2147
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.820412013871055e-05,
+      "loss": 0.1473,
+      "step": 2148
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.820224054805197e-05,
+      "loss": 0.1663,
+      "step": 2149
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8200360071447253e-05,
+      "loss": 0.1783,
+      "step": 2150
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.819847870909951e-05,
+      "loss": 0.1723,
+      "step": 2151
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8196596461211954e-05,
+      "loss": 0.1596,
+      "step": 2152
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.819471332798789e-05,
+      "loss": 0.1787,
+      "step": 2153
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8192829309630724e-05,
+      "loss": 0.1669,
+      "step": 2154
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.819094440634395e-05,
+      "loss": 0.1602,
+      "step": 2155
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.8189058618331165e-05,
+      "loss": 0.1777,
+      "step": 2156
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.818717194579606e-05,
+      "loss": 0.1549,
+      "step": 2157
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.818528438894241e-05,
+      "loss": 0.1665,
+      "step": 2158
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8183395947974107e-05,
+      "loss": 0.1765,
+      "step": 2159
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8181506623095114e-05,
+      "loss": 0.1781,
+      "step": 2160
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.817961641450951e-05,
+      "loss": 0.1503,
+      "step": 2161
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8177725322421464e-05,
+      "loss": 0.1804,
+      "step": 2162
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8175833347035233e-05,
+      "loss": 0.1739,
+      "step": 2163
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8173940488555172e-05,
+      "loss": 0.176,
+      "step": 2164
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8172046747185744e-05,
+      "loss": 0.1374,
+      "step": 2165
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8170152123131485e-05,
+      "loss": 0.1432,
+      "step": 2166
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8168256616597048e-05,
+      "loss": 0.1683,
+      "step": 2167
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8166360227787163e-05,
+      "loss": 0.1653,
+      "step": 2168
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.816446295690667e-05,
+      "loss": 0.1698,
+      "step": 2169
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.81625648041605e-05,
+      "loss": 0.1807,
+      "step": 2170
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.8160665769753674e-05,
+      "loss": 0.1734,
+      "step": 2171
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8158765853891313e-05,
+      "loss": 0.1815,
+      "step": 2172
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.815686505677863e-05,
+      "loss": 0.1862,
+      "step": 2173
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.815496337862094e-05,
+      "loss": 0.1843,
+      "step": 2174
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.815306081962364e-05,
+      "loss": 0.1765,
+      "step": 2175
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.8151157379992237e-05,
+      "loss": 0.1695,
+      "step": 2176
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.814925305993233e-05,
+      "loss": 0.1592,
+      "step": 2177
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.81473478596496e-05,
+      "loss": 0.1648,
+      "step": 2178
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.814544177934984e-05,
+      "loss": 0.1746,
+      "step": 2179
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.814353481923893e-05,
+      "loss": 0.1668,
+      "step": 2180
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8141626979522843e-05,
+      "loss": 0.1662,
+      "step": 2181
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8139718260407653e-05,
+      "loss": 0.1551,
+      "step": 2182
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8137808662099522e-05,
+      "loss": 0.1678,
+      "step": 2183
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8135898184804718e-05,
+      "loss": 0.1987,
+      "step": 2184
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.813398682872959e-05,
+      "loss": 0.1618,
+      "step": 2185
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8132074594080592e-05,
+      "loss": 0.1627,
+      "step": 2186
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8130161481064266e-05,
+      "loss": 0.1693,
+      "step": 2187
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8128247489887257e-05,
+      "loss": 0.1574,
+      "step": 2188
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.81263326207563e-05,
+      "loss": 0.1581,
+      "step": 2189
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.8124416873878223e-05,
+      "loss": 0.1771,
+      "step": 2190
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.812250024945995e-05,
+      "loss": 0.1637,
+      "step": 2191
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8120582747708503e-05,
+      "loss": 0.1654,
+      "step": 2192
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8118664368830995e-05,
+      "loss": 0.1756,
+      "step": 2193
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8116745113034638e-05,
+      "loss": 0.1581,
+      "step": 2194
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8114824980526733e-05,
+      "loss": 0.1637,
+      "step": 2195
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.811290397151468e-05,
+      "loss": 0.1493,
+      "step": 2196
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8110982086205973e-05,
+      "loss": 0.2072,
+      "step": 2197
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8109059324808198e-05,
+      "loss": 0.1585,
+      "step": 2198
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8107135687529044e-05,
+      "loss": 0.1724,
+      "step": 2199
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.810521117457628e-05,
+      "loss": 0.1686,
+      "step": 2200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.810328578615778e-05,
+      "loss": 0.176,
+      "step": 2201
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.810135952248151e-05,
+      "loss": 0.1442,
+      "step": 2202
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8099432383755533e-05,
+      "loss": 0.1553,
+      "step": 2203
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.8097504370188006e-05,
+      "loss": 0.1875,
+      "step": 2204
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8095575481987177e-05,
+      "loss": 0.1611,
+      "step": 2205
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.809364571936139e-05,
+      "loss": 0.1714,
+      "step": 2206
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8091715082519084e-05,
+      "loss": 0.1798,
+      "step": 2207
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.808978357166879e-05,
+      "loss": 0.152,
+      "step": 2208
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.808785118701914e-05,
+      "loss": 0.1849,
+      "step": 2209
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8085917928778856e-05,
+      "loss": 0.1632,
+      "step": 2210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.808398379715675e-05,
+      "loss": 0.1779,
+      "step": 2211
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8082048792361736e-05,
+      "loss": 0.1634,
+      "step": 2212
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8080112914602818e-05,
+      "loss": 0.1386,
+      "step": 2213
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.80781761640891e-05,
+      "loss": 0.1792,
+      "step": 2214
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.8076238541029768e-05,
+      "loss": 0.1702,
+      "step": 2215
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.807430004563411e-05,
+      "loss": 0.1885,
+      "step": 2216
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.807236067811152e-05,
+      "loss": 0.1456,
+      "step": 2217
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8070420438671462e-05,
+      "loss": 0.1706,
+      "step": 2218
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8068479327523508e-05,
+      "loss": 0.1689,
+      "step": 2219
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8066537344877327e-05,
+      "loss": 0.1821,
+      "step": 2220
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8064594490942675e-05,
+      "loss": 0.1844,
+      "step": 2221
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.806265076592941e-05,
+      "loss": 0.1584,
+      "step": 2222
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.806070617004747e-05,
+      "loss": 0.178,
+      "step": 2223
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.80587607035069e-05,
+      "loss": 0.1694,
+      "step": 2224
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8056814366517836e-05,
+      "loss": 0.1701,
+      "step": 2225
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.805486715929051e-05,
+      "loss": 0.1798,
+      "step": 2226
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.805291908203524e-05,
+      "loss": 0.1643,
+      "step": 2227
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.8050970134962447e-05,
+      "loss": 0.157,
+      "step": 2228
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.804902031828264e-05,
+      "loss": 0.152,
+      "step": 2229
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.804706963220642e-05,
+      "loss": 0.1645,
+      "step": 2230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8045118076944494e-05,
+      "loss": 0.1815,
+      "step": 2231
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.804316565270765e-05,
+      "loss": 0.1554,
+      "step": 2232
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8041212359706773e-05,
+      "loss": 0.169,
+      "step": 2233
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8039258198152852e-05,
+      "loss": 0.1749,
+      "step": 2234
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.803730316825695e-05,
+      "loss": 0.1698,
+      "step": 2235
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.803534727023024e-05,
+      "loss": 0.1791,
+      "step": 2236
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.8033390504283984e-05,
+      "loss": 0.178,
+      "step": 2237
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.803143287062954e-05,
+      "loss": 0.1731,
+      "step": 2238
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8029474369478356e-05,
+      "loss": 0.1689,
+      "step": 2239
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.802751500104197e-05,
+      "loss": 0.1755,
+      "step": 2240
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8025554765532023e-05,
+      "loss": 0.1733,
+      "step": 2241
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8023593663160247e-05,
+      "loss": 0.1533,
+      "step": 2242
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.802163169413846e-05,
+      "loss": 0.1613,
+      "step": 2243
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.801966885867859e-05,
+      "loss": 0.1694,
+      "step": 2244
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8017705156992633e-05,
+      "loss": 0.177,
+      "step": 2245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8015740589292707e-05,
+      "loss": 0.1688,
+      "step": 2246
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8013775155791004e-05,
+      "loss": 0.156,
+      "step": 2247
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.801180885669982e-05,
+      "loss": 0.1661,
+      "step": 2248
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.8009841692231534e-05,
+      "loss": 0.1634,
+      "step": 2249
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.800787366259863e-05,
+      "loss": 0.1619,
+      "step": 2250
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.800590476801368e-05,
+      "loss": 0.1687,
+      "step": 2251
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.800393500868935e-05,
+      "loss": 0.1776,
+      "step": 2252
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8001964384838394e-05,
+      "loss": 0.1654,
+      "step": 2253
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7999992896673667e-05,
+      "loss": 0.1863,
+      "step": 2254
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7998020544408113e-05,
+      "loss": 0.1963,
+      "step": 2255
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.7996047328254778e-05,
+      "loss": 0.1672,
+      "step": 2256
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7994073248426785e-05,
+      "loss": 0.1689,
+      "step": 2257
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.799209830513737e-05,
+      "loss": 0.1839,
+      "step": 2258
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.799012249859984e-05,
+      "loss": 0.1623,
+      "step": 2259
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.798814582902762e-05,
+      "loss": 0.1695,
+      "step": 2260
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7986168296634207e-05,
+      "loss": 0.1851,
+      "step": 2261
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.79841899016332e-05,
+      "loss": 0.1624,
+      "step": 2262
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.7982210644238294e-05,
+      "loss": 0.1837,
+      "step": 2263
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7980230524663273e-05,
+      "loss": 0.1867,
+      "step": 2264
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7978249543122012e-05,
+      "loss": 0.1768,
+      "step": 2265
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7976267699828486e-05,
+      "loss": 0.1816,
+      "step": 2266
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7974284994996757e-05,
+      "loss": 0.208,
+      "step": 2267
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7972301428840984e-05,
+      "loss": 0.1547,
+      "step": 2268
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.7970317001575413e-05,
+      "loss": 0.1709,
+      "step": 2269
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7968331713414398e-05,
+      "loss": 0.1641,
+      "step": 2270
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.796634556457236e-05,
+      "loss": 0.1736,
+      "step": 2271
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7964358555263845e-05,
+      "loss": 0.1728,
+      "step": 2272
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.796237068570346e-05,
+      "loss": 0.172,
+      "step": 2273
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.796038195610593e-05,
+      "loss": 0.1609,
+      "step": 2274
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7958392366686057e-05,
+      "loss": 0.1585,
+      "step": 2275
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.7956401917658747e-05,
+      "loss": 0.1745,
+      "step": 2276
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7954410609238994e-05,
+      "loss": 0.1842,
+      "step": 2277
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7952418441641878e-05,
+      "loss": 0.1886,
+      "step": 2278
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7950425415082585e-05,
+      "loss": 0.1851,
+      "step": 2279
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.794843152977638e-05,
+      "loss": 0.151,
+      "step": 2280
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7946436785938636e-05,
+      "loss": 0.1776,
+      "step": 2281
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7944441183784807e-05,
+      "loss": 0.18,
+      "step": 2282
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.7942444723530448e-05,
+      "loss": 0.1682,
+      "step": 2283
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7940447405391196e-05,
+      "loss": 0.1849,
+      "step": 2284
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7938449229582783e-05,
+      "loss": 0.1907,
+      "step": 2285
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7936450196321048e-05,
+      "loss": 0.1801,
+      "step": 2286
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7934450305821905e-05,
+      "loss": 0.1724,
+      "step": 2287
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.793244955830137e-05,
+      "loss": 0.1738,
+      "step": 2288
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.7930447953975548e-05,
+      "loss": 0.1692,
+      "step": 2289
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.792844549306064e-05,
+      "loss": 0.1826,
+      "step": 2290
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7926442175772933e-05,
+      "loss": 0.1839,
+      "step": 2291
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7924438002328813e-05,
+      "loss": 0.1647,
+      "step": 2292
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7922432972944758e-05,
+      "loss": 0.1754,
+      "step": 2293
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7920427087837336e-05,
+      "loss": 0.1664,
+      "step": 2294
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7918420347223206e-05,
+      "loss": 0.1848,
+      "step": 2295
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.7916412751319118e-05,
+      "loss": 0.1684,
+      "step": 2296
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.791440430034193e-05,
+      "loss": 0.1785,
+      "step": 2297
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7912394994508568e-05,
+      "loss": 0.1621,
+      "step": 2298
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7910384834036067e-05,
+      "loss": 0.1729,
+      "step": 2299
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7908373819141554e-05,
+      "loss": 0.1773,
+      "step": 2300
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7906361950042235e-05,
+      "loss": 0.1769,
+      "step": 2301
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.7904349226955426e-05,
+      "loss": 0.1804,
+      "step": 2302
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.790233565009852e-05,
+      "loss": 0.1671,
+      "step": 2303
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7900321219689013e-05,
+      "loss": 0.1935,
+      "step": 2304
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.789830593594449e-05,
+      "loss": 0.1722,
+      "step": 2305
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7896289799082626e-05,
+      "loss": 0.1854,
+      "step": 2306
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7894272809321186e-05,
+      "loss": 0.1706,
+      "step": 2307
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.7892254966878035e-05,
+      "loss": 0.1562,
+      "step": 2308
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.789023627197112e-05,
+      "loss": 0.1833,
+      "step": 2309
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7888216724818493e-05,
+      "loss": 0.1763,
+      "step": 2310
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.788619632563829e-05,
+      "loss": 0.1889,
+      "step": 2311
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7884175074648735e-05,
+      "loss": 0.1872,
+      "step": 2312
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.788215297206815e-05,
+      "loss": 0.1778,
+      "step": 2313
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.788013001811495e-05,
+      "loss": 0.1595,
+      "step": 2314
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.7878106213007635e-05,
+      "loss": 0.1679,
+      "step": 2315
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7876081556964805e-05,
+      "loss": 0.1686,
+      "step": 2316
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7874056050205152e-05,
+      "loss": 0.1461,
+      "step": 2317
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.787202969294745e-05,
+      "loss": 0.1878,
+      "step": 2318
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7870002485410576e-05,
+      "loss": 0.1786,
+      "step": 2319
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.786797442781349e-05,
+      "loss": 0.177,
+      "step": 2320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.7865945520375255e-05,
+      "loss": 0.1622,
+      "step": 2321
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.786391576331501e-05,
+      "loss": 0.1823,
+      "step": 2322
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7861885156852e-05,
+      "loss": 0.1749,
+      "step": 2323
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7859853701205556e-05,
+      "loss": 0.1924,
+      "step": 2324
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.78578213965951e-05,
+      "loss": 0.1722,
+      "step": 2325
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7855788243240147e-05,
+      "loss": 0.1747,
+      "step": 2326
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7853754241360303e-05,
+      "loss": 0.1794,
+      "step": 2327
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.785171939117526e-05,
+      "loss": 0.1687,
+      "step": 2328
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.7849683692904823e-05,
+      "loss": 0.1711,
+      "step": 2329
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.784764714676886e-05,
+      "loss": 0.1776,
+      "step": 2330
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.784560975298735e-05,
+      "loss": 0.1897,
+      "step": 2331
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.784357151178035e-05,
+      "loss": 0.1735,
+      "step": 2332
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7841532423368025e-05,
+      "loss": 0.1912,
+      "step": 2333
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7839492487970615e-05,
+      "loss": 0.1786,
+      "step": 2334
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.7837451705808468e-05,
+      "loss": 0.1798,
+      "step": 2335
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7835410077102007e-05,
+      "loss": 0.1761,
+      "step": 2336
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7833367602071755e-05,
+      "loss": 0.1987,
+      "step": 2337
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7831324280938327e-05,
+      "loss": 0.1732,
+      "step": 2338
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7829280113922426e-05,
+      "loss": 0.1759,
+      "step": 2339
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7827235101244852e-05,
+      "loss": 0.1721,
+      "step": 2340
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7825189243126482e-05,
+      "loss": 0.184,
+      "step": 2341
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.7823142539788307e-05,
+      "loss": 0.179,
+      "step": 2342
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7821094991451392e-05,
+      "loss": 0.1642,
+      "step": 2343
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7819046598336897e-05,
+      "loss": 0.1885,
+      "step": 2344
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7816997360666074e-05,
+      "loss": 0.1713,
+      "step": 2345
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7814947278660266e-05,
+      "loss": 0.1551,
+      "step": 2346
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7812896352540912e-05,
+      "loss": 0.1915,
+      "step": 2347
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.7810844582529538e-05,
+      "loss": 0.195,
+      "step": 2348
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7808791968847757e-05,
+      "loss": 0.1639,
+      "step": 2349
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.780673851171728e-05,
+      "loss": 0.1661,
+      "step": 2350
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.780468421135991e-05,
+      "loss": 0.1711,
+      "step": 2351
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.780262906799753e-05,
+      "loss": 0.1679,
+      "step": 2352
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7800573081852124e-05,
+      "loss": 0.1754,
+      "step": 2353
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.7798516253145767e-05,
+      "loss": 0.1872,
+      "step": 2354
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.779645858210062e-05,
+      "loss": 0.1917,
+      "step": 2355
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7794400068938938e-05,
+      "loss": 0.1688,
+      "step": 2356
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7792340713883073e-05,
+      "loss": 0.1699,
+      "step": 2357
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.779028051715545e-05,
+      "loss": 0.1661,
+      "step": 2358
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7788219478978604e-05,
+      "loss": 0.1985,
+      "step": 2359
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7786157599575153e-05,
+      "loss": 0.1752,
+      "step": 2360
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.7784094879167803e-05,
+      "loss": 0.1725,
+      "step": 2361
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7782031317979358e-05,
+      "loss": 0.1611,
+      "step": 2362
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7779966916232706e-05,
+      "loss": 0.1705,
+      "step": 2363
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7777901674150827e-05,
+      "loss": 0.1842,
+      "step": 2364
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.77758355919568e-05,
+      "loss": 0.1888,
+      "step": 2365
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7773768669873782e-05,
+      "loss": 0.1873,
+      "step": 2366
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7771700908125027e-05,
+      "loss": 0.1816,
+      "step": 2367
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.7769632306933885e-05,
+      "loss": 0.1858,
+      "step": 2368
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7767562866523784e-05,
+      "loss": 0.1707,
+      "step": 2369
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.776549258711825e-05,
+      "loss": 0.1676,
+      "step": 2370
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.776342146894091e-05,
+      "loss": 0.1743,
+      "step": 2371
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7761349512215456e-05,
+      "loss": 0.175,
+      "step": 2372
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.77592767171657e-05,
+      "loss": 0.17,
+      "step": 2373
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.7757203084015522e-05,
+      "loss": 0.1726,
+      "step": 2374
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.77551286129889e-05,
+      "loss": 0.167,
+      "step": 2375
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7753053304309908e-05,
+      "loss": 0.1753,
+      "step": 2376
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.77509771582027e-05,
+      "loss": 0.1767,
+      "step": 2377
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.774890017489153e-05,
+      "loss": 0.1749,
+      "step": 2378
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.774682235460074e-05,
+      "loss": 0.1903,
+      "step": 2379
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7744743697554757e-05,
+      "loss": 0.1683,
+      "step": 2380
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.7742664203978104e-05,
+      "loss": 0.1953,
+      "step": 2381
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7740583874095393e-05,
+      "loss": 0.1736,
+      "step": 2382
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.773850270813133e-05,
+      "loss": 0.1776,
+      "step": 2383
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7736420706310698e-05,
+      "loss": 0.1641,
+      "step": 2384
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7734337868858386e-05,
+      "loss": 0.1673,
+      "step": 2385
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.773225419599937e-05,
+      "loss": 0.2019,
+      "step": 2386
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7730169687958703e-05,
+      "loss": 0.1917,
+      "step": 2387
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.7728084344961547e-05,
+      "loss": 0.1697,
+      "step": 2388
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7725998167233144e-05,
+      "loss": 0.1769,
+      "step": 2389
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.772391115499882e-05,
+      "loss": 0.1786,
+      "step": 2390
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7721823308484016e-05,
+      "loss": 0.1742,
+      "step": 2391
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7719734627914227e-05,
+      "loss": 0.1898,
+      "step": 2392
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7717645113515067e-05,
+      "loss": 0.1711,
+      "step": 2393
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.7715554765512234e-05,
+      "loss": 0.1675,
+      "step": 2394
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.77134635841315e-05,
+      "loss": 0.177,
+      "step": 2395
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.771137156959875e-05,
+      "loss": 0.1809,
+      "step": 2396
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.770927872213994e-05,
+      "loss": 0.1621,
+      "step": 2397
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.770718504198113e-05,
+      "loss": 0.1976,
+      "step": 2398
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.770509052934846e-05,
+      "loss": 0.1638,
+      "step": 2399
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.770299518446817e-05,
+      "loss": 0.1618,
+      "step": 2400
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7700899007566575e-05,
+      "loss": 0.1632,
+      "step": 2401
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7698801998870095e-05,
+      "loss": 0.1874,
+      "step": 2402
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7696704158605233e-05,
+      "loss": 0.161,
+      "step": 2403
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7694605486998583e-05,
+      "loss": 0.188,
+      "step": 2404
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7692505984276822e-05,
+      "loss": 0.1857,
+      "step": 2405
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7690405650666732e-05,
+      "loss": 0.174,
+      "step": 2406
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7688304486395167e-05,
+      "loss": 0.183,
+      "step": 2407
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7686202491689086e-05,
+      "loss": 0.1809,
+      "step": 2408
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.768409966677553e-05,
+      "loss": 0.1743,
+      "step": 2409
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.768199601188163e-05,
+      "loss": 0.1702,
+      "step": 2410
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7679891527234606e-05,
+      "loss": 0.1754,
+      "step": 2411
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7677786213061767e-05,
+      "loss": 0.1854,
+      "step": 2412
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7675680069590522e-05,
+      "loss": 0.1859,
+      "step": 2413
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7673573097048352e-05,
+      "loss": 0.1586,
+      "step": 2414
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7671465295662843e-05,
+      "loss": 0.18,
+      "step": 2415
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7669356665661662e-05,
+      "loss": 0.1863,
+      "step": 2416
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7667247207272568e-05,
+      "loss": 0.1682,
+      "step": 2417
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.766513692072341e-05,
+      "loss": 0.1636,
+      "step": 2418
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7663025806242126e-05,
+      "loss": 0.1788,
+      "step": 2419
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7660913864056735e-05,
+      "loss": 0.1938,
+      "step": 2420
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7658801094395368e-05,
+      "loss": 0.1839,
+      "step": 2421
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7656687497486222e-05,
+      "loss": 0.1837,
+      "step": 2422
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7654573073557596e-05,
+      "loss": 0.1868,
+      "step": 2423
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.765245782283787e-05,
+      "loss": 0.1803,
+      "step": 2424
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.765034174555552e-05,
+      "loss": 0.1717,
+      "step": 2425
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.764822484193911e-05,
+      "loss": 0.1834,
+      "step": 2426
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7646107112217296e-05,
+      "loss": 0.1699,
+      "step": 2427
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.764398855661881e-05,
+      "loss": 0.1607,
+      "step": 2428
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7641869175372493e-05,
+      "loss": 0.1651,
+      "step": 2429
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.763974896870726e-05,
+      "loss": 0.1739,
+      "step": 2430
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.763762793685212e-05,
+      "loss": 0.1766,
+      "step": 2431
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7635506080036174e-05,
+      "loss": 0.1859,
+      "step": 2432
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.7633383398488608e-05,
+      "loss": 0.1656,
+      "step": 2433
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.76312598924387e-05,
+      "loss": 0.1695,
+      "step": 2434
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.762913556211581e-05,
+      "loss": 0.1792,
+      "step": 2435
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7627010407749402e-05,
+      "loss": 0.1768,
+      "step": 2436
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7624884429569013e-05,
+      "loss": 0.1604,
+      "step": 2437
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7622757627804275e-05,
+      "loss": 0.1955,
+      "step": 2438
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7620630002684916e-05,
+      "loss": 0.1954,
+      "step": 2439
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.7618501554440737e-05,
+      "loss": 0.1732,
+      "step": 2440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7616372283301644e-05,
+      "loss": 0.1827,
+      "step": 2441
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7614242189497628e-05,
+      "loss": 0.1824,
+      "step": 2442
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7612111273258762e-05,
+      "loss": 0.1713,
+      "step": 2443
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7609979534815205e-05,
+      "loss": 0.1842,
+      "step": 2444
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7607846974397226e-05,
+      "loss": 0.1823,
+      "step": 2445
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.760571359223516e-05,
+      "loss": 0.1873,
+      "step": 2446
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.7603579388559442e-05,
+      "loss": 0.1598,
+      "step": 2447
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7601444363600593e-05,
+      "loss": 0.1898,
+      "step": 2448
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.759930851758922e-05,
+      "loss": 0.1701,
+      "step": 2449
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7597171850756022e-05,
+      "loss": 0.1627,
+      "step": 2450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.759503436333179e-05,
+      "loss": 0.1844,
+      "step": 2451
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.75928960555474e-05,
+      "loss": 0.1866,
+      "step": 2452
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.7590756927633808e-05,
+      "loss": 0.1822,
+      "step": 2453
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.758861697982208e-05,
+      "loss": 0.1927,
+      "step": 2454
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7586476212343343e-05,
+      "loss": 0.17,
+      "step": 2455
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7584334625428837e-05,
+      "loss": 0.1641,
+      "step": 2456
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.758219221930988e-05,
+      "loss": 0.1778,
+      "step": 2457
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7580048994217874e-05,
+      "loss": 0.1933,
+      "step": 2458
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.757790495038432e-05,
+      "loss": 0.1767,
+      "step": 2459
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.7575760088040797e-05,
+      "loss": 0.1903,
+      "step": 2460
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.757361440741898e-05,
+      "loss": 0.1581,
+      "step": 2461
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7571467908750633e-05,
+      "loss": 0.1866,
+      "step": 2462
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.75693205922676e-05,
+      "loss": 0.1742,
+      "step": 2463
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7567172458201822e-05,
+      "loss": 0.1674,
+      "step": 2464
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7565023506785323e-05,
+      "loss": 0.1851,
+      "step": 2465
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7562873738250214e-05,
+      "loss": 0.1716,
+      "step": 2466
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.7560723152828704e-05,
+      "loss": 0.1582,
+      "step": 2467
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7558571750753076e-05,
+      "loss": 0.1802,
+      "step": 2468
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7556419532255713e-05,
+      "loss": 0.165,
+      "step": 2469
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7554266497569084e-05,
+      "loss": 0.1672,
+      "step": 2470
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.755211264692574e-05,
+      "loss": 0.2006,
+      "step": 2471
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.754995798055833e-05,
+      "loss": 0.1599,
+      "step": 2472
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.7547802498699576e-05,
+      "loss": 0.1775,
+      "step": 2473
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7545646201582304e-05,
+      "loss": 0.1865,
+      "step": 2474
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7543489089439417e-05,
+      "loss": 0.171,
+      "step": 2475
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.754133116250392e-05,
+      "loss": 0.1749,
+      "step": 2476
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7539172421008887e-05,
+      "loss": 0.1917,
+      "step": 2477
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7537012865187494e-05,
+      "loss": 0.1787,
+      "step": 2478
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7534852495272994e-05,
+      "loss": 0.1613,
+      "step": 2479
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7532691311498746e-05,
+      "loss": 0.1781,
+      "step": 2480
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7530529314098174e-05,
+      "loss": 0.1918,
+      "step": 2481
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.752836650330481e-05,
+      "loss": 0.1727,
+      "step": 2482
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.752620287935226e-05,
+      "loss": 0.1731,
+      "step": 2483
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.752403844247422e-05,
+      "loss": 0.1941,
+      "step": 2484
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.7521873192904485e-05,
+      "loss": 0.2028,
+      "step": 2485
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.751970713087692e-05,
+      "loss": 0.1643,
+      "step": 2486
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7517540256625496e-05,
+      "loss": 0.1854,
+      "step": 2487
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.751537257038426e-05,
+      "loss": 0.1761,
+      "step": 2488
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7513204072387345e-05,
+      "loss": 0.1774,
+      "step": 2489
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.751103476286898e-05,
+      "loss": 0.1705,
+      "step": 2490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7508864642063482e-05,
+      "loss": 0.1686,
+      "step": 2491
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.7506693710205244e-05,
+      "loss": 0.1899,
+      "step": 2492
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.750452196752876e-05,
+      "loss": 0.1824,
+      "step": 2493
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7502349414268606e-05,
+      "loss": 0.1798,
+      "step": 2494
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7500176050659438e-05,
+      "loss": 0.1895,
+      "step": 2495
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7498001876936013e-05,
+      "loss": 0.2043,
+      "step": 2496
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.749582689333317e-05,
+      "loss": 0.1904,
+      "step": 2497
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.7493651100085834e-05,
+      "loss": 0.1803,
+      "step": 2498
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.749147449742901e-05,
+      "loss": 0.1575,
+      "step": 2499
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7489297085597813e-05,
+      "loss": 0.1681,
+      "step": 2500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7487118864827422e-05,
+      "loss": 0.187,
+      "step": 2501
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7484939835353118e-05,
+      "loss": 0.1668,
+      "step": 2502
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7482759997410256e-05,
+      "loss": 0.1867,
+      "step": 2503
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7480579351234293e-05,
+      "loss": 0.1651,
+      "step": 2504
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7478397897060767e-05,
+      "loss": 0.1846,
+      "step": 2505
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.7476215635125292e-05,
+      "loss": 0.1721,
+      "step": 2506
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7474032565663595e-05,
+      "loss": 0.1785,
+      "step": 2507
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7471848688911465e-05,
+      "loss": 0.1555,
+      "step": 2508
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7469664005104794e-05,
+      "loss": 0.1675,
+      "step": 2509
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.746747851447955e-05,
+      "loss": 0.178,
+      "step": 2510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.74652922172718e-05,
+      "loss": 0.1975,
+      "step": 2511
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7463105113717686e-05,
+      "loss": 0.1907,
+      "step": 2512
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.7460917204053446e-05,
+      "loss": 0.1833,
+      "step": 2513
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7458728488515406e-05,
+      "loss": 0.1821,
+      "step": 2514
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7456538967339966e-05,
+      "loss": 0.1679,
+      "step": 2515
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.745434864076363e-05,
+      "loss": 0.1736,
+      "step": 2516
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7452157509022977e-05,
+      "loss": 0.1689,
+      "step": 2517
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.7449965572354675e-05,
+      "loss": 0.1821,
+      "step": 2518
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.744777283099549e-05,
+      "loss": 0.1895,
+      "step": 2519
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7445579285182258e-05,
+      "loss": 0.1719,
+      "step": 2520
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7443384935151912e-05,
+      "loss": 0.1459,
+      "step": 2521
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7441189781141466e-05,
+      "loss": 0.2009,
+      "step": 2522
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.743899382338803e-05,
+      "loss": 0.1838,
+      "step": 2523
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7436797062128793e-05,
+      "loss": 0.1776,
+      "step": 2524
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7434599497601036e-05,
+      "loss": 0.2011,
+      "step": 2525
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.7432401130042118e-05,
+      "loss": 0.1691,
+      "step": 2526
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7430201959689495e-05,
+      "loss": 0.1931,
+      "step": 2527
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7428001986780704e-05,
+      "loss": 0.1687,
+      "step": 2528
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.742580121155337e-05,
+      "loss": 0.1832,
+      "step": 2529
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7423599634245207e-05,
+      "loss": 0.1746,
+      "step": 2530
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7421397255094012e-05,
+      "loss": 0.1696,
+      "step": 2531
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.7419194074337666e-05,
+      "loss": 0.172,
+      "step": 2532
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7416990092214145e-05,
+      "loss": 0.1948,
+      "step": 2533
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7414785308961505e-05,
+      "loss": 0.1752,
+      "step": 2534
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7412579724817893e-05,
+      "loss": 0.1891,
+      "step": 2535
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7410373340021537e-05,
+      "loss": 0.1894,
+      "step": 2536
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7408166154810756e-05,
+      "loss": 0.1749,
+      "step": 2537
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.7405958169423954e-05,
+      "loss": 0.1991,
+      "step": 2538
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.740374938409962e-05,
+      "loss": 0.1784,
+      "step": 2539
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7401539799076337e-05,
+      "loss": 0.1681,
+      "step": 2540
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.739932941459276e-05,
+      "loss": 0.178,
+      "step": 2541
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.739711823088764e-05,
+      "loss": 0.1853,
+      "step": 2542
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7394906248199822e-05,
+      "loss": 0.1784,
+      "step": 2543
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7392693466768213e-05,
+      "loss": 0.1717,
+      "step": 2544
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.7390479886831835e-05,
+      "loss": 0.1831,
+      "step": 2545
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7388265508629775e-05,
+      "loss": 0.1796,
+      "step": 2546
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.738605033240122e-05,
+      "loss": 0.1794,
+      "step": 2547
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.738383435838543e-05,
+      "loss": 0.1938,
+      "step": 2548
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7381617586821766e-05,
+      "loss": 0.1735,
+      "step": 2549
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.737940001794966e-05,
+      "loss": 0.1728,
+      "step": 2550
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7377181652008644e-05,
+      "loss": 0.1988,
+      "step": 2551
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.7374962489238324e-05,
+      "loss": 0.1804,
+      "step": 2552
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.73727425298784e-05,
+      "loss": 0.1777,
+      "step": 2553
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.737052177416866e-05,
+      "loss": 0.1893,
+      "step": 2554
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7368300222348965e-05,
+      "loss": 0.191,
+      "step": 2555
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.736607787465928e-05,
+      "loss": 0.1776,
+      "step": 2556
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7363854731339643e-05,
+      "loss": 0.1849,
+      "step": 2557
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.736163079263018e-05,
+      "loss": 0.212,
+      "step": 2558
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.7359406058771108e-05,
+      "loss": 0.1806,
+      "step": 2559
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.735718053000272e-05,
+      "loss": 0.1781,
+      "step": 2560
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.735495420656541e-05,
+      "loss": 0.1747,
+      "step": 2561
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7352727088699645e-05,
+      "loss": 0.1802,
+      "step": 2562
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.735049917664598e-05,
+      "loss": 0.1756,
+      "step": 2563
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7348270470645057e-05,
+      "loss": 0.1773,
+      "step": 2564
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.7346040970937614e-05,
+      "loss": 0.1892,
+      "step": 2565
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7343810677764457e-05,
+      "loss": 0.181,
+      "step": 2566
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7341579591366487e-05,
+      "loss": 0.1863,
+      "step": 2567
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7339347711984688e-05,
+      "loss": 0.1861,
+      "step": 2568
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.733711503986014e-05,
+      "loss": 0.1721,
+      "step": 2569
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.733488157523399e-05,
+      "loss": 0.1879,
+      "step": 2570
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.7332647318347483e-05,
+      "loss": 0.1849,
+      "step": 2571
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.733041226944195e-05,
+      "loss": 0.1827,
+      "step": 2572
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.73281764287588e-05,
+      "loss": 0.1824,
+      "step": 2573
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7325939796539537e-05,
+      "loss": 0.2051,
+      "step": 2574
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7323702373025745e-05,
+      "loss": 0.1709,
+      "step": 2575
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7321464158459093e-05,
+      "loss": 0.1981,
+      "step": 2576
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7319225153081337e-05,
+      "loss": 0.1718,
+      "step": 2577
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.7316985357134318e-05,
+      "loss": 0.179,
+      "step": 2578
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.731474477085996e-05,
+      "loss": 0.177,
+      "step": 2579
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7312503394500278e-05,
+      "loss": 0.1948,
+      "step": 2580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7310261228297367e-05,
+      "loss": 0.1742,
+      "step": 2581
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7308018272493413e-05,
+      "loss": 0.1867,
+      "step": 2582
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.730577452733068e-05,
+      "loss": 0.1921,
+      "step": 2583
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.730352999305152e-05,
+      "loss": 0.1818,
+      "step": 2584
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7301284669898374e-05,
+      "loss": 0.1654,
+      "step": 2585
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7299038558113762e-05,
+      "loss": 0.1753,
+      "step": 2586
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.72967916579403e-05,
+      "loss": 0.1904,
+      "step": 2587
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7294543969620678e-05,
+      "loss": 0.1742,
+      "step": 2588
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7292295493397668e-05,
+      "loss": 0.1831,
+      "step": 2589
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7290046229514146e-05,
+      "loss": 0.1694,
+      "step": 2590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.7287796178213052e-05,
+      "loss": 0.1979,
+      "step": 2591
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7285545339737423e-05,
+      "loss": 0.1979,
+      "step": 2592
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7283293714330378e-05,
+      "loss": 0.1917,
+      "step": 2593
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7281041302235123e-05,
+      "loss": 0.1823,
+      "step": 2594
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7278788103694944e-05,
+      "loss": 0.1744,
+      "step": 2595
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7276534118953217e-05,
+      "loss": 0.1766,
+      "step": 2596
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.72742793482534e-05,
+      "loss": 0.1947,
+      "step": 2597
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.7272023791839037e-05,
+      "loss": 0.1961,
+      "step": 2598
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7269767449953756e-05,
+      "loss": 0.1672,
+      "step": 2599
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7267510322841273e-05,
+      "loss": 0.1931,
+      "step": 2600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7265252410745386e-05,
+      "loss": 0.168,
+      "step": 2601
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7262993713909978e-05,
+      "loss": 0.1701,
+      "step": 2602
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7260734232579013e-05,
+      "loss": 0.1775,
+      "step": 2603
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.7258473966996547e-05,
+      "loss": 0.1755,
+      "step": 2604
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.725621291740672e-05,
+      "loss": 0.1631,
+      "step": 2605
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.725395108405375e-05,
+      "loss": 0.1907,
+      "step": 2606
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7251688467181944e-05,
+      "loss": 0.1695,
+      "step": 2607
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7249425067035695e-05,
+      "loss": 0.169,
+      "step": 2608
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.724716088385948e-05,
+      "loss": 0.1724,
+      "step": 2609
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7244895917897856e-05,
+      "loss": 0.1854,
+      "step": 2610
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7242630169395468e-05,
+      "loss": 0.1863,
+      "step": 2611
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7240363638597052e-05,
+      "loss": 0.1814,
+      "step": 2612
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7238096325747417e-05,
+      "loss": 0.1735,
+      "step": 2613
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7235828231091457e-05,
+      "loss": 0.1872,
+      "step": 2614
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.723355935487417e-05,
+      "loss": 0.1886,
+      "step": 2615
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7231289697340605e-05,
+      "loss": 0.2019,
+      "step": 2616
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7229019258735923e-05,
+      "loss": 0.1691,
+      "step": 2617
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7226748039305365e-05,
+      "loss": 0.1821,
+      "step": 2618
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7224476039294243e-05,
+      "loss": 0.1733,
+      "step": 2619
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7222203258947968e-05,
+      "loss": 0.1841,
+      "step": 2620
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7219929698512022e-05,
+      "loss": 0.183,
+      "step": 2621
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7217655358231985e-05,
+      "loss": 0.1647,
+      "step": 2622
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7215380238353513e-05,
+      "loss": 0.1671,
+      "step": 2623
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7213104339122346e-05,
+      "loss": 0.1787,
+      "step": 2624
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.721082766078431e-05,
+      "loss": 0.1694,
+      "step": 2625
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.720855020358532e-05,
+      "loss": 0.1767,
+      "step": 2626
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7206271967771364e-05,
+      "loss": 0.1833,
+      "step": 2627
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.720399295358852e-05,
+      "loss": 0.1693,
+      "step": 2628
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.720171316128296e-05,
+      "loss": 0.102,
+      "step": 2629
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.719943259110092e-05,
+      "loss": 0.097,
+      "step": 2630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.7197151243288738e-05,
+      "loss": 0.0997,
+      "step": 2631
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7194869118092824e-05,
+      "loss": 0.1055,
+      "step": 2632
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.719258621575967e-05,
+      "loss": 0.1034,
+      "step": 2633
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7190302536535876e-05,
+      "loss": 0.1119,
+      "step": 2634
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7188018080668095e-05,
+      "loss": 0.1043,
+      "step": 2635
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7185732848403082e-05,
+      "loss": 0.1194,
+      "step": 2636
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7183446839987667e-05,
+      "loss": 0.1032,
+      "step": 2637
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7181160055668772e-05,
+      "loss": 0.0988,
+      "step": 2638
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7178872495693397e-05,
+      "loss": 0.1185,
+      "step": 2639
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.717658416030863e-05,
+      "loss": 0.101,
+      "step": 2640
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7174295049761636e-05,
+      "loss": 0.1144,
+      "step": 2641
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.717200516429967e-05,
+      "loss": 0.0907,
+      "step": 2642
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.716971450417007e-05,
+      "loss": 0.1047,
+      "step": 2643
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.7167423069620258e-05,
+      "loss": 0.1167,
+      "step": 2644
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7165130860897733e-05,
+      "loss": 0.1342,
+      "step": 2645
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.716283787825008e-05,
+      "loss": 0.0983,
+      "step": 2646
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7160544121924982e-05,
+      "loss": 0.112,
+      "step": 2647
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7158249592170185e-05,
+      "loss": 0.1091,
+      "step": 2648
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.7155954289233532e-05,
+      "loss": 0.0937,
+      "step": 2649
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.715365821336294e-05,
+      "loss": 0.1007,
+      "step": 2650
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7151361364806418e-05,
+      "loss": 0.1085,
+      "step": 2651
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.714906374381205e-05,
+      "loss": 0.1143,
+      "step": 2652
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7146765350628015e-05,
+      "loss": 0.1047,
+      "step": 2653
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7144466185502562e-05,
+      "loss": 0.1026,
+      "step": 2654
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.714216624868404e-05,
+      "loss": 0.1034,
+      "step": 2655
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.7139865540420857e-05,
+      "loss": 0.1067,
+      "step": 2656
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.713756406096153e-05,
+      "loss": 0.0952,
+      "step": 2657
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7135261810554645e-05,
+      "loss": 0.1039,
+      "step": 2658
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.713295878944887e-05,
+      "loss": 0.0987,
+      "step": 2659
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.713065499789297e-05,
+      "loss": 0.1071,
+      "step": 2660
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7128350436135777e-05,
+      "loss": 0.1045,
+      "step": 2661
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.712604510442621e-05,
+      "loss": 0.0939,
+      "step": 2662
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7123739003013283e-05,
+      "loss": 0.0847,
+      "step": 2663
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.7121432132146077e-05,
+      "loss": 0.1128,
+      "step": 2664
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7119124492073768e-05,
+      "loss": 0.1084,
+      "step": 2665
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7116816083045603e-05,
+      "loss": 0.0882,
+      "step": 2666
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7114506905310928e-05,
+      "loss": 0.0821,
+      "step": 2667
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.711219695911916e-05,
+      "loss": 0.1064,
+      "step": 2668
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.71098862447198e-05,
+      "loss": 0.1119,
+      "step": 2669
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.710757476236244e-05,
+      "loss": 0.1064,
+      "step": 2670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7105262512296748e-05,
+      "loss": 0.1049,
+      "step": 2671
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7102949494772472e-05,
+      "loss": 0.0895,
+      "step": 2672
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7100635710039452e-05,
+      "loss": 0.095,
+      "step": 2673
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.70983211583476e-05,
+      "loss": 0.105,
+      "step": 2674
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7096005839946928e-05,
+      "loss": 0.1014,
+      "step": 2675
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.709368975508751e-05,
+      "loss": 0.0925,
+      "step": 2676
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7091372904019515e-05,
+      "loss": 0.1031,
+      "step": 2677
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7089055286993196e-05,
+      "loss": 0.1034,
+      "step": 2678
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7086736904258878e-05,
+      "loss": 0.1015,
+      "step": 2679
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.708441775606698e-05,
+      "loss": 0.1131,
+      "step": 2680
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7082097842668006e-05,
+      "loss": 0.0956,
+      "step": 2681
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7079777164312522e-05,
+      "loss": 0.0965,
+      "step": 2682
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.70774557212512e-05,
+      "loss": 0.1081,
+      "step": 2683
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7075133513734785e-05,
+      "loss": 0.1038,
+      "step": 2684
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7072810542014104e-05,
+      "loss": 0.1118,
+      "step": 2685
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7070486806340065e-05,
+      "loss": 0.134,
+      "step": 2686
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7068162306963666e-05,
+      "loss": 0.1173,
+      "step": 2687
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7065837044135973e-05,
+      "loss": 0.113,
+      "step": 2688
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7063511018108155e-05,
+      "loss": 0.1032,
+      "step": 2689
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.7061184229131446e-05,
+      "loss": 0.1108,
+      "step": 2690
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.705885667745717e-05,
+      "loss": 0.1121,
+      "step": 2691
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7056528363336733e-05,
+      "loss": 0.1143,
+      "step": 2692
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7054199287021625e-05,
+      "loss": 0.1101,
+      "step": 2693
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7051869448763406e-05,
+      "loss": 0.0962,
+      "step": 2694
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.704953884881374e-05,
+      "loss": 0.1023,
+      "step": 2695
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.7047207487424358e-05,
+      "loss": 0.1164,
+      "step": 2696
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.704487536484707e-05,
+      "loss": 0.1012,
+      "step": 2697
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.704254248133378e-05,
+      "loss": 0.1072,
+      "step": 2698
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.704020883713647e-05,
+      "loss": 0.1182,
+      "step": 2699
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7037874432507204e-05,
+      "loss": 0.1121,
+      "step": 2700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7035539267698125e-05,
+      "loss": 0.1065,
+      "step": 2701
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7033203342961458e-05,
+      "loss": 0.104,
+      "step": 2702
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.7030866658549516e-05,
+      "loss": 0.101,
+      "step": 2703
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.702852921471469e-05,
+      "loss": 0.1051,
+      "step": 2704
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7026191011709456e-05,
+      "loss": 0.102,
+      "step": 2705
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7023852049786364e-05,
+      "loss": 0.1154,
+      "step": 2706
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7021512329198056e-05,
+      "loss": 0.1042,
+      "step": 2707
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7019171850197254e-05,
+      "loss": 0.1217,
+      "step": 2708
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7016830613036753e-05,
+      "loss": 0.1018,
+      "step": 2709
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.7014488617969436e-05,
+      "loss": 0.0962,
+      "step": 2710
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.701214586524827e-05,
+      "loss": 0.1034,
+      "step": 2711
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.700980235512631e-05,
+      "loss": 0.0888,
+      "step": 2712
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7007458087856678e-05,
+      "loss": 0.0991,
+      "step": 2713
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.700511306369258e-05,
+      "loss": 0.0919,
+      "step": 2714
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7002767282887313e-05,
+      "loss": 0.1198,
+      "step": 2715
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.7000420745694256e-05,
+      "loss": 0.1047,
+      "step": 2716
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6998073452366858e-05,
+      "loss": 0.1094,
+      "step": 2717
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.699572540315866e-05,
+      "loss": 0.1141,
+      "step": 2718
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6993376598323274e-05,
+      "loss": 0.1139,
+      "step": 2719
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.699102703811441e-05,
+      "loss": 0.1188,
+      "step": 2720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6988676722785848e-05,
+      "loss": 0.1085,
+      "step": 2721
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.698632565259145e-05,
+      "loss": 0.1059,
+      "step": 2722
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6983973827785165e-05,
+      "loss": 0.11,
+      "step": 2723
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6981621248621013e-05,
+      "loss": 0.1319,
+      "step": 2724
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.697926791535311e-05,
+      "loss": 0.1171,
+      "step": 2725
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.697691382823564e-05,
+      "loss": 0.0947,
+      "step": 2726
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6974558987522876e-05,
+      "loss": 0.104,
+      "step": 2727
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6972203393469174e-05,
+      "loss": 0.1089,
+      "step": 2728
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.696984704632896e-05,
+      "loss": 0.1132,
+      "step": 2729
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6967489946356762e-05,
+      "loss": 0.1039,
+      "step": 2730
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.696513209380717e-05,
+      "loss": 0.1174,
+      "step": 2731
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6962773488934857e-05,
+      "loss": 0.1269,
+      "step": 2732
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6960414131994584e-05,
+      "loss": 0.1073,
+      "step": 2733
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6958054023241202e-05,
+      "loss": 0.108,
+      "step": 2734
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6955693162929622e-05,
+      "loss": 0.1171,
+      "step": 2735
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.695333155131485e-05,
+      "loss": 0.1064,
+      "step": 2736
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.695096918865197e-05,
+      "loss": 0.1191,
+      "step": 2737
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6948606075196148e-05,
+      "loss": 0.1114,
+      "step": 2738
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.694624221120263e-05,
+      "loss": 0.1167,
+      "step": 2739
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.694387759692674e-05,
+      "loss": 0.1077,
+      "step": 2740
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6941512232623894e-05,
+      "loss": 0.1115,
+      "step": 2741
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6939146118549574e-05,
+      "loss": 0.1154,
+      "step": 2742
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.693677925495935e-05,
+      "loss": 0.1322,
+      "step": 2743
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.693441164210888e-05,
+      "loss": 0.0983,
+      "step": 2744
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6932043280253892e-05,
+      "loss": 0.1104,
+      "step": 2745
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.69296741696502e-05,
+      "loss": 0.1029,
+      "step": 2746
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6927304310553697e-05,
+      "loss": 0.1011,
+      "step": 2747
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6924933703220358e-05,
+      "loss": 0.1195,
+      "step": 2748
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.692256234790624e-05,
+      "loss": 0.0945,
+      "step": 2749
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.692019024486748e-05,
+      "loss": 0.121,
+      "step": 2750
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.691781739436029e-05,
+      "loss": 0.0974,
+      "step": 2751
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6915443796640977e-05,
+      "loss": 0.0949,
+      "step": 2752
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6913069451965912e-05,
+      "loss": 0.1154,
+      "step": 2753
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.691069436059156e-05,
+      "loss": 0.1117,
+      "step": 2754
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6908318522774455e-05,
+      "loss": 0.1171,
+      "step": 2755
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6905941938771226e-05,
+      "loss": 0.1057,
+      "step": 2756
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6903564608838565e-05,
+      "loss": 0.1434,
+      "step": 2757
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6901186533233256e-05,
+      "loss": 0.1157,
+      "step": 2758
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6898807712212164e-05,
+      "loss": 0.1043,
+      "step": 2759
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6896428146032234e-05,
+      "loss": 0.1195,
+      "step": 2760
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6894047834950486e-05,
+      "loss": 0.1128,
+      "step": 2761
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6891666779224023e-05,
+      "loss": 0.1075,
+      "step": 2762
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.688928497911003e-05,
+      "loss": 0.1123,
+      "step": 2763
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6886902434865775e-05,
+      "loss": 0.107,
+      "step": 2764
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6884519146748596e-05,
+      "loss": 0.1233,
+      "step": 2765
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6882135115015925e-05,
+      "loss": 0.1174,
+      "step": 2766
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6879750339925266e-05,
+      "loss": 0.1068,
+      "step": 2767
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6877364821734205e-05,
+      "loss": 0.1109,
+      "step": 2768
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.687497856070041e-05,
+      "loss": 0.1179,
+      "step": 2769
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.6872591557081617e-05,
+      "loss": 0.1078,
+      "step": 2770
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.6870203811135665e-05,
+      "loss": 0.1068,
+      "step": 2771
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.686781532312046e-05,
+      "loss": 0.1055,
+      "step": 2772
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.686542609329398e-05,
+      "loss": 0.119,
+      "step": 2773
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.6863036121914302e-05,
+      "loss": 0.1138,
+      "step": 2774
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.686064540923957e-05,
+      "loss": 0.1193,
+      "step": 2775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6858253955528007e-05,
+      "loss": 0.132,
+      "step": 2776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6855861761037923e-05,
+      "loss": 0.1142,
+      "step": 2777
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.685346882602771e-05,
+      "loss": 0.117,
+      "step": 2778
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6851075150755832e-05,
+      "loss": 0.1139,
+      "step": 2779
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.684868073548083e-05,
+      "loss": 0.1106,
+      "step": 2780
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6846285580461342e-05,
+      "loss": 0.1024,
+      "step": 2781
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6843889685956073e-05,
+      "loss": 0.1132,
+      "step": 2782
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.68414930522238e-05,
+      "loss": 0.1003,
+      "step": 2783
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6839095679523402e-05,
+      "loss": 0.1072,
+      "step": 2784
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6836697568113823e-05,
+      "loss": 0.1021,
+      "step": 2785
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6834298718254083e-05,
+      "loss": 0.0976,
+      "step": 2786
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6831899130203294e-05,
+      "loss": 0.1102,
+      "step": 2787
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.682949880422064e-05,
+      "loss": 0.0948,
+      "step": 2788
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.682709774056539e-05,
+      "loss": 0.1096,
+      "step": 2789
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6824695939496885e-05,
+      "loss": 0.1109,
+      "step": 2790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.682229340127455e-05,
+      "loss": 0.112,
+      "step": 2791
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6819890126157898e-05,
+      "loss": 0.1111,
+      "step": 2792
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.68174861144065e-05,
+      "loss": 0.1047,
+      "step": 2793
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6815081366280027e-05,
+      "loss": 0.1056,
+      "step": 2794
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6812675882038224e-05,
+      "loss": 0.1085,
+      "step": 2795
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.681026966194091e-05,
+      "loss": 0.1083,
+      "step": 2796
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.680786270624799e-05,
+      "loss": 0.1078,
+      "step": 2797
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6805455015219444e-05,
+      "loss": 0.1094,
+      "step": 2798
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6803046589115336e-05,
+      "loss": 0.1103,
+      "step": 2799
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.68006374281958e-05,
+      "loss": 0.1192,
+      "step": 2800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6798227532721062e-05,
+      "loss": 0.1071,
+      "step": 2801
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6795816902951423e-05,
+      "loss": 0.1242,
+      "step": 2802
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6793405539147257e-05,
+      "loss": 0.1153,
+      "step": 2803
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6790993441569022e-05,
+      "loss": 0.1172,
+      "step": 2804
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6788580610477258e-05,
+      "loss": 0.1197,
+      "step": 2805
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6786167046132577e-05,
+      "loss": 0.0953,
+      "step": 2806
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.678375274879568e-05,
+      "loss": 0.1234,
+      "step": 2807
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6781337718727337e-05,
+      "loss": 0.1187,
+      "step": 2808
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6778921956188406e-05,
+      "loss": 0.0974,
+      "step": 2809
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.677650546143982e-05,
+      "loss": 0.1123,
+      "step": 2810
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6774088234742587e-05,
+      "loss": 0.1086,
+      "step": 2811
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6771670276357803e-05,
+      "loss": 0.114,
+      "step": 2812
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.676925158654663e-05,
+      "loss": 0.1061,
+      "step": 2813
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6766832165570327e-05,
+      "loss": 0.0946,
+      "step": 2814
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.6764412013690222e-05,
+      "loss": 0.1153,
+      "step": 2815
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6761991131167712e-05,
+      "loss": 0.1052,
+      "step": 2816
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.675956951826429e-05,
+      "loss": 0.1035,
+      "step": 2817
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6757147175241523e-05,
+      "loss": 0.1062,
+      "step": 2818
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.675472410236105e-05,
+      "loss": 0.1132,
+      "step": 2819
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6752300299884597e-05,
+      "loss": 0.1255,
+      "step": 2820
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.6749875768073964e-05,
+      "loss": 0.1212,
+      "step": 2821
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.674745050719103e-05,
+      "loss": 0.1213,
+      "step": 2822
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.674502451749776e-05,
+      "loss": 0.1189,
+      "step": 2823
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.6742597799256182e-05,
+      "loss": 0.1195,
+      "step": 2824
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.6740170352728416e-05,
+      "loss": 0.1119,
+      "step": 2825
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.6737742178176666e-05,
+      "loss": 0.1173,
+      "step": 2826
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.6735313275863192e-05,
+      "loss": 0.1038,
+      "step": 2827
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.6732883646050356e-05,
+      "loss": 0.1321,
+      "step": 2828
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.673045328900058e-05,
+      "loss": 0.1122,
+      "step": 2829
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.6728022204976383e-05,
+      "loss": 0.1028,
+      "step": 2830
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.672559039424035e-05,
+      "loss": 0.1242,
+      "step": 2831
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.6723157857055145e-05,
+      "loss": 0.1184,
+      "step": 2832
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.6720724593683512e-05,
+      "loss": 0.1144,
+      "step": 2833
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.671829060438828e-05,
+      "loss": 0.1119,
+      "step": 2834
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.671585588943234e-05,
+      "loss": 0.122,
+      "step": 2835
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6713420449078688e-05,
+      "loss": 0.1019,
+      "step": 2836
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.671098428359037e-05,
+      "loss": 0.1122,
+      "step": 2837
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6708547393230526e-05,
+      "loss": 0.1008,
+      "step": 2838
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6706109778262373e-05,
+      "loss": 0.1195,
+      "step": 2839
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6703671438949202e-05,
+      "loss": 0.1274,
+      "step": 2840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.6701232375554384e-05,
+      "loss": 0.1113,
+      "step": 2841
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6698792588341372e-05,
+      "loss": 0.1127,
+      "step": 2842
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6696352077573697e-05,
+      "loss": 0.1098,
+      "step": 2843
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6693910843514955e-05,
+      "loss": 0.1177,
+      "step": 2844
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.669146888642884e-05,
+      "loss": 0.1098,
+      "step": 2845
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6689026206579105e-05,
+      "loss": 0.1177,
+      "step": 2846
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.66865828042296e-05,
+      "loss": 0.1191,
+      "step": 2847
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.6684138679644237e-05,
+      "loss": 0.1209,
+      "step": 2848
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6681693833087018e-05,
+      "loss": 0.1131,
+      "step": 2849
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.667924826482201e-05,
+      "loss": 0.1064,
+      "step": 2850
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6676801975113374e-05,
+      "loss": 0.1351,
+      "step": 2851
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.667435496422533e-05,
+      "loss": 0.1227,
+      "step": 2852
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6671907232422196e-05,
+      "loss": 0.1333,
+      "step": 2853
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.6669458779968357e-05,
+      "loss": 0.1183,
+      "step": 2854
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6667009607128273e-05,
+      "loss": 0.1164,
+      "step": 2855
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6664559714166486e-05,
+      "loss": 0.1109,
+      "step": 2856
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6662109101347613e-05,
+      "loss": 0.1132,
+      "step": 2857
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.665965776893636e-05,
+      "loss": 0.1145,
+      "step": 2858
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6657205717197495e-05,
+      "loss": 0.1112,
+      "step": 2859
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6654752946395872e-05,
+      "loss": 0.1237,
+      "step": 2860
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.6652299456796422e-05,
+      "loss": 0.1177,
+      "step": 2861
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6649845248664154e-05,
+      "loss": 0.1036,
+      "step": 2862
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6647390322264153e-05,
+      "loss": 0.1072,
+      "step": 2863
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6644934677861582e-05,
+      "loss": 0.1304,
+      "step": 2864
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6642478315721687e-05,
+      "loss": 0.1303,
+      "step": 2865
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6640021236109774e-05,
+      "loss": 0.0931,
+      "step": 2866
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.6637563439291246e-05,
+      "loss": 0.1163,
+      "step": 2867
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.663510492553158e-05,
+      "loss": 0.1241,
+      "step": 2868
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6632645695096318e-05,
+      "loss": 0.117,
+      "step": 2869
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.66301857482511e-05,
+      "loss": 0.1221,
+      "step": 2870
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.662772508526162e-05,
+      "loss": 0.1133,
+      "step": 2871
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6625263706393664e-05,
+      "loss": 0.1035,
+      "step": 2872
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.6622801611913097e-05,
+      "loss": 0.1309,
+      "step": 2873
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.662033880208585e-05,
+      "loss": 0.1152,
+      "step": 2874
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.661787527717794e-05,
+      "loss": 0.1131,
+      "step": 2875
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6615411037455462e-05,
+      "loss": 0.1037,
+      "step": 2876
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6612946083184582e-05,
+      "loss": 0.1231,
+      "step": 2877
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.661048041463155e-05,
+      "loss": 0.1179,
+      "step": 2878
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6608014032062683e-05,
+      "loss": 0.1267,
+      "step": 2879
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.6605546935744386e-05,
+      "loss": 0.1182,
+      "step": 2880
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6603079125943136e-05,
+      "loss": 0.1199,
+      "step": 2881
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.660061060292549e-05,
+      "loss": 0.1125,
+      "step": 2882
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6598141366958076e-05,
+      "loss": 0.1142,
+      "step": 2883
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6595671418307606e-05,
+      "loss": 0.1157,
+      "step": 2884
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6593200757240864e-05,
+      "loss": 0.1243,
+      "step": 2885
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6590729384024714e-05,
+      "loss": 0.1104,
+      "step": 2886
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.6588257298926093e-05,
+      "loss": 0.1143,
+      "step": 2887
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.658578450221202e-05,
+      "loss": 0.1177,
+      "step": 2888
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6583310994149585e-05,
+      "loss": 0.1213,
+      "step": 2889
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6580836775005965e-05,
+      "loss": 0.1257,
+      "step": 2890
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6578361845048404e-05,
+      "loss": 0.1211,
+      "step": 2891
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6575886204544223e-05,
+      "loss": 0.1131,
+      "step": 2892
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6573409853760826e-05,
+      "loss": 0.1282,
+      "step": 2893
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.6570932792965687e-05,
+      "loss": 0.1154,
+      "step": 2894
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6568455022426365e-05,
+      "loss": 0.1151,
+      "step": 2895
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.656597654241049e-05,
+      "loss": 0.118,
+      "step": 2896
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.656349735318576e-05,
+      "loss": 0.1106,
+      "step": 2897
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.656101745501997e-05,
+      "loss": 0.1082,
+      "step": 2898
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6558536848180974e-05,
+      "loss": 0.1144,
+      "step": 2899
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.6556055532936714e-05,
+      "loss": 0.1132,
+      "step": 2900
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.65535735095552e-05,
+      "loss": 0.1136,
+      "step": 2901
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6551090778304525e-05,
+      "loss": 0.1283,
+      "step": 2902
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.1195,
+      "step": 2903
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6546123193268426e-05,
+      "loss": 0.1059,
+      "step": 2904
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6543638340019567e-05,
+      "loss": 0.1097,
+      "step": 2905
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.654115277997467e-05,
+      "loss": 0.1154,
+      "step": 2906
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.6538666513402202e-05,
+      "loss": 0.1213,
+      "step": 2907
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6536179540570723e-05,
+      "loss": 0.1184,
+      "step": 2908
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6533691861748846e-05,
+      "loss": 0.1159,
+      "step": 2909
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6531203477205274e-05,
+      "loss": 0.1116,
+      "step": 2910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6528714387208794e-05,
+      "loss": 0.1165,
+      "step": 2911
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6526224592028247e-05,
+      "loss": 0.0996,
+      "step": 2912
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6523734091932567e-05,
+      "loss": 0.1226,
+      "step": 2913
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6521242887190764e-05,
+      "loss": 0.1067,
+      "step": 2914
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6518750978071914e-05,
+      "loss": 0.1295,
+      "step": 2915
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6516258364845176e-05,
+      "loss": 0.1358,
+      "step": 2916
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6513765047779785e-05,
+      "loss": 0.1233,
+      "step": 2917
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6511271027145054e-05,
+      "loss": 0.1203,
+      "step": 2918
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6508776303210363e-05,
+      "loss": 0.1161,
+      "step": 2919
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6506280876245172e-05,
+      "loss": 0.1318,
+      "step": 2920
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6503784746519026e-05,
+      "loss": 0.0957,
+      "step": 2921
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.650128791430154e-05,
+      "loss": 0.1322,
+      "step": 2922
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6498790379862396e-05,
+      "loss": 0.1286,
+      "step": 2923
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.649629214347136e-05,
+      "loss": 0.1002,
+      "step": 2924
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6493793205398282e-05,
+      "loss": 0.1124,
+      "step": 2925
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6491293565913073e-05,
+      "loss": 0.1121,
+      "step": 2926
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.648879322528572e-05,
+      "loss": 0.1191,
+      "step": 2927
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.64862921837863e-05,
+      "loss": 0.1168,
+      "step": 2928
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6483790441684957e-05,
+      "loss": 0.1243,
+      "step": 2929
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.648128799925191e-05,
+      "loss": 0.1136,
+      "step": 2930
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.647878485675745e-05,
+      "loss": 0.1311,
+      "step": 2931
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.647628101447195e-05,
+      "loss": 0.1213,
+      "step": 2932
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.6473776472665858e-05,
+      "loss": 0.1195,
+      "step": 2933
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.64712712316097e-05,
+      "loss": 0.1357,
+      "step": 2934
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.646876529157407e-05,
+      "loss": 0.1225,
+      "step": 2935
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6466258652829638e-05,
+      "loss": 0.1218,
+      "step": 2936
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6463751315647162e-05,
+      "loss": 0.118,
+      "step": 2937
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6461243280297458e-05,
+      "loss": 0.1065,
+      "step": 2938
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6458734547051426e-05,
+      "loss": 0.1036,
+      "step": 2939
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.6456225116180044e-05,
+      "loss": 0.106,
+      "step": 2940
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6453714987954367e-05,
+      "loss": 0.1265,
+      "step": 2941
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6451204162645512e-05,
+      "loss": 0.1166,
+      "step": 2942
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6448692640524684e-05,
+      "loss": 0.1286,
+      "step": 2943
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6446180421863157e-05,
+      "loss": 0.1092,
+      "step": 2944
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6443667506932285e-05,
+      "loss": 0.1001,
+      "step": 2945
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.6441153896003496e-05,
+      "loss": 0.1309,
+      "step": 2946
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.643863958934829e-05,
+      "loss": 0.1359,
+      "step": 2947
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6436124587238245e-05,
+      "loss": 0.0981,
+      "step": 2948
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.643360888994501e-05,
+      "loss": 0.1041,
+      "step": 2949
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.643109249774031e-05,
+      "loss": 0.102,
+      "step": 2950
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6428575410895963e-05,
+      "loss": 0.1128,
+      "step": 2951
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6426057629683826e-05,
+      "loss": 0.1265,
+      "step": 2952
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.6423539154375864e-05,
+      "loss": 0.1169,
+      "step": 2953
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.64210199852441e-05,
+      "loss": 0.118,
+      "step": 2954
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6418500122560638e-05,
+      "loss": 0.111,
+      "step": 2955
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6415979566597654e-05,
+      "loss": 0.1314,
+      "step": 2956
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.64134583176274e-05,
+      "loss": 0.1191,
+      "step": 2957
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.64109363759222e-05,
+      "loss": 0.1101,
+      "step": 2958
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.6408413741754466e-05,
+      "loss": 0.109,
+      "step": 2959
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6405890415396664e-05,
+      "loss": 0.1309,
+      "step": 2960
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6403366397121345e-05,
+      "loss": 0.1113,
+      "step": 2961
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6400841687201143e-05,
+      "loss": 0.1121,
+      "step": 2962
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6398316285908754e-05,
+      "loss": 0.1227,
+      "step": 2963
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6395790193516954e-05,
+      "loss": 0.1138,
+      "step": 2964
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6393263410298592e-05,
+      "loss": 0.1292,
+      "step": 2965
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.6390735936526594e-05,
+      "loss": 0.1293,
+      "step": 2966
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.638820777247396e-05,
+      "loss": 0.1094,
+      "step": 2967
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6385678918413763e-05,
+      "loss": 0.1261,
+      "step": 2968
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.638314937461915e-05,
+      "loss": 0.1216,
+      "step": 2969
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6380619141363343e-05,
+      "loss": 0.1166,
+      "step": 2970
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.6378088218919645e-05,
+      "loss": 0.119,
+      "step": 2971
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.637555660756142e-05,
+      "loss": 0.1172,
+      "step": 2972
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.637302430756212e-05,
+      "loss": 0.1107,
+      "step": 2973
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6370491319195266e-05,
+      "loss": 0.1142,
+      "step": 2974
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.636795764273445e-05,
+      "loss": 0.1183,
+      "step": 2975
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.636542327845334e-05,
+      "loss": 0.1289,
+      "step": 2976
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6362888226625684e-05,
+      "loss": 0.1325,
+      "step": 2977
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6360352487525292e-05,
+      "loss": 0.1204,
+      "step": 2978
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.6357816061426065e-05,
+      "loss": 0.1239,
+      "step": 2979
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6355278948601968e-05,
+      "loss": 0.12,
+      "step": 2980
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6352741149327035e-05,
+      "loss": 0.1423,
+      "step": 2981
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6350202663875385e-05,
+      "loss": 0.1154,
+      "step": 2982
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6347663492521207e-05,
+      "loss": 0.1166,
+      "step": 2983
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6345123635538764e-05,
+      "loss": 0.1083,
+      "step": 2984
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6342583093202388e-05,
+      "loss": 0.1146,
+      "step": 2985
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.6340041865786495e-05,
+      "loss": 0.1414,
+      "step": 2986
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.633749995356557e-05,
+      "loss": 0.1317,
+      "step": 2987
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6334957356814168e-05,
+      "loss": 0.112,
+      "step": 2988
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6332414075806923e-05,
+      "loss": 0.1316,
+      "step": 2989
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6329870110818546e-05,
+      "loss": 0.123,
+      "step": 2990
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.6327325462123816e-05,
+      "loss": 0.1311,
+      "step": 2991
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.632478012999758e-05,
+      "loss": 0.104,
+      "step": 2992
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6322234114714774e-05,
+      "loss": 0.1157,
+      "step": 2993
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6319687416550393e-05,
+      "loss": 0.1172,
+      "step": 2994
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6317140035779525e-05,
+      "loss": 0.146,
+      "step": 2995
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.631459197267731e-05,
+      "loss": 0.1318,
+      "step": 2996
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.631204322751897e-05,
+      "loss": 0.1158,
+      "step": 2997
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6309493800579805e-05,
+      "loss": 0.1181,
+      "step": 2998
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.6306943692135188e-05,
+      "loss": 0.108,
+      "step": 2999
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6304392902460558e-05,
+      "loss": 0.1212,
+      "step": 3000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6301841431831436e-05,
+      "loss": 0.124,
+      "step": 3001
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.629928928052341e-05,
+      "loss": 0.1235,
+      "step": 3002
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6296736448812148e-05,
+      "loss": 0.129,
+      "step": 3003
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6294182936973387e-05,
+      "loss": 0.1117,
+      "step": 3004
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.6291628745282943e-05,
+      "loss": 0.1076,
+      "step": 3005
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6289073874016695e-05,
+      "loss": 0.1119,
+      "step": 3006
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6286518323450603e-05,
+      "loss": 0.1218,
+      "step": 3007
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6283962093860698e-05,
+      "loss": 0.1216,
+      "step": 3008
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.628140518552309e-05,
+      "loss": 0.1154,
+      "step": 3009
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6278847598713952e-05,
+      "loss": 0.1095,
+      "step": 3010
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.627628933370954e-05,
+      "loss": 0.1154,
+      "step": 3011
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6273730390786177e-05,
+      "loss": 0.124,
+      "step": 3012
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6271170770220258e-05,
+      "loss": 0.1225,
+      "step": 3013
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.626861047228826e-05,
+      "loss": 0.1111,
+      "step": 3014
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.626604949726673e-05,
+      "loss": 0.1249,
+      "step": 3015
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.626348784543228e-05,
+      "loss": 0.1268,
+      "step": 3016
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6260925517061604e-05,
+      "loss": 0.1237,
+      "step": 3017
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.625836251243146e-05,
+      "loss": 0.1349,
+      "step": 3018
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6255798831818693e-05,
+      "loss": 0.1447,
+      "step": 3019
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.625323447550021e-05,
+      "loss": 0.1253,
+      "step": 3020
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6250669443752994e-05,
+      "loss": 0.1112,
+      "step": 3021
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.62481037368541e-05,
+      "loss": 0.1176,
+      "step": 3022
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6245537355080657e-05,
+      "loss": 0.1211,
+      "step": 3023
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6242970298709867e-05,
+      "loss": 0.1245,
+      "step": 3024
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6240402568019007e-05,
+      "loss": 0.1171,
+      "step": 3025
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6237834163285423e-05,
+      "loss": 0.1434,
+      "step": 3026
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6235265084786532e-05,
+      "loss": 0.1127,
+      "step": 3027
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6232695332799835e-05,
+      "loss": 0.1167,
+      "step": 3028
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6230124907602892e-05,
+      "loss": 0.1172,
+      "step": 3029
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.622755380947334e-05,
+      "loss": 0.1115,
+      "step": 3030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6224982038688893e-05,
+      "loss": 0.1079,
+      "step": 3031
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.6222409595527334e-05,
+      "loss": 0.1318,
+      "step": 3032
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6219836480266528e-05,
+      "loss": 0.1261,
+      "step": 3033
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.621726269318439e-05,
+      "loss": 0.1452,
+      "step": 3034
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6214688234558927e-05,
+      "loss": 0.1236,
+      "step": 3035
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6212113104668216e-05,
+      "loss": 0.1196,
+      "step": 3036
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6209537303790402e-05,
+      "loss": 0.1137,
+      "step": 3037
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.6206960832203704e-05,
+      "loss": 0.1304,
+      "step": 3038
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6204383690186417e-05,
+      "loss": 0.1342,
+      "step": 3039
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.62018058780169e-05,
+      "loss": 0.1145,
+      "step": 3040
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6199227395973588e-05,
+      "loss": 0.1047,
+      "step": 3041
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6196648244334997e-05,
+      "loss": 0.1225,
+      "step": 3042
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6194068423379702e-05,
+      "loss": 0.1229,
+      "step": 3043
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.6191487933386364e-05,
+      "loss": 0.1111,
+      "step": 3044
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.61889067746337e-05,
+      "loss": 0.121,
+      "step": 3045
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.618632494740051e-05,
+      "loss": 0.1294,
+      "step": 3046
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6183742451965668e-05,
+      "loss": 0.1293,
+      "step": 3047
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.618115928860811e-05,
+      "loss": 0.1235,
+      "step": 3048
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.617857545760686e-05,
+      "loss": 0.127,
+      "step": 3049
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.6175990959240996e-05,
+      "loss": 0.1357,
+      "step": 3050
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.617340579378968e-05,
+      "loss": 0.138,
+      "step": 3051
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.617081996153214e-05,
+      "loss": 0.1301,
+      "step": 3052
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6168233462747685e-05,
+      "loss": 0.1208,
+      "step": 3053
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.616564629771568e-05,
+      "loss": 0.1129,
+      "step": 3054
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6163058466715584e-05,
+      "loss": 0.1181,
+      "step": 3055
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6160469970026906e-05,
+      "loss": 0.1455,
+      "step": 3056
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.615788080792924e-05,
+      "loss": 0.1206,
+      "step": 3057
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.6155290980702245e-05,
+      "loss": 0.1133,
+      "step": 3058
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.615270048862566e-05,
+      "loss": 0.1224,
+      "step": 3059
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6150109331979288e-05,
+      "loss": 0.1298,
+      "step": 3060
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.614751751104301e-05,
+      "loss": 0.1202,
+      "step": 3061
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6144925026096774e-05,
+      "loss": 0.1184,
+      "step": 3062
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.6142331877420598e-05,
+      "loss": 0.1325,
+      "step": 3063
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.613973806529458e-05,
+      "loss": 0.1407,
+      "step": 3064
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.613714358999888e-05,
+      "loss": 0.124,
+      "step": 3065
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6134548451813738e-05,
+      "loss": 0.1351,
+      "step": 3066
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.6131952651019463e-05,
+      "loss": 0.1083,
+      "step": 3067
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.612935618789643e-05,
+      "loss": 0.1288,
+      "step": 3068
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.612675906272509e-05,
+      "loss": 0.1254,
+      "step": 3069
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.612416127578597e-05,
+      "loss": 0.1186,
+      "step": 3070
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.612156282735966e-05,
+      "loss": 0.113,
+      "step": 3071
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.611896371772683e-05,
+      "loss": 0.1177,
+      "step": 3072
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.611636394716821e-05,
+      "loss": 0.1134,
+      "step": 3073
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.6113763515964614e-05,
+      "loss": 0.1143,
+      "step": 3074
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.611116242439692e-05,
+      "loss": 0.1355,
+      "step": 3075
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.610856067274608e-05,
+      "loss": 0.1144,
+      "step": 3076
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.610595826129311e-05,
+      "loss": 0.12,
+      "step": 3077
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.610335519031911e-05,
+      "loss": 0.1303,
+      "step": 3078
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6100751460105244e-05,
+      "loss": 0.1373,
+      "step": 3079
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6098147070932748e-05,
+      "loss": 0.1225,
+      "step": 3080
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6095542023082925e-05,
+      "loss": 0.1186,
+      "step": 3081
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.609293631683716e-05,
+      "loss": 0.1168,
+      "step": 3082
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6090329952476897e-05,
+      "loss": 0.113,
+      "step": 3083
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.6087722930283657e-05,
+      "loss": 0.1068,
+      "step": 3084
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.608511525053903e-05,
+      "loss": 0.127,
+      "step": 3085
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.608250691352469e-05,
+      "loss": 0.1303,
+      "step": 3086
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6079897919522352e-05,
+      "loss": 0.1309,
+      "step": 3087
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6077288268813834e-05,
+      "loss": 0.1262,
+      "step": 3088
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.607467796168101e-05,
+      "loss": 0.1267,
+      "step": 3089
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.607206699840582e-05,
+      "loss": 0.1153,
+      "step": 3090
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.6069455379270284e-05,
+      "loss": 0.1198,
+      "step": 3091
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6066843104556494e-05,
+      "loss": 0.1298,
+      "step": 3092
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.606423017454661e-05,
+      "loss": 0.1285,
+      "step": 3093
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6061616589522852e-05,
+      "loss": 0.1251,
+      "step": 3094
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6059002349767523e-05,
+      "loss": 0.1346,
+      "step": 3095
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6056387455563e-05,
+      "loss": 0.1214,
+      "step": 3096
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6053771907191724e-05,
+      "loss": 0.1158,
+      "step": 3097
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.60511557049362e-05,
+      "loss": 0.1166,
+      "step": 3098
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6048538849079024e-05,
+      "loss": 0.1155,
+      "step": 3099
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.604592133990284e-05,
+      "loss": 0.1188,
+      "step": 3100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.604330317769037e-05,
+      "loss": 0.1228,
+      "step": 3101
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.604068436272441e-05,
+      "loss": 0.1288,
+      "step": 3102
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6038064895287835e-05,
+      "loss": 0.1277,
+      "step": 3103
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6035444775663573e-05,
+      "loss": 0.1291,
+      "step": 3104
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.603282400413463e-05,
+      "loss": 0.1243,
+      "step": 3105
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6030202580984083e-05,
+      "loss": 0.1322,
+      "step": 3106
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6027580506495078e-05,
+      "loss": 0.1325,
+      "step": 3107
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6024957780950837e-05,
+      "loss": 0.1251,
+      "step": 3108
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.6022334404634646e-05,
+      "loss": 0.1213,
+      "step": 3109
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.601971037782986e-05,
+      "loss": 0.1262,
+      "step": 3110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.601708570081991e-05,
+      "loss": 0.1216,
+      "step": 3111
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6014460373888293e-05,
+      "loss": 0.1224,
+      "step": 3112
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6011834397318576e-05,
+      "loss": 0.1063,
+      "step": 3113
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.60092077713944e-05,
+      "loss": 0.1196,
+      "step": 3114
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.6006580496399477e-05,
+      "loss": 0.1292,
+      "step": 3115
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.600395257261758e-05,
+      "loss": 0.1182,
+      "step": 3116
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.600132400033256e-05,
+      "loss": 0.123,
+      "step": 3117
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.599869477982834e-05,
+      "loss": 0.1186,
+      "step": 3118
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5996064911388907e-05,
+      "loss": 0.1438,
+      "step": 3119
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5993434395298314e-05,
+      "loss": 0.1255,
+      "step": 3120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.59908032318407e-05,
+      "loss": 0.1261,
+      "step": 3121
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.598817142130026e-05,
+      "loss": 0.1248,
+      "step": 3122
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5985538963961263e-05,
+      "loss": 0.1393,
+      "step": 3123
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.5982905860108045e-05,
+      "loss": 0.1308,
+      "step": 3124
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.598027211002502e-05,
+      "loss": 0.1246,
+      "step": 3125
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5977637713996664e-05,
+      "loss": 0.1305,
+      "step": 3126
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5975002672307524e-05,
+      "loss": 0.1145,
+      "step": 3127
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5972366985242216e-05,
+      "loss": 0.1391,
+      "step": 3128
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5969730653085434e-05,
+      "loss": 0.1278,
+      "step": 3129
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5967093676121928e-05,
+      "loss": 0.1302,
+      "step": 3130
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.596445605463653e-05,
+      "loss": 0.1213,
+      "step": 3131
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.596181778891414e-05,
+      "loss": 0.1163,
+      "step": 3132
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5959178879239716e-05,
+      "loss": 0.1417,
+      "step": 3133
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.59565393258983e-05,
+      "loss": 0.1238,
+      "step": 3134
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.595389912917499e-05,
+      "loss": 0.1285,
+      "step": 3135
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.595125828935497e-05,
+      "loss": 0.1169,
+      "step": 3136
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5948616806723478e-05,
+      "loss": 0.1223,
+      "step": 3137
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.594597468156583e-05,
+      "loss": 0.1209,
+      "step": 3138
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5943331914167405e-05,
+      "loss": 0.138,
+      "step": 3139
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5940688504813664e-05,
+      "loss": 0.1168,
+      "step": 3140
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.593804445379012e-05,
+      "loss": 0.1059,
+      "step": 3141
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5935399761382372e-05,
+      "loss": 0.1381,
+      "step": 3142
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5932754427876073e-05,
+      "loss": 0.1156,
+      "step": 3143
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5930108453556955e-05,
+      "loss": 0.1327,
+      "step": 3144
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5927461838710818e-05,
+      "loss": 0.1261,
+      "step": 3145
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5924814583623528e-05,
+      "loss": 0.1359,
+      "step": 3146
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.592216668858103e-05,
+      "loss": 0.1226,
+      "step": 3147
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.591951815386932e-05,
+      "loss": 0.128,
+      "step": 3148
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5916868979774478e-05,
+      "loss": 0.129,
+      "step": 3149
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5914219166582645e-05,
+      "loss": 0.1242,
+      "step": 3150
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.591156871458004e-05,
+      "loss": 0.117,
+      "step": 3151
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5908917624052943e-05,
+      "loss": 0.1359,
+      "step": 3152
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5906265895287708e-05,
+      "loss": 0.136,
+      "step": 3153
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5903613528570748e-05,
+      "loss": 0.1257,
+      "step": 3154
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.590096052418856e-05,
+      "loss": 0.1244,
+      "step": 3155
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5898306882427696e-05,
+      "loss": 0.1323,
+      "step": 3156
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5895652603574787e-05,
+      "loss": 0.1154,
+      "step": 3157
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5892997687916528e-05,
+      "loss": 0.142,
+      "step": 3158
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5890342135739687e-05,
+      "loss": 0.1426,
+      "step": 3159
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.588768594733109e-05,
+      "loss": 0.1301,
+      "step": 3160
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5885029122977645e-05,
+      "loss": 0.1209,
+      "step": 3161
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.588237166296632e-05,
+      "loss": 0.1294,
+      "step": 3162
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5879713567584152e-05,
+      "loss": 0.1243,
+      "step": 3163
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.587705483711826e-05,
+      "loss": 0.1342,
+      "step": 3164
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5874395471855805e-05,
+      "loss": 0.1301,
+      "step": 3165
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5871735472084043e-05,
+      "loss": 0.1159,
+      "step": 3166
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5869074838090288e-05,
+      "loss": 0.1202,
+      "step": 3167
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.5866413570161912e-05,
+      "loss": 0.12,
+      "step": 3168
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.586375166858638e-05,
+      "loss": 0.1305,
+      "step": 3169
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.58610891336512e-05,
+      "loss": 0.1365,
+      "step": 3170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5858425965643966e-05,
+      "loss": 0.1217,
+      "step": 3171
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.585576216485233e-05,
+      "loss": 0.1314,
+      "step": 3172
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5853097731564017e-05,
+      "loss": 0.1206,
+      "step": 3173
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5850432666066825e-05,
+      "loss": 0.1245,
+      "step": 3174
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5847766968648608e-05,
+      "loss": 0.1286,
+      "step": 3175
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5845100639597294e-05,
+      "loss": 0.135,
+      "step": 3176
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5842433679200887e-05,
+      "loss": 0.118,
+      "step": 3177
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.583976608774745e-05,
+      "loss": 0.1338,
+      "step": 3178
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5837097865525116e-05,
+      "loss": 0.1279,
+      "step": 3179
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5834429012822088e-05,
+      "loss": 0.1155,
+      "step": 3180
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5831759529926635e-05,
+      "loss": 0.1177,
+      "step": 3181
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.5829089417127095e-05,
+      "loss": 0.1272,
+      "step": 3182
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.582641867471187e-05,
+      "loss": 0.1179,
+      "step": 3183
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5823747302969446e-05,
+      "loss": 0.1308,
+      "step": 3184
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.582107530218835e-05,
+      "loss": 0.1324,
+      "step": 3185
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5818402672657205e-05,
+      "loss": 0.1126,
+      "step": 3186
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5815729414664682e-05,
+      "loss": 0.1216,
+      "step": 3187
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.581305552849953e-05,
+      "loss": 0.1291,
+      "step": 3188
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5810381014450557e-05,
+      "loss": 0.124,
+      "step": 3189
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.580770587280665e-05,
+      "loss": 0.128,
+      "step": 3190
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.580503010385676e-05,
+      "loss": 0.1366,
+      "step": 3191
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.58023537078899e-05,
+      "loss": 0.1238,
+      "step": 3192
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.579967668519515e-05,
+      "loss": 0.1231,
+      "step": 3193
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.579699903606167e-05,
+      "loss": 0.167,
+      "step": 3194
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.579432076077868e-05,
+      "loss": 0.1384,
+      "step": 3195
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5791641859635464e-05,
+      "loss": 0.124,
+      "step": 3196
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5788962332921384e-05,
+      "loss": 0.1278,
+      "step": 3197
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5786282180925854e-05,
+      "loss": 0.1493,
+      "step": 3198
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.578360140393837e-05,
+      "loss": 0.1309,
+      "step": 3199
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5780920002248484e-05,
+      "loss": 0.1166,
+      "step": 3200
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.577823797614583e-05,
+      "loss": 0.1218,
+      "step": 3201
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5775555325920096e-05,
+      "loss": 0.1149,
+      "step": 3202
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5772872051861042e-05,
+      "loss": 0.1196,
+      "step": 3203
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.57701881542585e-05,
+      "loss": 0.1208,
+      "step": 3204
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.576750363340236e-05,
+      "loss": 0.1279,
+      "step": 3205
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.576481848958258e-05,
+      "loss": 0.1088,
+      "step": 3206
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5762132723089205e-05,
+      "loss": 0.1195,
+      "step": 3207
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.575944633421232e-05,
+      "loss": 0.1272,
+      "step": 3208
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.575675932324209e-05,
+      "loss": 0.1353,
+      "step": 3209
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.575407169046875e-05,
+      "loss": 0.1159,
+      "step": 3210
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.575138343618259e-05,
+      "loss": 0.1304,
+      "step": 3211
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.574869456067399e-05,
+      "loss": 0.1263,
+      "step": 3212
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5746005064233373e-05,
+      "loss": 0.111,
+      "step": 3213
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.574331494715124e-05,
+      "loss": 0.1354,
+      "step": 3214
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5740624209718157e-05,
+      "loss": 0.1216,
+      "step": 3215
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.573793285222476e-05,
+      "loss": 0.115,
+      "step": 3216
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.573524087496175e-05,
+      "loss": 0.1221,
+      "step": 3217
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.573254827821989e-05,
+      "loss": 0.1328,
+      "step": 3218
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5729855062290024e-05,
+      "loss": 0.125,
+      "step": 3219
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5727161227463045e-05,
+      "loss": 0.1233,
+      "step": 3220
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5724466774029924e-05,
+      "loss": 0.1288,
+      "step": 3221
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.5721771702281694e-05,
+      "loss": 0.1182,
+      "step": 3222
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.571907601250946e-05,
+      "loss": 0.123,
+      "step": 3223
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5716379705004397e-05,
+      "loss": 0.1401,
+      "step": 3224
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.571368278005773e-05,
+      "loss": 0.1281,
+      "step": 3225
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5710985237960762e-05,
+      "loss": 0.1372,
+      "step": 3226
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5708287079004865e-05,
+      "loss": 0.1138,
+      "step": 3227
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5705588303481474e-05,
+      "loss": 0.1275,
+      "step": 3228
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5702888911682092e-05,
+      "loss": 0.1291,
+      "step": 3229
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5700188903898287e-05,
+      "loss": 0.1221,
+      "step": 3230
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5697488280421693e-05,
+      "loss": 0.1276,
+      "step": 3231
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5694787041544014e-05,
+      "loss": 0.1303,
+      "step": 3232
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5692085187557012e-05,
+      "loss": 0.1446,
+      "step": 3233
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5689382718752528e-05,
+      "loss": 0.1323,
+      "step": 3234
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5686679635422463e-05,
+      "loss": 0.1439,
+      "step": 3235
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5683975937858778e-05,
+      "loss": 0.1261,
+      "step": 3236
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5681271626353515e-05,
+      "loss": 0.1258,
+      "step": 3237
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.567856670119877e-05,
+      "loss": 0.1382,
+      "step": 3238
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5675861162686708e-05,
+      "loss": 0.1385,
+      "step": 3239
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5673155011109563e-05,
+      "loss": 0.1156,
+      "step": 3240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5670448246759633e-05,
+      "loss": 0.1295,
+      "step": 3241
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5667740869929283e-05,
+      "loss": 0.134,
+      "step": 3242
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5665032880910946e-05,
+      "loss": 0.1268,
+      "step": 3243
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5662324279997115e-05,
+      "loss": 0.1071,
+      "step": 3244
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.565961506748036e-05,
+      "loss": 0.1179,
+      "step": 3245
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5656905243653303e-05,
+      "loss": 0.1233,
+      "step": 3246
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.565419480880864e-05,
+      "loss": 0.1304,
+      "step": 3247
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.565148376323914e-05,
+      "loss": 0.129,
+      "step": 3248
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5648772107237622e-05,
+      "loss": 0.1257,
+      "step": 3249
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5646059841096985e-05,
+      "loss": 0.1395,
+      "step": 3250
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5643346965110186e-05,
+      "loss": 0.1343,
+      "step": 3251
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5640633479570252e-05,
+      "loss": 0.1356,
+      "step": 3252
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5637919384770266e-05,
+      "loss": 0.1174,
+      "step": 3253
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5635204681003397e-05,
+      "loss": 0.1166,
+      "step": 3254
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5632489368562853e-05,
+      "loss": 0.1141,
+      "step": 3255
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5629773447741935e-05,
+      "loss": 0.1375,
+      "step": 3256
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5627056918833993e-05,
+      "loss": 0.1231,
+      "step": 3257
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5624339782132445e-05,
+      "loss": 0.123,
+      "step": 3258
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5621622037930776e-05,
+      "loss": 0.1378,
+      "step": 3259
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.561890368652254e-05,
+      "loss": 0.1295,
+      "step": 3260
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5616184728201348e-05,
+      "loss": 0.1136,
+      "step": 3261
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.561346516326089e-05,
+      "loss": 0.1222,
+      "step": 3262
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5610744991994908e-05,
+      "loss": 0.1213,
+      "step": 3263
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5608024214697213e-05,
+      "loss": 0.1249,
+      "step": 3264
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5605302831661688e-05,
+      "loss": 0.1227,
+      "step": 3265
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5602580843182272e-05,
+      "loss": 0.1195,
+      "step": 3266
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5599858249552984e-05,
+      "loss": 0.1272,
+      "step": 3267
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.559713505106789e-05,
+      "loss": 0.1379,
+      "step": 3268
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5594411248021135e-05,
+      "loss": 0.1408,
+      "step": 3269
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.559168684070692e-05,
+      "loss": 0.1229,
+      "step": 3270
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5588961829419513e-05,
+      "loss": 0.1365,
+      "step": 3271
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.558623621445326e-05,
+      "loss": 0.1311,
+      "step": 3272
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5583509996102552e-05,
+      "loss": 0.1341,
+      "step": 3273
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.558078317466186e-05,
+      "loss": 0.1228,
+      "step": 3274
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.557805575042572e-05,
+      "loss": 0.1375,
+      "step": 3275
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5575327723688718e-05,
+      "loss": 0.1268,
+      "step": 3276
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5572599094745522e-05,
+      "loss": 0.1342,
+      "step": 3277
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5569869863890858e-05,
+      "loss": 0.1353,
+      "step": 3278
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5567140031419516e-05,
+      "loss": 0.1263,
+      "step": 3279
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5564409597626353e-05,
+      "loss": 0.1226,
+      "step": 3280
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5561678562806292e-05,
+      "loss": 0.1221,
+      "step": 3281
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.555894692725432e-05,
+      "loss": 0.1259,
+      "step": 3282
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5556214691265482e-05,
+      "loss": 0.1327,
+      "step": 3283
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5553481855134903e-05,
+      "loss": 0.1414,
+      "step": 3284
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5550748419157762e-05,
+      "loss": 0.1362,
+      "step": 3285
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5548014383629298e-05,
+      "loss": 0.0768,
+      "step": 3286
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5545279748844828e-05,
+      "loss": 0.0684,
+      "step": 3287
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.554254451509973e-05,
+      "loss": 0.0743,
+      "step": 3288
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5539808682689435e-05,
+      "loss": 0.0747,
+      "step": 3289
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5537072251909453e-05,
+      "loss": 0.0755,
+      "step": 3290
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.553433522305535e-05,
+      "loss": 0.0754,
+      "step": 3291
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5531597596422767e-05,
+      "loss": 0.0686,
+      "step": 3292
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5528859372307393e-05,
+      "loss": 0.0569,
+      "step": 3293
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5526120551005e-05,
+      "loss": 0.0701,
+      "step": 3294
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.552338113281141e-05,
+      "loss": 0.0732,
+      "step": 3295
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5520641118022515e-05,
+      "loss": 0.0742,
+      "step": 3296
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5517900506934277e-05,
+      "loss": 0.0679,
+      "step": 3297
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.551515929984271e-05,
+      "loss": 0.0645,
+      "step": 3298
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5512417497043898e-05,
+      "loss": 0.0686,
+      "step": 3299
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5509675098834e-05,
+      "loss": 0.077,
+      "step": 3300
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.550693210550922e-05,
+      "loss": 0.075,
+      "step": 3301
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.550418851736584e-05,
+      "loss": 0.0853,
+      "step": 3302
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5501444334700204e-05,
+      "loss": 0.0797,
+      "step": 3303
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5498699557808718e-05,
+      "loss": 0.0667,
+      "step": 3304
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.549595418698785e-05,
+      "loss": 0.0621,
+      "step": 3305
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5493208222534134e-05,
+      "loss": 0.069,
+      "step": 3306
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5490461664744176e-05,
+      "loss": 0.0688,
+      "step": 3307
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5487714513914634e-05,
+      "loss": 0.0739,
+      "step": 3308
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5484966770342237e-05,
+      "loss": 0.0721,
+      "step": 3309
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5482218434323772e-05,
+      "loss": 0.0755,
+      "step": 3310
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.54794695061561e-05,
+      "loss": 0.0848,
+      "step": 3311
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5476719986136136e-05,
+      "loss": 0.0676,
+      "step": 3312
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5473969874560867e-05,
+      "loss": 0.0649,
+      "step": 3313
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5471219171727335e-05,
+      "loss": 0.0722,
+      "step": 3314
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5468467877932654e-05,
+      "loss": 0.0799,
+      "step": 3315
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5465715993474e-05,
+      "loss": 0.07,
+      "step": 3316
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5462963518648607e-05,
+      "loss": 0.0723,
+      "step": 3317
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5460210453753783e-05,
+      "loss": 0.0664,
+      "step": 3318
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5457456799086886e-05,
+      "loss": 0.0829,
+      "step": 3319
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5454702554945353e-05,
+      "loss": 0.0655,
+      "step": 3320
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5451947721626676e-05,
+      "loss": 0.0727,
+      "step": 3321
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5449192299428413e-05,
+      "loss": 0.066,
+      "step": 3322
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5446436288648178e-05,
+      "loss": 0.0834,
+      "step": 3323
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.544367968958366e-05,
+      "loss": 0.0641,
+      "step": 3324
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.544092250253261e-05,
+      "loss": 0.0676,
+      "step": 3325
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.543816472779283e-05,
+      "loss": 0.0712,
+      "step": 3326
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5435406365662203e-05,
+      "loss": 0.0754,
+      "step": 3327
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5432647416438667e-05,
+      "loss": 0.0724,
+      "step": 3328
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.542988788042022e-05,
+      "loss": 0.0807,
+      "step": 3329
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5427127757904922e-05,
+      "loss": 0.0646,
+      "step": 3330
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.542436704919091e-05,
+      "loss": 0.072,
+      "step": 3331
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5421605754576376e-05,
+      "loss": 0.085,
+      "step": 3332
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5418843874359568e-05,
+      "loss": 0.0687,
+      "step": 3333
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5416081408838807e-05,
+      "loss": 0.0901,
+      "step": 3334
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.541331835831248e-05,
+      "loss": 0.0913,
+      "step": 3335
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5410554723079022e-05,
+      "loss": 0.0787,
+      "step": 3336
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5407790503436947e-05,
+      "loss": 0.0762,
+      "step": 3337
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5405025699684824e-05,
+      "loss": 0.0747,
+      "step": 3338
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5402260312121287e-05,
+      "loss": 0.0769,
+      "step": 3339
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.539949434104503e-05,
+      "loss": 0.0673,
+      "step": 3340
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.539672778675482e-05,
+      "loss": 0.0765,
+      "step": 3341
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5393960649549475e-05,
+      "loss": 0.0711,
+      "step": 3342
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5391192929727884e-05,
+      "loss": 0.0736,
+      "step": 3343
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.538842462758899e-05,
+      "loss": 0.0761,
+      "step": 3344
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.538565574343181e-05,
+      "loss": 0.0688,
+      "step": 3345
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5382886277555417e-05,
+      "loss": 0.0773,
+      "step": 3346
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.538011623025895e-05,
+      "loss": 0.0807,
+      "step": 3347
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5377345601841608e-05,
+      "loss": 0.0764,
+      "step": 3348
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5374574392602653e-05,
+      "loss": 0.0636,
+      "step": 3349
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5371802602841412e-05,
+      "loss": 0.0687,
+      "step": 3350
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5369030232857275e-05,
+      "loss": 0.0657,
+      "step": 3351
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5366257282949687e-05,
+      "loss": 0.0844,
+      "step": 3352
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.536348375341817e-05,
+      "loss": 0.0705,
+      "step": 3353
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5360709644562296e-05,
+      "loss": 0.0691,
+      "step": 3354
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.535793495668171e-05,
+      "loss": 0.0767,
+      "step": 3355
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5355159690076102e-05,
+      "loss": 0.0712,
+      "step": 3356
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5352383845045245e-05,
+      "loss": 0.0855,
+      "step": 3357
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5349607421888963e-05,
+      "loss": 0.072,
+      "step": 3358
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.534683042090714e-05,
+      "loss": 0.0794,
+      "step": 3359
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5344052842399734e-05,
+      "loss": 0.073,
+      "step": 3360
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.534127468666676e-05,
+      "loss": 0.0776,
+      "step": 3361
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.5338495954008287e-05,
+      "loss": 0.0741,
+      "step": 3362
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.5335716644724464e-05,
+      "loss": 0.0826,
+      "step": 3363
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.533293675911548e-05,
+      "loss": 0.0727,
+      "step": 3364
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.53301562974816e-05,
+      "loss": 0.0591,
+      "step": 3365
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.532737526012316e-05,
+      "loss": 0.0674,
+      "step": 3366
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.5324593647340535e-05,
+      "loss": 0.0719,
+      "step": 3367
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.532181145943418e-05,
+      "loss": 0.0657,
+      "step": 3368
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5319028696704606e-05,
+      "loss": 0.0797,
+      "step": 3369
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5316245359452387e-05,
+      "loss": 0.0739,
+      "step": 3370
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5313461447978157e-05,
+      "loss": 0.0843,
+      "step": 3371
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5310676962582617e-05,
+      "loss": 0.0899,
+      "step": 3372
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5307891903566523e-05,
+      "loss": 0.0849,
+      "step": 3373
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.5305106271230697e-05,
+      "loss": 0.0673,
+      "step": 3374
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.5302320065876025e-05,
+      "loss": 0.0762,
+      "step": 3375
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.529953328780345e-05,
+      "loss": 0.0854,
+      "step": 3376
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.529674593731399e-05,
+      "loss": 0.0691,
+      "step": 3377
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.5293958014708695e-05,
+      "loss": 0.0763,
+      "step": 3378
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.5291169520288713e-05,
+      "loss": 0.0609,
+      "step": 3379
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.528838045435523e-05,
+      "loss": 0.0769,
+      "step": 3380
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.5285590817209494e-05,
+      "loss": 0.0675,
+      "step": 3381
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.5282800609152833e-05,
+      "loss": 0.0889,
+      "step": 3382
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.528000983048662e-05,
+      "loss": 0.0688,
+      "step": 3383
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.5277218481512294e-05,
+      "loss": 0.0854,
+      "step": 3384
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.5274426562531357e-05,
+      "loss": 0.0716,
+      "step": 3385
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.527163407384537e-05,
+      "loss": 0.0692,
+      "step": 3386
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.526884101575596e-05,
+      "loss": 0.0766,
+      "step": 3387
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5266047388564812e-05,
+      "loss": 0.0804,
+      "step": 3388
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5263253192573674e-05,
+      "loss": 0.0868,
+      "step": 3389
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.526045842808435e-05,
+      "loss": 0.0702,
+      "step": 3390
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5257663095398712e-05,
+      "loss": 0.088,
+      "step": 3391
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5254867194818694e-05,
+      "loss": 0.0814,
+      "step": 3392
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5252070726646286e-05,
+      "loss": 0.0915,
+      "step": 3393
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5249273691183548e-05,
+      "loss": 0.0605,
+      "step": 3394
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.524647608873259e-05,
+      "loss": 0.0762,
+      "step": 3395
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5243677919595585e-05,
+      "loss": 0.0733,
+      "step": 3396
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5240879184074777e-05,
+      "loss": 0.0603,
+      "step": 3397
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5238079882472465e-05,
+      "loss": 0.077,
+      "step": 3398
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5235280015091007e-05,
+      "loss": 0.0855,
+      "step": 3399
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5232479582232822e-05,
+      "loss": 0.0837,
+      "step": 3400
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.52296785842004e-05,
+      "loss": 0.0891,
+      "step": 3401
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5226877021296274e-05,
+      "loss": 0.0862,
+      "step": 3402
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5224074893823055e-05,
+      "loss": 0.0802,
+      "step": 3403
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5221272202083407e-05,
+      "loss": 0.0716,
+      "step": 3404
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.521846894638006e-05,
+      "loss": 0.0792,
+      "step": 3405
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5215665127015792e-05,
+      "loss": 0.0856,
+      "step": 3406
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5212860744293458e-05,
+      "loss": 0.0743,
+      "step": 3407
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5210055798515968e-05,
+      "loss": 0.0748,
+      "step": 3408
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.520725028998629e-05,
+      "loss": 0.0679,
+      "step": 3409
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.520444421900745e-05,
+      "loss": 0.0822,
+      "step": 3410
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5201637585882545e-05,
+      "loss": 0.0762,
+      "step": 3411
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5198830390914726e-05,
+      "loss": 0.0911,
+      "step": 3412
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5196022634407201e-05,
+      "loss": 0.0657,
+      "step": 3413
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5193214316663253e-05,
+      "loss": 0.0657,
+      "step": 3414
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5190405437986203e-05,
+      "loss": 0.0868,
+      "step": 3415
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5187595998679458e-05,
+      "loss": 0.0678,
+      "step": 3416
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.518478599904647e-05,
+      "loss": 0.0737,
+      "step": 3417
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5181975439390746e-05,
+      "loss": 0.0583,
+      "step": 3418
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.5179164320015871e-05,
+      "loss": 0.082,
+      "step": 3419
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5176352641225482e-05,
+      "loss": 0.0815,
+      "step": 3420
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.517354040332327e-05,
+      "loss": 0.0745,
+      "step": 3421
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5170727606612997e-05,
+      "loss": 0.0732,
+      "step": 3422
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5167914251398482e-05,
+      "loss": 0.0933,
+      "step": 3423
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5165100337983596e-05,
+      "loss": 0.0832,
+      "step": 3424
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5162285866672283e-05,
+      "loss": 0.0711,
+      "step": 3425
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.5159470837768544e-05,
+      "loss": 0.0855,
+      "step": 3426
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5156655251576431e-05,
+      "loss": 0.0954,
+      "step": 3427
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5153839108400065e-05,
+      "loss": 0.0641,
+      "step": 3428
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.515102240854363e-05,
+      "loss": 0.0758,
+      "step": 3429
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5148205152311364e-05,
+      "loss": 0.0805,
+      "step": 3430
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5145387340007562e-05,
+      "loss": 0.0691,
+      "step": 3431
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.5142568971936584e-05,
+      "loss": 0.0799,
+      "step": 3432
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5139750048402854e-05,
+      "loss": 0.0962,
+      "step": 3433
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5136930569710848e-05,
+      "loss": 0.0733,
+      "step": 3434
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5134110536165105e-05,
+      "loss": 0.0768,
+      "step": 3435
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5131289948070227e-05,
+      "loss": 0.0838,
+      "step": 3436
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5128468805730875e-05,
+      "loss": 0.078,
+      "step": 3437
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5125647109451764e-05,
+      "loss": 0.0647,
+      "step": 3438
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5122824859537675e-05,
+      "loss": 0.0871,
+      "step": 3439
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5120002056293445e-05,
+      "loss": 0.0685,
+      "step": 3440
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5117178700023975e-05,
+      "loss": 0.0781,
+      "step": 3441
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5114354791034225e-05,
+      "loss": 0.0744,
+      "step": 3442
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5111530329629207e-05,
+      "loss": 0.0797,
+      "step": 3443
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5108705316114e-05,
+      "loss": 0.076,
+      "step": 3444
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5105879750793748e-05,
+      "loss": 0.0948,
+      "step": 3445
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5103053633973642e-05,
+      "loss": 0.0828,
+      "step": 3446
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5100226965958942e-05,
+      "loss": 0.0684,
+      "step": 3447
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5097399747054958e-05,
+      "loss": 0.0783,
+      "step": 3448
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5094571977567069e-05,
+      "loss": 0.0912,
+      "step": 3449
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5091743657800712e-05,
+      "loss": 0.0877,
+      "step": 3450
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5088914788061379e-05,
+      "loss": 0.0791,
+      "step": 3451
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5086085368654627e-05,
+      "loss": 0.0846,
+      "step": 3452
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5083255399886069e-05,
+      "loss": 0.0714,
+      "step": 3453
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.508042488206137e-05,
+      "loss": 0.0766,
+      "step": 3454
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5077593815486271e-05,
+      "loss": 0.0646,
+      "step": 3455
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5074762200466557e-05,
+      "loss": 0.0713,
+      "step": 3456
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5071930037308082e-05,
+      "loss": 0.0807,
+      "step": 3457
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5069097326316757e-05,
+      "loss": 0.0729,
+      "step": 3458
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.5066264067798547e-05,
+      "loss": 0.0762,
+      "step": 3459
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5063430262059479e-05,
+      "loss": 0.0684,
+      "step": 3460
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5060595909405647e-05,
+      "loss": 0.0864,
+      "step": 3461
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5057761010143187e-05,
+      "loss": 0.0851,
+      "step": 3462
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.505492556457831e-05,
+      "loss": 0.0848,
+      "step": 3463
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5052089573017283e-05,
+      "loss": 0.08,
+      "step": 3464
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.5049253035766422e-05,
+      "loss": 0.0731,
+      "step": 3465
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5046415953132116e-05,
+      "loss": 0.0824,
+      "step": 3466
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.50435783254208e-05,
+      "loss": 0.071,
+      "step": 3467
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5040740152938979e-05,
+      "loss": 0.0884,
+      "step": 3468
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.503790143599321e-05,
+      "loss": 0.0742,
+      "step": 3469
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5035062174890105e-05,
+      "loss": 0.0807,
+      "step": 3470
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5032222369936347e-05,
+      "loss": 0.0786,
+      "step": 3471
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.5029382021438668e-05,
+      "loss": 0.0775,
+      "step": 3472
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5026541129703863e-05,
+      "loss": 0.0829,
+      "step": 3473
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5023699695038784e-05,
+      "loss": 0.0806,
+      "step": 3474
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5020857717750345e-05,
+      "loss": 0.0778,
+      "step": 3475
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.501801519814551e-05,
+      "loss": 0.0668,
+      "step": 3476
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.5015172136531311e-05,
+      "loss": 0.0921,
+      "step": 3477
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.501232853321483e-05,
+      "loss": 0.0779,
+      "step": 3478
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5009484388503217e-05,
+      "loss": 0.0909,
+      "step": 3479
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5006639702703681e-05,
+      "loss": 0.0703,
+      "step": 3480
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.500379447612347e-05,
+      "loss": 0.0763,
+      "step": 3481
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.5000948709069919e-05,
+      "loss": 0.0718,
+      "step": 3482
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4998102401850394e-05,
+      "loss": 0.0882,
+      "step": 3483
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.499525555477234e-05,
+      "loss": 0.071,
+      "step": 3484
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.4992408168143253e-05,
+      "loss": 0.0729,
+      "step": 3485
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4989560242270686e-05,
+      "loss": 0.0849,
+      "step": 3486
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4986711777462249e-05,
+      "loss": 0.087,
+      "step": 3487
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4983862774025612e-05,
+      "loss": 0.0703,
+      "step": 3488
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4981013232268511e-05,
+      "loss": 0.0803,
+      "step": 3489
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4978163152498723e-05,
+      "loss": 0.0792,
+      "step": 3490
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4975312535024099e-05,
+      "loss": 0.0769,
+      "step": 3491
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4972461380152537e-05,
+      "loss": 0.0822,
+      "step": 3492
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4969609688192002e-05,
+      "loss": 0.0793,
+      "step": 3493
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.496675745945051e-05,
+      "loss": 0.0746,
+      "step": 3494
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.496390469423614e-05,
+      "loss": 0.0863,
+      "step": 3495
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.496105139285703e-05,
+      "loss": 0.0896,
+      "step": 3496
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4958197555621367e-05,
+      "loss": 0.0786,
+      "step": 3497
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4955343182837403e-05,
+      "loss": 0.0783,
+      "step": 3498
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4952488274813448e-05,
+      "loss": 0.0955,
+      "step": 3499
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4949632831857868e-05,
+      "loss": 0.0648,
+      "step": 3500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4946776854279084e-05,
+      "loss": 0.0929,
+      "step": 3501
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4943920342385585e-05,
+      "loss": 0.0729,
+      "step": 3502
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4941063296485903e-05,
+      "loss": 0.0769,
+      "step": 3503
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4938205716888641e-05,
+      "loss": 0.0828,
+      "step": 3504
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4935347603902451e-05,
+      "loss": 0.0808,
+      "step": 3505
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4932488957836043e-05,
+      "loss": 0.0808,
+      "step": 3506
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4929629778998193e-05,
+      "loss": 0.0817,
+      "step": 3507
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4926770067697723e-05,
+      "loss": 0.0842,
+      "step": 3508
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4923909824243522e-05,
+      "loss": 0.0858,
+      "step": 3509
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4921049048944529e-05,
+      "loss": 0.0901,
+      "step": 3510
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4918187742109751e-05,
+      "loss": 0.0831,
+      "step": 3511
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4915325904048237e-05,
+      "loss": 0.103,
+      "step": 3512
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4912463535069109e-05,
+      "loss": 0.0691,
+      "step": 3513
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.490960063548153e-05,
+      "loss": 0.077,
+      "step": 3514
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4906737205594741e-05,
+      "loss": 0.0766,
+      "step": 3515
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4903873245718021e-05,
+      "loss": 0.0787,
+      "step": 3516
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4901008756160715e-05,
+      "loss": 0.0942,
+      "step": 3517
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4898143737232227e-05,
+      "loss": 0.0836,
+      "step": 3518
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4895278189242017e-05,
+      "loss": 0.0857,
+      "step": 3519
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4892412112499592e-05,
+      "loss": 0.0686,
+      "step": 3520
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.488954550731453e-05,
+      "loss": 0.0861,
+      "step": 3521
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4886678373996465e-05,
+      "loss": 0.0936,
+      "step": 3522
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4883810712855079e-05,
+      "loss": 0.0708,
+      "step": 3523
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4880942524200116e-05,
+      "loss": 0.08,
+      "step": 3524
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4878073808341375e-05,
+      "loss": 0.0817,
+      "step": 3525
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4875204565588719e-05,
+      "loss": 0.0652,
+      "step": 3526
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4872334796252056e-05,
+      "loss": 0.0781,
+      "step": 3527
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4869464500641363e-05,
+      "loss": 0.0865,
+      "step": 3528
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.486659367906667e-05,
+      "loss": 0.0717,
+      "step": 3529
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4863722331838056e-05,
+      "loss": 0.0819,
+      "step": 3530
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.4860850459265665e-05,
+      "loss": 0.0731,
+      "step": 3531
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4857978061659696e-05,
+      "loss": 0.0856,
+      "step": 3532
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4855105139330408e-05,
+      "loss": 0.0889,
+      "step": 3533
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4852231692588109e-05,
+      "loss": 0.0731,
+      "step": 3534
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4849357721743169e-05,
+      "loss": 0.0746,
+      "step": 3535
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.484648322710601e-05,
+      "loss": 0.0876,
+      "step": 3536
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.484360820898712e-05,
+      "loss": 0.0779,
+      "step": 3537
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4840732667697034e-05,
+      "loss": 0.0826,
+      "step": 3538
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4837856603546346e-05,
+      "loss": 0.0885,
+      "step": 3539
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.483498001684571e-05,
+      "loss": 0.0928,
+      "step": 3540
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4832102907905834e-05,
+      "loss": 0.0716,
+      "step": 3541
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4829225277037477e-05,
+      "loss": 0.0731,
+      "step": 3542
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4826347124551468e-05,
+      "loss": 0.0807,
+      "step": 3543
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4823468450758678e-05,
+      "loss": 0.0769,
+      "step": 3544
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4820589255970043e-05,
+      "loss": 0.0778,
+      "step": 3545
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.481770954049655e-05,
+      "loss": 0.0705,
+      "step": 3546
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4814829304649247e-05,
+      "loss": 0.0798,
+      "step": 3547
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4811948548739239e-05,
+      "loss": 0.0694,
+      "step": 3548
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4809067273077676e-05,
+      "loss": 0.0871,
+      "step": 3549
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4806185477975779e-05,
+      "loss": 0.0924,
+      "step": 3550
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4803303163744817e-05,
+      "loss": 0.0796,
+      "step": 3551
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4800420330696118e-05,
+      "loss": 0.0849,
+      "step": 3552
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4797536979141058e-05,
+      "loss": 0.0885,
+      "step": 3553
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4794653109391084e-05,
+      "loss": 0.065,
+      "step": 3554
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4791768721757686e-05,
+      "loss": 0.0817,
+      "step": 3555
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4788883816552418e-05,
+      "loss": 0.0738,
+      "step": 3556
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.478599839408688e-05,
+      "loss": 0.076,
+      "step": 3557
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.478311245467274e-05,
+      "loss": 0.09,
+      "step": 3558
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.478022599862171e-05,
+      "loss": 0.0819,
+      "step": 3559
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4777339026245572e-05,
+      "loss": 0.0806,
+      "step": 3560
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4774451537856152e-05,
+      "loss": 0.0827,
+      "step": 3561
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4771563533765335e-05,
+      "loss": 0.091,
+      "step": 3562
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4768675014285063e-05,
+      "loss": 0.0885,
+      "step": 3563
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4765785979727329e-05,
+      "loss": 0.0787,
+      "step": 3564
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.476289643040419e-05,
+      "loss": 0.089,
+      "step": 3565
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4760006366627753e-05,
+      "loss": 0.0974,
+      "step": 3566
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4757115788710183e-05,
+      "loss": 0.0887,
+      "step": 3567
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4754224696963696e-05,
+      "loss": 0.0982,
+      "step": 3568
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4751333091700564e-05,
+      "loss": 0.1006,
+      "step": 3569
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4748440973233125e-05,
+      "loss": 0.0771,
+      "step": 3570
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4745548341873759e-05,
+      "loss": 0.0877,
+      "step": 3571
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4742655197934907e-05,
+      "loss": 0.0754,
+      "step": 3572
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.473976154172907e-05,
+      "loss": 0.0779,
+      "step": 3573
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4736867373568795e-05,
+      "loss": 0.0839,
+      "step": 3574
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4733972693766689e-05,
+      "loss": 0.0904,
+      "step": 3575
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4731077502635416e-05,
+      "loss": 0.0836,
+      "step": 3576
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4728181800487693e-05,
+      "loss": 0.0793,
+      "step": 3577
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4725285587636293e-05,
+      "loss": 0.0782,
+      "step": 3578
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4722388864394043e-05,
+      "loss": 0.0781,
+      "step": 3579
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4719491631073826e-05,
+      "loss": 0.0732,
+      "step": 3580
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4716593887988579e-05,
+      "loss": 0.0865,
+      "step": 3581
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4713695635451296e-05,
+      "loss": 0.0865,
+      "step": 3582
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.4710796873775025e-05,
+      "loss": 0.0917,
+      "step": 3583
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4707897603272871e-05,
+      "loss": 0.0857,
+      "step": 3584
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4704997824257992e-05,
+      "loss": 0.0689,
+      "step": 3585
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4702097537043596e-05,
+      "loss": 0.0856,
+      "step": 3586
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4699196741942952e-05,
+      "loss": 0.0997,
+      "step": 3587
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4696295439269387e-05,
+      "loss": 0.0931,
+      "step": 3588
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4693393629336279e-05,
+      "loss": 0.0841,
+      "step": 3589
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4690491312457054e-05,
+      "loss": 0.093,
+      "step": 3590
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4687588488945204e-05,
+      "loss": 0.0945,
+      "step": 3591
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4684685159114271e-05,
+      "loss": 0.0802,
+      "step": 3592
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4681781323277848e-05,
+      "loss": 0.0715,
+      "step": 3593
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.467887698174959e-05,
+      "loss": 0.084,
+      "step": 3594
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.46759721348432e-05,
+      "loss": 0.0943,
+      "step": 3595
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4673066782872436e-05,
+      "loss": 0.0793,
+      "step": 3596
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.467016092615112e-05,
+      "loss": 0.0834,
+      "step": 3597
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4667254564993116e-05,
+      "loss": 0.0772,
+      "step": 3598
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4664347699712352e-05,
+      "loss": 0.0791,
+      "step": 3599
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4661440330622803e-05,
+      "loss": 0.0821,
+      "step": 3600
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4658532458038502e-05,
+      "loss": 0.0909,
+      "step": 3601
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4655624082273536e-05,
+      "loss": 0.0869,
+      "step": 3602
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4652715203642047e-05,
+      "loss": 0.0949,
+      "step": 3603
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4649805822458231e-05,
+      "loss": 0.0691,
+      "step": 3604
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4646895939036343e-05,
+      "loss": 0.0701,
+      "step": 3605
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4643985553690679e-05,
+      "loss": 0.0751,
+      "step": 3606
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4641074666735601e-05,
+      "loss": 0.0782,
+      "step": 3607
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4638163278485523e-05,
+      "loss": 0.0818,
+      "step": 3608
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.463525138925491e-05,
+      "loss": 0.0777,
+      "step": 3609
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4632338999358282e-05,
+      "loss": 0.0904,
+      "step": 3610
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.462942610911022e-05,
+      "loss": 0.0773,
+      "step": 3611
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4626512718825348e-05,
+      "loss": 0.0738,
+      "step": 3612
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4623598828818349e-05,
+      "loss": 0.0883,
+      "step": 3613
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4620684439403962e-05,
+      "loss": 0.086,
+      "step": 3614
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4617769550896979e-05,
+      "loss": 0.0949,
+      "step": 3615
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4614854163612243e-05,
+      "loss": 0.074,
+      "step": 3616
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4611938277864653e-05,
+      "loss": 0.0856,
+      "step": 3617
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.460902189396916e-05,
+      "loss": 0.0944,
+      "step": 3618
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4606105012240772e-05,
+      "loss": 0.0902,
+      "step": 3619
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4603187632994552e-05,
+      "loss": 0.0755,
+      "step": 3620
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4600269756545613e-05,
+      "loss": 0.0988,
+      "step": 3621
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4597351383209121e-05,
+      "loss": 0.0818,
+      "step": 3622
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4594432513300297e-05,
+      "loss": 0.0797,
+      "step": 3623
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4591513147134415e-05,
+      "loss": 0.0835,
+      "step": 3624
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4588593285026806e-05,
+      "loss": 0.089,
+      "step": 3625
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4585672927292853e-05,
+      "loss": 0.0868,
+      "step": 3626
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4582752074247992e-05,
+      "loss": 0.079,
+      "step": 3627
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4579830726207708e-05,
+      "loss": 0.0967,
+      "step": 3628
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.4576908883487549e-05,
+      "loss": 0.0887,
+      "step": 3629
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4573986546403105e-05,
+      "loss": 0.0955,
+      "step": 3630
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.457106371527003e-05,
+      "loss": 0.0803,
+      "step": 3631
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4568140390404026e-05,
+      "loss": 0.083,
+      "step": 3632
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4565216572120853e-05,
+      "loss": 0.0951,
+      "step": 3633
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.456229226073631e-05,
+      "loss": 0.0885,
+      "step": 3634
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4559367456566269e-05,
+      "loss": 0.081,
+      "step": 3635
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.4556442159926648e-05,
+      "loss": 0.091,
+      "step": 3636
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4553516371133404e-05,
+      "loss": 0.0848,
+      "step": 3637
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4550590090502572e-05,
+      "loss": 0.0821,
+      "step": 3638
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.454766331835022e-05,
+      "loss": 0.0734,
+      "step": 3639
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4544736054992479e-05,
+      "loss": 0.092,
+      "step": 3640
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4541808300745528e-05,
+      "loss": 0.0841,
+      "step": 3641
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4538880055925605e-05,
+      "loss": 0.0871,
+      "step": 3642
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.4535951320849003e-05,
+      "loss": 0.0869,
+      "step": 3643
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4533022095832052e-05,
+      "loss": 0.0719,
+      "step": 3644
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.453009238119115e-05,
+      "loss": 0.0742,
+      "step": 3645
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.452716217724274e-05,
+      "loss": 0.0905,
+      "step": 3646
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4524231484303326e-05,
+      "loss": 0.0781,
+      "step": 3647
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4521300302689461e-05,
+      "loss": 0.0908,
+      "step": 3648
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4518368632717747e-05,
+      "loss": 0.0784,
+      "step": 3649
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.451543647470484e-05,
+      "loss": 0.0865,
+      "step": 3650
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4512503828967454e-05,
+      "loss": 0.067,
+      "step": 3651
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.450957069582235e-05,
+      "loss": 0.088,
+      "step": 3652
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.450663707558634e-05,
+      "loss": 0.086,
+      "step": 3653
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4503702968576304e-05,
+      "loss": 0.0859,
+      "step": 3654
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4500768375109146e-05,
+      "loss": 0.0771,
+      "step": 3655
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4497833295501852e-05,
+      "loss": 0.0806,
+      "step": 3656
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4494897730071444e-05,
+      "loss": 0.0865,
+      "step": 3657
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4491961679135003e-05,
+      "loss": 0.0986,
+      "step": 3658
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4489025143009656e-05,
+      "loss": 0.0884,
+      "step": 3659
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4486088122012588e-05,
+      "loss": 0.0874,
+      "step": 3660
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.448315061646103e-05,
+      "loss": 0.0872,
+      "step": 3661
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4480212626672278e-05,
+      "loss": 0.0698,
+      "step": 3662
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4477274152963666e-05,
+      "loss": 0.0837,
+      "step": 3663
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4474335195652587e-05,
+      "loss": 0.0816,
+      "step": 3664
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4471395755056492e-05,
+      "loss": 0.0997,
+      "step": 3665
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.446845583149287e-05,
+      "loss": 0.0814,
+      "step": 3666
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4465515425279275e-05,
+      "loss": 0.092,
+      "step": 3667
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4462574536733305e-05,
+      "loss": 0.0811,
+      "step": 3668
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4459633166172616e-05,
+      "loss": 0.0825,
+      "step": 3669
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.445669131391491e-05,
+      "loss": 0.0797,
+      "step": 3670
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.445374898027795e-05,
+      "loss": 0.0908,
+      "step": 3671
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4450806165579539e-05,
+      "loss": 0.0979,
+      "step": 3672
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4447862870137542e-05,
+      "loss": 0.0866,
+      "step": 3673
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4444919094269874e-05,
+      "loss": 0.0823,
+      "step": 3674
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4441974838294495e-05,
+      "loss": 0.0822,
+      "step": 3675
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4439030102529427e-05,
+      "loss": 0.087,
+      "step": 3676
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4436084887292733e-05,
+      "loss": 0.0903,
+      "step": 3677
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.443313919290254e-05,
+      "loss": 0.0901,
+      "step": 3678
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4430193019677016e-05,
+      "loss": 0.0887,
+      "step": 3679
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.442724636793439e-05,
+      "loss": 0.0741,
+      "step": 3680
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4424299237992933e-05,
+      "loss": 0.0852,
+      "step": 3681
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.4421351630170976e-05,
+      "loss": 0.0871,
+      "step": 3682
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4418403544786894e-05,
+      "loss": 0.0784,
+      "step": 3683
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4415454982159121e-05,
+      "loss": 0.0812,
+      "step": 3684
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4412505942606136e-05,
+      "loss": 0.0834,
+      "step": 3685
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4409556426446475e-05,
+      "loss": 0.0779,
+      "step": 3686
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.4406606433998725e-05,
+      "loss": 0.0943,
+      "step": 3687
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.440365596558152e-05,
+      "loss": 0.0782,
+      "step": 3688
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.440070502151355e-05,
+      "loss": 0.0838,
+      "step": 3689
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4397753602113551e-05,
+      "loss": 0.0792,
+      "step": 3690
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4394801707700316e-05,
+      "loss": 0.0831,
+      "step": 3691
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4391849338592687e-05,
+      "loss": 0.079,
+      "step": 3692
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.438889649510956e-05,
+      "loss": 0.0891,
+      "step": 3693
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4385943177569873e-05,
+      "loss": 0.1003,
+      "step": 3694
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.4382989386292627e-05,
+      "loss": 0.0854,
+      "step": 3695
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4380035121596869e-05,
+      "loss": 0.0775,
+      "step": 3696
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4377080383801696e-05,
+      "loss": 0.0686,
+      "step": 3697
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4374125173226256e-05,
+      "loss": 0.088,
+      "step": 3698
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4371169490189751e-05,
+      "loss": 0.0831,
+      "step": 3699
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4368213335011429e-05,
+      "loss": 0.0808,
+      "step": 3700
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4365256708010598e-05,
+      "loss": 0.0903,
+      "step": 3701
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4362299609506605e-05,
+      "loss": 0.1025,
+      "step": 3702
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4359342039818863e-05,
+      "loss": 0.0752,
+      "step": 3703
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4356383999266818e-05,
+      "loss": 0.0744,
+      "step": 3704
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4353425488169979e-05,
+      "loss": 0.0945,
+      "step": 3705
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4350466506847901e-05,
+      "loss": 0.0734,
+      "step": 3706
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4347507055620199e-05,
+      "loss": 0.0865,
+      "step": 3707
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4344547134806522e-05,
+      "loss": 0.0821,
+      "step": 3708
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4341586744726583e-05,
+      "loss": 0.0832,
+      "step": 3709
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4338625885700142e-05,
+      "loss": 0.104,
+      "step": 3710
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.433566455804701e-05,
+      "loss": 0.083,
+      "step": 3711
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4332702762087047e-05,
+      "loss": 0.0869,
+      "step": 3712
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4329740498140165e-05,
+      "loss": 0.0876,
+      "step": 3713
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4326777766526328e-05,
+      "loss": 0.0905,
+      "step": 3714
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4323814567565544e-05,
+      "loss": 0.0854,
+      "step": 3715
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4320850901577878e-05,
+      "loss": 0.0844,
+      "step": 3716
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4317886768883445e-05,
+      "loss": 0.0851,
+      "step": 3717
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.431492216980241e-05,
+      "loss": 0.0726,
+      "step": 3718
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4311957104654982e-05,
+      "loss": 0.0901,
+      "step": 3719
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.430899157376143e-05,
+      "loss": 0.0936,
+      "step": 3720
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4306025577442069e-05,
+      "loss": 0.0858,
+      "step": 3721
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4303059116017264e-05,
+      "loss": 0.0907,
+      "step": 3722
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4300092189807429e-05,
+      "loss": 0.0928,
+      "step": 3723
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4297124799133029e-05,
+      "loss": 0.0836,
+      "step": 3724
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4294156944314585e-05,
+      "loss": 0.0842,
+      "step": 3725
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.429118862567266e-05,
+      "loss": 0.0831,
+      "step": 3726
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4288219843527866e-05,
+      "loss": 0.0882,
+      "step": 3727
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4285250598200875e-05,
+      "loss": 0.0789,
+      "step": 3728
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.42822808900124e-05,
+      "loss": 0.0933,
+      "step": 3729
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4279310719283207e-05,
+      "loss": 0.093,
+      "step": 3730
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4276340086334118e-05,
+      "loss": 0.0767,
+      "step": 3731
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4273368991485989e-05,
+      "loss": 0.0793,
+      "step": 3732
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4270397435059747e-05,
+      "loss": 0.0793,
+      "step": 3733
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4267425417376349e-05,
+      "loss": 0.0823,
+      "step": 3734
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4264452938756813e-05,
+      "loss": 0.0913,
+      "step": 3735
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4261479999522205e-05,
+      "loss": 0.075,
+      "step": 3736
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4258506599993642e-05,
+      "loss": 0.0839,
+      "step": 3737
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4255532740492286e-05,
+      "loss": 0.0837,
+      "step": 3738
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4252558421339355e-05,
+      "loss": 0.0806,
+      "step": 3739
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4249583642856113e-05,
+      "loss": 0.0778,
+      "step": 3740
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4246608405363868e-05,
+      "loss": 0.0957,
+      "step": 3741
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.424363270918399e-05,
+      "loss": 0.0831,
+      "step": 3742
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4240656554637889e-05,
+      "loss": 0.0842,
+      "step": 3743
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4237679942047027e-05,
+      "loss": 0.1,
+      "step": 3744
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4234702871732917e-05,
+      "loss": 0.0922,
+      "step": 3745
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.423172534401712e-05,
+      "loss": 0.0883,
+      "step": 3746
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.422874735922125e-05,
+      "loss": 0.0799,
+      "step": 3747
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4225768917666965e-05,
+      "loss": 0.0793,
+      "step": 3748
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4222790019675969e-05,
+      "loss": 0.0872,
+      "step": 3749
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.421981066557003e-05,
+      "loss": 0.0811,
+      "step": 3750
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4216830855670948e-05,
+      "loss": 0.0767,
+      "step": 3751
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.421385059030059e-05,
+      "loss": 0.0922,
+      "step": 3752
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.421086986978085e-05,
+      "loss": 0.0891,
+      "step": 3753
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.420788869443369e-05,
+      "loss": 0.0913,
+      "step": 3754
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4204907064581122e-05,
+      "loss": 0.0851,
+      "step": 3755
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4201924980545187e-05,
+      "loss": 0.0964,
+      "step": 3756
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4198942442647996e-05,
+      "loss": 0.0841,
+      "step": 3757
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.41959594512117e-05,
+      "loss": 0.0813,
+      "step": 3758
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4192976006558497e-05,
+      "loss": 0.0871,
+      "step": 3759
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4189992109010638e-05,
+      "loss": 0.0852,
+      "step": 3760
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4187007758890423e-05,
+      "loss": 0.0807,
+      "step": 3761
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4184022956520201e-05,
+      "loss": 0.0964,
+      "step": 3762
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4181037702222366e-05,
+      "loss": 0.088,
+      "step": 3763
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4178051996319363e-05,
+      "loss": 0.0905,
+      "step": 3764
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4175065839133688e-05,
+      "loss": 0.087,
+      "step": 3765
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4172079230987882e-05,
+      "loss": 0.0861,
+      "step": 3766
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4169092172204539e-05,
+      "loss": 0.0982,
+      "step": 3767
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.41661046631063e-05,
+      "loss": 0.0927,
+      "step": 3768
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.416311670401585e-05,
+      "loss": 0.0858,
+      "step": 3769
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4160128295255931e-05,
+      "loss": 0.0837,
+      "step": 3770
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4157139437149324e-05,
+      "loss": 0.0785,
+      "step": 3771
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.0754,
+      "step": 3772
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.415116037418744e-05,
+      "loss": 0.0711,
+      "step": 3773
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4148170169977982e-05,
+      "loss": 0.0801,
+      "step": 3774
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4145179517713463e-05,
+      "loss": 0.0999,
+      "step": 3775
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.414218841771692e-05,
+      "loss": 0.0881,
+      "step": 3776
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4139196870311424e-05,
+      "loss": 0.0877,
+      "step": 3777
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4136204875820104e-05,
+      "loss": 0.0835,
+      "step": 3778
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.413321243456613e-05,
+      "loss": 0.0885,
+      "step": 3779
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4130219546872727e-05,
+      "loss": 0.0809,
+      "step": 3780
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4127226213063163e-05,
+      "loss": 0.083,
+      "step": 3781
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4124232433460756e-05,
+      "loss": 0.0807,
+      "step": 3782
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4121238208388872e-05,
+      "loss": 0.0871,
+      "step": 3783
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4118243538170929e-05,
+      "loss": 0.0905,
+      "step": 3784
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4115248423130385e-05,
+      "loss": 0.0874,
+      "step": 3785
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.411225286359075e-05,
+      "loss": 0.0893,
+      "step": 3786
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4109256859875585e-05,
+      "loss": 0.073,
+      "step": 3787
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4106260412308494e-05,
+      "loss": 0.1028,
+      "step": 3788
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4103263521213135e-05,
+      "loss": 0.0739,
+      "step": 3789
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4100266186913207e-05,
+      "loss": 0.0884,
+      "step": 3790
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4097268409732463e-05,
+      "loss": 0.0915,
+      "step": 3791
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.40942701899947e-05,
+      "loss": 0.0925,
+      "step": 3792
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4091271528023764e-05,
+      "loss": 0.0809,
+      "step": 3793
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4088272424143546e-05,
+      "loss": 0.0933,
+      "step": 3794
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4085272878677989e-05,
+      "loss": 0.0782,
+      "step": 3795
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4082272891951085e-05,
+      "loss": 0.0873,
+      "step": 3796
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4079272464286864e-05,
+      "loss": 0.0828,
+      "step": 3797
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4076271596009416e-05,
+      "loss": 0.0789,
+      "step": 3798
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4073270287442873e-05,
+      "loss": 0.0856,
+      "step": 3799
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.407026853891141e-05,
+      "loss": 0.0891,
+      "step": 3800
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4067266350739259e-05,
+      "loss": 0.0789,
+      "step": 3801
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4064263723250689e-05,
+      "loss": 0.091,
+      "step": 3802
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4061260656770024e-05,
+      "loss": 0.0874,
+      "step": 3803
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4058257151621635e-05,
+      "loss": 0.0773,
+      "step": 3804
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.405525320812994e-05,
+      "loss": 0.1088,
+      "step": 3805
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4052248826619401e-05,
+      "loss": 0.1013,
+      "step": 3806
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4049244007414528e-05,
+      "loss": 0.0931,
+      "step": 3807
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.404623875083988e-05,
+      "loss": 0.0859,
+      "step": 3808
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4043233057220066e-05,
+      "loss": 0.0994,
+      "step": 3809
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4040226926879734e-05,
+      "loss": 0.0848,
+      "step": 3810
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.403722036014359e-05,
+      "loss": 0.0917,
+      "step": 3811
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4034213357336374e-05,
+      "loss": 0.089,
+      "step": 3812
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4031205918782888e-05,
+      "loss": 0.096,
+      "step": 3813
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4028198044807972e-05,
+      "loss": 0.0794,
+      "step": 3814
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.402518973573651e-05,
+      "loss": 0.087,
+      "step": 3815
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4022180991893443e-05,
+      "loss": 0.1044,
+      "step": 3816
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4019171813603749e-05,
+      "loss": 0.0934,
+      "step": 3817
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4016162201192462e-05,
+      "loss": 0.0918,
+      "step": 3818
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4013152154984654e-05,
+      "loss": 0.0854,
+      "step": 3819
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.401014167530545e-05,
+      "loss": 0.0823,
+      "step": 3820
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4007130762480023e-05,
+      "loss": 0.0828,
+      "step": 3821
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4004119416833586e-05,
+      "loss": 0.089,
+      "step": 3822
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4001107638691407e-05,
+      "loss": 0.1016,
+      "step": 3823
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3998095428378787e-05,
+      "loss": 0.0927,
+      "step": 3824
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3995082786221091e-05,
+      "loss": 0.0866,
+      "step": 3825
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.399206971254372e-05,
+      "loss": 0.0846,
+      "step": 3826
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.3989056207672125e-05,
+      "loss": 0.0831,
+      "step": 3827
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3986042271931806e-05,
+      "loss": 0.0847,
+      "step": 3828
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3983027905648298e-05,
+      "loss": 0.0873,
+      "step": 3829
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3980013109147199e-05,
+      "loss": 0.0837,
+      "step": 3830
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.397699788275414e-05,
+      "loss": 0.0857,
+      "step": 3831
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3973982226794806e-05,
+      "loss": 0.0819,
+      "step": 3832
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.3970966141594926e-05,
+      "loss": 0.0925,
+      "step": 3833
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3967949627480277e-05,
+      "loss": 0.0849,
+      "step": 3834
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3964932684776678e-05,
+      "loss": 0.0801,
+      "step": 3835
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.396191531381e-05,
+      "loss": 0.0813,
+      "step": 3836
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3958897514906153e-05,
+      "loss": 0.0804,
+      "step": 3837
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.39558792883911e-05,
+      "loss": 0.0852,
+      "step": 3838
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.395286063459085e-05,
+      "loss": 0.0969,
+      "step": 3839
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.3949841553831456e-05,
+      "loss": 0.0892,
+      "step": 3840
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3946822046439012e-05,
+      "loss": 0.0976,
+      "step": 3841
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3943802112739666e-05,
+      "loss": 0.0918,
+      "step": 3842
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3940781753059614e-05,
+      "loss": 0.1037,
+      "step": 3843
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3937760967725084e-05,
+      "loss": 0.0988,
+      "step": 3844
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3934739757062367e-05,
+      "loss": 0.0841,
+      "step": 3845
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.3931718121397785e-05,
+      "loss": 0.0808,
+      "step": 3846
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3928696061057721e-05,
+      "loss": 0.0852,
+      "step": 3847
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.392567357636859e-05,
+      "loss": 0.0861,
+      "step": 3848
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3922650667656861e-05,
+      "loss": 0.0868,
+      "step": 3849
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3919627335249046e-05,
+      "loss": 0.0757,
+      "step": 3850
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3916603579471705e-05,
+      "loss": 0.09,
+      "step": 3851
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3913579400651439e-05,
+      "loss": 0.0864,
+      "step": 3852
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.3910554799114899e-05,
+      "loss": 0.0829,
+      "step": 3853
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.390752977518878e-05,
+      "loss": 0.0862,
+      "step": 3854
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3904504329199824e-05,
+      "loss": 0.0894,
+      "step": 3855
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3901478461474817e-05,
+      "loss": 0.0788,
+      "step": 3856
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3898452172340593e-05,
+      "loss": 0.0818,
+      "step": 3857
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3895425462124028e-05,
+      "loss": 0.0759,
+      "step": 3858
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.3892398331152044e-05,
+      "loss": 0.0995,
+      "step": 3859
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.388937077975161e-05,
+      "loss": 0.0917,
+      "step": 3860
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3886342808249742e-05,
+      "loss": 0.0863,
+      "step": 3861
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3883314416973496e-05,
+      "loss": 0.0766,
+      "step": 3862
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3880285606249978e-05,
+      "loss": 0.0971,
+      "step": 3863
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3877256376406342e-05,
+      "loss": 0.0994,
+      "step": 3864
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.387422672776978e-05,
+      "loss": 0.0963,
+      "step": 3865
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.3871196660667533e-05,
+      "loss": 0.0851,
+      "step": 3866
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3868166175426885e-05,
+      "loss": 0.0827,
+      "step": 3867
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3865135272375168e-05,
+      "loss": 0.0762,
+      "step": 3868
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3862103951839756e-05,
+      "loss": 0.0741,
+      "step": 3869
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3859072214148075e-05,
+      "loss": 0.0923,
+      "step": 3870
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3856040059627588e-05,
+      "loss": 0.1007,
+      "step": 3871
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3853007488605808e-05,
+      "loss": 0.0717,
+      "step": 3872
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.3849974501410291e-05,
+      "loss": 0.0827,
+      "step": 3873
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3846941098368633e-05,
+      "loss": 0.0834,
+      "step": 3874
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3843907279808487e-05,
+      "loss": 0.0868,
+      "step": 3875
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3840873046057536e-05,
+      "loss": 0.0812,
+      "step": 3876
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3837838397443526e-05,
+      "loss": 0.0798,
+      "step": 3877
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.3834803334294233e-05,
+      "loss": 0.0904,
+      "step": 3878
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.383176785693748e-05,
+      "loss": 0.0907,
+      "step": 3879
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3828731965701138e-05,
+      "loss": 0.0757,
+      "step": 3880
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3825695660913124e-05,
+      "loss": 0.0783,
+      "step": 3881
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3822658942901394e-05,
+      "loss": 0.0907,
+      "step": 3882
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3819621811993957e-05,
+      "loss": 0.0957,
+      "step": 3883
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3816584268518859e-05,
+      "loss": 0.0854,
+      "step": 3884
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3813546312804192e-05,
+      "loss": 0.0796,
+      "step": 3885
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.3810507945178093e-05,
+      "loss": 0.0825,
+      "step": 3886
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.380746916596875e-05,
+      "loss": 0.0918,
+      "step": 3887
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3804429975504383e-05,
+      "loss": 0.0962,
+      "step": 3888
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3801390374113268e-05,
+      "loss": 0.0935,
+      "step": 3889
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3798350362123719e-05,
+      "loss": 0.0803,
+      "step": 3890
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3795309939864095e-05,
+      "loss": 0.0782,
+      "step": 3891
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.3792269107662799e-05,
+      "loss": 0.0953,
+      "step": 3892
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3789227865848282e-05,
+      "loss": 0.0795,
+      "step": 3893
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.378618621474904e-05,
+      "loss": 0.084,
+      "step": 3894
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3783144154693604e-05,
+      "loss": 0.0869,
+      "step": 3895
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3780101686010556e-05,
+      "loss": 0.0924,
+      "step": 3896
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3777058809028521e-05,
+      "loss": 0.0956,
+      "step": 3897
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3774015524076173e-05,
+      "loss": 0.0809,
+      "step": 3898
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.3770971831482218e-05,
+      "loss": 0.0962,
+      "step": 3899
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.376792773157542e-05,
+      "loss": 0.076,
+      "step": 3900
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3764883224684579e-05,
+      "loss": 0.0842,
+      "step": 3901
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.376183831113854e-05,
+      "loss": 0.0881,
+      "step": 3902
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3758792991266193e-05,
+      "loss": 0.095,
+      "step": 3903
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3755747265396466e-05,
+      "loss": 0.0836,
+      "step": 3904
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.3752701133858342e-05,
+      "loss": 0.0686,
+      "step": 3905
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3749654596980841e-05,
+      "loss": 0.1071,
+      "step": 3906
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3746607655093026e-05,
+      "loss": 0.0917,
+      "step": 3907
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3743560308524005e-05,
+      "loss": 0.0891,
+      "step": 3908
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3740512557602932e-05,
+      "loss": 0.0889,
+      "step": 3909
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3737464402659004e-05,
+      "loss": 0.0864,
+      "step": 3910
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3734415844021457e-05,
+      "loss": 0.091,
+      "step": 3911
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.3731366882019573e-05,
+      "loss": 0.081,
+      "step": 3912
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3728317516982682e-05,
+      "loss": 0.0957,
+      "step": 3913
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3725267749240155e-05,
+      "loss": 0.0759,
+      "step": 3914
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.37222175791214e-05,
+      "loss": 0.0804,
+      "step": 3915
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3719167006955884e-05,
+      "loss": 0.0829,
+      "step": 3916
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3716116033073099e-05,
+      "loss": 0.0917,
+      "step": 3917
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3713064657802591e-05,
+      "loss": 0.1053,
+      "step": 3918
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.3710012881473946e-05,
+      "loss": 0.0882,
+      "step": 3919
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3706960704416799e-05,
+      "loss": 0.0945,
+      "step": 3920
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3703908126960823e-05,
+      "loss": 0.0902,
+      "step": 3921
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.370085514943573e-05,
+      "loss": 0.0934,
+      "step": 3922
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3697801772171285e-05,
+      "loss": 0.0743,
+      "step": 3923
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.369474799549729e-05,
+      "loss": 0.0932,
+      "step": 3924
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.3691693819743593e-05,
+      "loss": 0.085,
+      "step": 3925
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3688639245240078e-05,
+      "loss": 0.0921,
+      "step": 3926
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3685584272316689e-05,
+      "loss": 0.1041,
+      "step": 3927
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3682528901303392e-05,
+      "loss": 0.0817,
+      "step": 3928
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.367947313253021e-05,
+      "loss": 0.0911,
+      "step": 3929
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3676416966327201e-05,
+      "loss": 0.077,
+      "step": 3930
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.3673360403024477e-05,
+      "loss": 0.0901,
+      "step": 3931
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.367030344295218e-05,
+      "loss": 0.0912,
+      "step": 3932
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3667246086440504e-05,
+      "loss": 0.0902,
+      "step": 3933
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3664188333819677e-05,
+      "loss": 0.0937,
+      "step": 3934
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.366113018541998e-05,
+      "loss": 0.0824,
+      "step": 3935
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3658071641571729e-05,
+      "loss": 0.0885,
+      "step": 3936
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3655012702605288e-05,
+      "loss": 0.0896,
+      "step": 3937
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.3651953368851062e-05,
+      "loss": 0.0784,
+      "step": 3938
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3648893640639492e-05,
+      "loss": 0.0803,
+      "step": 3939
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3645833518301074e-05,
+      "loss": 0.0934,
+      "step": 3940
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3642773002166338e-05,
+      "loss": 0.0919,
+      "step": 3941
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3639712092565855e-05,
+      "loss": 0.0873,
+      "step": 3942
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.363665078983025e-05,
+      "loss": 0.0488,
+      "step": 3943
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3633589094290174e-05,
+      "loss": 0.0459,
+      "step": 3944
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.3630527006276334e-05,
+      "loss": 0.0561,
+      "step": 3945
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3627464526119476e-05,
+      "loss": 0.0476,
+      "step": 3946
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3624401654150381e-05,
+      "loss": 0.0558,
+      "step": 3947
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3621338390699881e-05,
+      "loss": 0.0469,
+      "step": 3948
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3618274736098852e-05,
+      "loss": 0.0512,
+      "step": 3949
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3615210690678197e-05,
+      "loss": 0.057,
+      "step": 3950
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.3612146254768878e-05,
+      "loss": 0.056,
+      "step": 3951
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3609081428701895e-05,
+      "loss": 0.0559,
+      "step": 3952
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3606016212808289e-05,
+      "loss": 0.0651,
+      "step": 3953
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3602950607419134e-05,
+      "loss": 0.0614,
+      "step": 3954
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3599884612865564e-05,
+      "loss": 0.0606,
+      "step": 3955
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3596818229478736e-05,
+      "loss": 0.0499,
+      "step": 3956
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.3593751457589868e-05,
+      "loss": 0.0525,
+      "step": 3957
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.35906842975302e-05,
+      "loss": 0.0536,
+      "step": 3958
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3587616749631037e-05,
+      "loss": 0.0539,
+      "step": 3959
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3584548814223702e-05,
+      "loss": 0.0473,
+      "step": 3960
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3581480491639573e-05,
+      "loss": 0.0502,
+      "step": 3961
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3578411782210072e-05,
+      "loss": 0.0621,
+      "step": 3962
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3575342686266656e-05,
+      "loss": 0.0548,
+      "step": 3963
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.3572273204140827e-05,
+      "loss": 0.0513,
+      "step": 3964
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3569203336164128e-05,
+      "loss": 0.0424,
+      "step": 3965
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3566133082668143e-05,
+      "loss": 0.0566,
+      "step": 3966
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3563062443984496e-05,
+      "loss": 0.044,
+      "step": 3967
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.355999142044486e-05,
+      "loss": 0.0539,
+      "step": 3968
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3556920012380945e-05,
+      "loss": 0.043,
+      "step": 3969
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.3553848220124497e-05,
+      "loss": 0.0459,
+      "step": 3970
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.355077604400731e-05,
+      "loss": 0.0587,
+      "step": 3971
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3547703484361219e-05,
+      "loss": 0.0402,
+      "step": 3972
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.35446305415181e-05,
+      "loss": 0.0604,
+      "step": 3973
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3541557215809868e-05,
+      "loss": 0.0516,
+      "step": 3974
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3538483507568486e-05,
+      "loss": 0.0509,
+      "step": 3975
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3535409417125946e-05,
+      "loss": 0.0666,
+      "step": 3976
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3532334944814296e-05,
+      "loss": 0.0489,
+      "step": 3977
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.3529260090965614e-05,
+      "loss": 0.0703,
+      "step": 3978
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3526184855912022e-05,
+      "loss": 0.0477,
+      "step": 3979
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3523109239985688e-05,
+      "loss": 0.0651,
+      "step": 3980
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3520033243518818e-05,
+      "loss": 0.0561,
+      "step": 3981
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3516956866843653e-05,
+      "loss": 0.0551,
+      "step": 3982
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3513880110292488e-05,
+      "loss": 0.0644,
+      "step": 3983
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.3510802974197648e-05,
+      "loss": 0.047,
+      "step": 3984
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3507725458891501e-05,
+      "loss": 0.0477,
+      "step": 3985
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3504647564706459e-05,
+      "loss": 0.055,
+      "step": 3986
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3501569291974979e-05,
+      "loss": 0.0501,
+      "step": 3987
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3498490641029544e-05,
+      "loss": 0.0532,
+      "step": 3988
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3495411612202696e-05,
+      "loss": 0.0632,
+      "step": 3989
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3492332205827004e-05,
+      "loss": 0.0514,
+      "step": 3990
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.3489252422235086e-05,
+      "loss": 0.0676,
+      "step": 3991
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3486172261759598e-05,
+      "loss": 0.0719,
+      "step": 3992
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3483091724733231e-05,
+      "loss": 0.0498,
+      "step": 3993
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3480010811488726e-05,
+      "loss": 0.054,
+      "step": 3994
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3476929522358861e-05,
+      "loss": 0.0587,
+      "step": 3995
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3473847857676457e-05,
+      "loss": 0.0616,
+      "step": 3996
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.3470765817774365e-05,
+      "loss": 0.0509,
+      "step": 3997
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3467683402985494e-05,
+      "loss": 0.0574,
+      "step": 3998
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.346460061364278e-05,
+      "loss": 0.0536,
+      "step": 3999
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.34615174500792e-05,
+      "loss": 0.05,
+      "step": 4000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3458433912627777e-05,
+      "loss": 0.0543,
+      "step": 4001
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3455350001621575e-05,
+      "loss": 0.0471,
+      "step": 4002
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3452265717393693e-05,
+      "loss": 0.0569,
+      "step": 4003
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.3449181060277271e-05,
+      "loss": 0.0577,
+      "step": 4004
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3446096030605495e-05,
+      "loss": 0.0619,
+      "step": 4005
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3443010628711588e-05,
+      "loss": 0.0513,
+      "step": 4006
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3439924854928808e-05,
+      "loss": 0.0558,
+      "step": 4007
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3436838709590461e-05,
+      "loss": 0.0505,
+      "step": 4008
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3433752193029888e-05,
+      "loss": 0.0469,
+      "step": 4009
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.3430665305580472e-05,
+      "loss": 0.0486,
+      "step": 4010
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3427578047575636e-05,
+      "loss": 0.0541,
+      "step": 4011
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3424490419348845e-05,
+      "loss": 0.08,
+      "step": 4012
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3421402421233605e-05,
+      "loss": 0.0476,
+      "step": 4013
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3418314053563455e-05,
+      "loss": 0.0651,
+      "step": 4014
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3415225316671973e-05,
+      "loss": 0.053,
+      "step": 4015
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.341213621089279e-05,
+      "loss": 0.0476,
+      "step": 4016
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.3409046736559564e-05,
+      "loss": 0.0521,
+      "step": 4017
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3405956894005998e-05,
+      "loss": 0.0552,
+      "step": 4018
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3402866683565838e-05,
+      "loss": 0.052,
+      "step": 4019
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3399776105572861e-05,
+      "loss": 0.0569,
+      "step": 4020
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3396685160360893e-05,
+      "loss": 0.0501,
+      "step": 4021
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3393593848263792e-05,
+      "loss": 0.0519,
+      "step": 4022
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.339050216961546e-05,
+      "loss": 0.0635,
+      "step": 4023
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.3387410124749837e-05,
+      "loss": 0.0488,
+      "step": 4024
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3384317714000907e-05,
+      "loss": 0.0538,
+      "step": 4025
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3381224937702683e-05,
+      "loss": 0.0548,
+      "step": 4026
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3378131796189228e-05,
+      "loss": 0.0499,
+      "step": 4027
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3375038289794644e-05,
+      "loss": 0.0444,
+      "step": 4028
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.337194441885306e-05,
+      "loss": 0.0528,
+      "step": 4029
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.3368850183698663e-05,
+      "loss": 0.0454,
+      "step": 4030
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3365755584665663e-05,
+      "loss": 0.0604,
+      "step": 4031
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3362660622088317e-05,
+      "loss": 0.0549,
+      "step": 4032
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.335956529630092e-05,
+      "loss": 0.054,
+      "step": 4033
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.335646960763781e-05,
+      "loss": 0.0485,
+      "step": 4034
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.335337355643336e-05,
+      "loss": 0.051,
+      "step": 4035
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3350277143021981e-05,
+      "loss": 0.0674,
+      "step": 4036
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.3347180367738124e-05,
+      "loss": 0.0549,
+      "step": 4037
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.334408323091628e-05,
+      "loss": 0.0534,
+      "step": 4038
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.3340985732890982e-05,
+      "loss": 0.0483,
+      "step": 4039
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.33378878739968e-05,
+      "loss": 0.0547,
+      "step": 4040
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.3334789654568339e-05,
+      "loss": 0.0656,
+      "step": 4041
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.3331691074940246e-05,
+      "loss": 0.0541,
+      "step": 4042
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.3328592135447207e-05,
+      "loss": 0.0541,
+      "step": 4043
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3325492836423947e-05,
+      "loss": 0.0463,
+      "step": 4044
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3322393178205233e-05,
+      "loss": 0.0592,
+      "step": 4045
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3319293161125862e-05,
+      "loss": 0.0614,
+      "step": 4046
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.331619278552068e-05,
+      "loss": 0.0568,
+      "step": 4047
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3313092051724563e-05,
+      "loss": 0.0502,
+      "step": 4048
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3309990960072432e-05,
+      "loss": 0.0647,
+      "step": 4049
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.3306889510899246e-05,
+      "loss": 0.0637,
+      "step": 4050
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3303787704539994e-05,
+      "loss": 0.0543,
+      "step": 4051
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.330068554132972e-05,
+      "loss": 0.0497,
+      "step": 4052
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3297583021603489e-05,
+      "loss": 0.05,
+      "step": 4053
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3294480145696418e-05,
+      "loss": 0.0394,
+      "step": 4054
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3291376913943651e-05,
+      "loss": 0.0557,
+      "step": 4055
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.3288273326680381e-05,
+      "loss": 0.0574,
+      "step": 4056
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3285169384241839e-05,
+      "loss": 0.0562,
+      "step": 4057
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.328206508696328e-05,
+      "loss": 0.0633,
+      "step": 4058
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3278960435180016e-05,
+      "loss": 0.0605,
+      "step": 4059
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3275855429227383e-05,
+      "loss": 0.0571,
+      "step": 4060
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3272750069440763e-05,
+      "loss": 0.0576,
+      "step": 4061
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3269644356155575e-05,
+      "loss": 0.0551,
+      "step": 4062
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.3266538289707279e-05,
+      "loss": 0.0514,
+      "step": 4063
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3263431870431362e-05,
+      "loss": 0.0579,
+      "step": 4064
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3260325098663364e-05,
+      "loss": 0.0475,
+      "step": 4065
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.325721797473885e-05,
+      "loss": 0.0557,
+      "step": 4066
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3254110498993431e-05,
+      "loss": 0.0556,
+      "step": 4067
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3251002671762756e-05,
+      "loss": 0.0526,
+      "step": 4068
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3247894493382507e-05,
+      "loss": 0.0637,
+      "step": 4069
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.3244785964188407e-05,
+      "loss": 0.0623,
+      "step": 4070
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3241677084516215e-05,
+      "loss": 0.0607,
+      "step": 4071
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3238567854701738e-05,
+      "loss": 0.0599,
+      "step": 4072
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3235458275080801e-05,
+      "loss": 0.0603,
+      "step": 4073
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3232348345989288e-05,
+      "loss": 0.0548,
+      "step": 4074
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.3229238067763101e-05,
+      "loss": 0.0673,
+      "step": 4075
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.32261274407382e-05,
+      "loss": 0.0644,
+      "step": 4076
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3223016465250562e-05,
+      "loss": 0.0654,
+      "step": 4077
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3219905141636215e-05,
+      "loss": 0.0602,
+      "step": 4078
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3216793470231228e-05,
+      "loss": 0.0559,
+      "step": 4079
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3213681451371697e-05,
+      "loss": 0.0523,
+      "step": 4080
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3210569085393756e-05,
+      "loss": 0.0575,
+      "step": 4081
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.3207456372633583e-05,
+      "loss": 0.0489,
+      "step": 4082
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.320434331342739e-05,
+      "loss": 0.0546,
+      "step": 4083
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3201229908111427e-05,
+      "loss": 0.0535,
+      "step": 4084
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3198116157021982e-05,
+      "loss": 0.06,
+      "step": 4085
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3195002060495381e-05,
+      "loss": 0.0543,
+      "step": 4086
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.3191887618867983e-05,
+      "loss": 0.0532,
+      "step": 4087
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.318877283247619e-05,
+      "loss": 0.052,
+      "step": 4088
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.318565770165643e-05,
+      "loss": 0.0491,
+      "step": 4089
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3182542226745192e-05,
+      "loss": 0.0464,
+      "step": 4090
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3179426408078975e-05,
+      "loss": 0.0455,
+      "step": 4091
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3176310245994328e-05,
+      "loss": 0.0517,
+      "step": 4092
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3173193740827838e-05,
+      "loss": 0.0596,
+      "step": 4093
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3170076892916133e-05,
+      "loss": 0.0594,
+      "step": 4094
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3166959702595863e-05,
+      "loss": 0.0603,
+      "step": 4095
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.3163842170203731e-05,
+      "loss": 0.0447,
+      "step": 4096
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3160724296076463e-05,
+      "loss": 0.0626,
+      "step": 4097
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3157606080550835e-05,
+      "loss": 0.0524,
+      "step": 4098
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3154487523963653e-05,
+      "loss": 0.0586,
+      "step": 4099
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3151368626651757e-05,
+      "loss": 0.057,
+      "step": 4100
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3148249388952032e-05,
+      "loss": 0.0601,
+      "step": 4101
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.3145129811201395e-05,
+      "loss": 0.0508,
+      "step": 4102
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3142009893736796e-05,
+      "loss": 0.054,
+      "step": 4103
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3138889636895227e-05,
+      "loss": 0.0581,
+      "step": 4104
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3135769041013717e-05,
+      "loss": 0.0565,
+      "step": 4105
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.313264810642933e-05,
+      "loss": 0.0522,
+      "step": 4106
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3129526833479166e-05,
+      "loss": 0.0579,
+      "step": 4107
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3126405222500362e-05,
+      "loss": 0.0504,
+      "step": 4108
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.3123283273830094e-05,
+      "loss": 0.0573,
+      "step": 4109
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3120160987805568e-05,
+      "loss": 0.0527,
+      "step": 4110
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3117038364764033e-05,
+      "loss": 0.052,
+      "step": 4111
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3113915405042769e-05,
+      "loss": 0.0708,
+      "step": 4112
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.31107921089791e-05,
+      "loss": 0.0534,
+      "step": 4113
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.310766847691038e-05,
+      "loss": 0.0457,
+      "step": 4114
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3104544509173998e-05,
+      "loss": 0.0558,
+      "step": 4115
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3101420206107385e-05,
+      "loss": 0.0535,
+      "step": 4116
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3098295568048007e-05,
+      "loss": 0.0603,
+      "step": 4117
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3095170595333359e-05,
+      "loss": 0.0562,
+      "step": 4118
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3092045288300983e-05,
+      "loss": 0.0634,
+      "step": 4119
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.308891964728845e-05,
+      "loss": 0.06,
+      "step": 4120
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3085793672633366e-05,
+      "loss": 0.0528,
+      "step": 4121
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3082667364673382e-05,
+      "loss": 0.0578,
+      "step": 4122
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3079540723746175e-05,
+      "loss": 0.0585,
+      "step": 4123
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3076413750189468e-05,
+      "loss": 0.0644,
+      "step": 4124
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3073286444341003e-05,
+      "loss": 0.0609,
+      "step": 4125
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3070158806538573e-05,
+      "loss": 0.0559,
+      "step": 4126
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3067030837120005e-05,
+      "loss": 0.0503,
+      "step": 4127
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3063902536423161e-05,
+      "loss": 0.0529,
+      "step": 4128
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3060773904785932e-05,
+      "loss": 0.0566,
+      "step": 4129
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3057644942546253e-05,
+      "loss": 0.0522,
+      "step": 4130
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3054515650042094e-05,
+      "loss": 0.0616,
+      "step": 4131
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3051386027611453e-05,
+      "loss": 0.0527,
+      "step": 4132
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.304825607559237e-05,
+      "loss": 0.0471,
+      "step": 4133
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3045125794322924e-05,
+      "loss": 0.0535,
+      "step": 4134
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.304199518414122e-05,
+      "loss": 0.0577,
+      "step": 4135
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3038864245385406e-05,
+      "loss": 0.0724,
+      "step": 4136
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3035732978393663e-05,
+      "loss": 0.0573,
+      "step": 4137
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3032601383504208e-05,
+      "loss": 0.0604,
+      "step": 4138
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3029469461055295e-05,
+      "loss": 0.054,
+      "step": 4139
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3026337211385207e-05,
+      "loss": 0.0678,
+      "step": 4140
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3023204634832266e-05,
+      "loss": 0.0531,
+      "step": 4141
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3020071731734838e-05,
+      "loss": 0.0532,
+      "step": 4142
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3016938502431307e-05,
+      "loss": 0.0496,
+      "step": 4143
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3013804947260104e-05,
+      "loss": 0.0596,
+      "step": 4144
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3010671066559701e-05,
+      "loss": 0.0603,
+      "step": 4145
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3007536860668588e-05,
+      "loss": 0.0635,
+      "step": 4146
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3004402329925299e-05,
+      "loss": 0.0618,
+      "step": 4147
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.3001267474668407e-05,
+      "loss": 0.0544,
+      "step": 4148
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2998132295236511e-05,
+      "loss": 0.0518,
+      "step": 4149
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.299499679196826e-05,
+      "loss": 0.0553,
+      "step": 4150
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2991860965202318e-05,
+      "loss": 0.0663,
+      "step": 4151
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2988724815277398e-05,
+      "loss": 0.0548,
+      "step": 4152
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2985588342532247e-05,
+      "loss": 0.0462,
+      "step": 4153
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.2982451547305639e-05,
+      "loss": 0.0598,
+      "step": 4154
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.297931442993639e-05,
+      "loss": 0.06,
+      "step": 4155
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.297617699076335e-05,
+      "loss": 0.0466,
+      "step": 4156
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2973039230125397e-05,
+      "loss": 0.0634,
+      "step": 4157
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2969901148361455e-05,
+      "loss": 0.0553,
+      "step": 4158
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2966762745810472e-05,
+      "loss": 0.0596,
+      "step": 4159
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.296362402281144e-05,
+      "loss": 0.061,
+      "step": 4160
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2960484979703381e-05,
+      "loss": 0.0498,
+      "step": 4161
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.2957345616825344e-05,
+      "loss": 0.0537,
+      "step": 4162
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2954205934516427e-05,
+      "loss": 0.0667,
+      "step": 4163
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2951065933115752e-05,
+      "loss": 0.0592,
+      "step": 4164
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2947925612962482e-05,
+      "loss": 0.0618,
+      "step": 4165
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.294478497439581e-05,
+      "loss": 0.0568,
+      "step": 4166
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2941644017754964e-05,
+      "loss": 0.067,
+      "step": 4167
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.2938502743379212e-05,
+      "loss": 0.0639,
+      "step": 4168
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2935361151607844e-05,
+      "loss": 0.0558,
+      "step": 4169
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2932219242780195e-05,
+      "loss": 0.0497,
+      "step": 4170
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2929077017235633e-05,
+      "loss": 0.0615,
+      "step": 4171
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2925934475313557e-05,
+      "loss": 0.0675,
+      "step": 4172
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.29227916173534e-05,
+      "loss": 0.0506,
+      "step": 4173
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2919648443694635e-05,
+      "loss": 0.0557,
+      "step": 4174
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.2916504954676759e-05,
+      "loss": 0.0501,
+      "step": 4175
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2913361150639311e-05,
+      "loss": 0.071,
+      "step": 4176
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2910217031921863e-05,
+      "loss": 0.0516,
+      "step": 4177
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2907072598864017e-05,
+      "loss": 0.0621,
+      "step": 4178
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2903927851805416e-05,
+      "loss": 0.0583,
+      "step": 4179
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2900782791085726e-05,
+      "loss": 0.059,
+      "step": 4180
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.2897637417044659e-05,
+      "loss": 0.0585,
+      "step": 4181
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2894491730021956e-05,
+      "loss": 0.0592,
+      "step": 4182
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2891345730357392e-05,
+      "loss": 0.0542,
+      "step": 4183
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2888199418390766e-05,
+      "loss": 0.0644,
+      "step": 4184
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2885052794461927e-05,
+      "loss": 0.0682,
+      "step": 4185
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2881905858910748e-05,
+      "loss": 0.0498,
+      "step": 4186
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.287875861207714e-05,
+      "loss": 0.0624,
+      "step": 4187
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.2875611054301047e-05,
+      "loss": 0.0651,
+      "step": 4188
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2872463185922439e-05,
+      "loss": 0.0532,
+      "step": 4189
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2869315007281334e-05,
+      "loss": 0.063,
+      "step": 4190
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2866166518717767e-05,
+      "loss": 0.0517,
+      "step": 4191
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2863017720571818e-05,
+      "loss": 0.056,
+      "step": 4192
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.28598686131836e-05,
+      "loss": 0.0634,
+      "step": 4193
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.2856719196893255e-05,
+      "loss": 0.0636,
+      "step": 4194
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2853569472040957e-05,
+      "loss": 0.047,
+      "step": 4195
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2850419438966917e-05,
+      "loss": 0.0547,
+      "step": 4196
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2847269098011385e-05,
+      "loss": 0.0634,
+      "step": 4197
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2844118449514629e-05,
+      "loss": 0.0647,
+      "step": 4198
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2840967493816967e-05,
+      "loss": 0.0554,
+      "step": 4199
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2837816231258736e-05,
+      "loss": 0.0607,
+      "step": 4200
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.2834664662180311e-05,
+      "loss": 0.057,
+      "step": 4201
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.283151278692211e-05,
+      "loss": 0.048,
+      "step": 4202
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2828360605824565e-05,
+      "loss": 0.058,
+      "step": 4203
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2825208119228165e-05,
+      "loss": 0.0525,
+      "step": 4204
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2822055327473409e-05,
+      "loss": 0.0623,
+      "step": 4205
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2818902230900837e-05,
+      "loss": 0.0594,
+      "step": 4206
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2815748829851028e-05,
+      "loss": 0.0517,
+      "step": 4207
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.2812595124664586e-05,
+      "loss": 0.0604,
+      "step": 4208
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2809441115682154e-05,
+      "loss": 0.0558,
+      "step": 4209
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.280628680324441e-05,
+      "loss": 0.0666,
+      "step": 4210
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2803132187692049e-05,
+      "loss": 0.0645,
+      "step": 4211
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2799977269365818e-05,
+      "loss": 0.0606,
+      "step": 4212
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2796822048606484e-05,
+      "loss": 0.0571,
+      "step": 4213
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.2793666525754849e-05,
+      "loss": 0.0603,
+      "step": 4214
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2790510701151756e-05,
+      "loss": 0.0567,
+      "step": 4215
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.278735457513807e-05,
+      "loss": 0.0524,
+      "step": 4216
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.278419814805469e-05,
+      "loss": 0.0578,
+      "step": 4217
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2781041420242555e-05,
+      "loss": 0.0488,
+      "step": 4218
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2777884392042631e-05,
+      "loss": 0.0558,
+      "step": 4219
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2774727063795918e-05,
+      "loss": 0.0515,
+      "step": 4220
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.2771569435843446e-05,
+      "loss": 0.0575,
+      "step": 4221
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2768411508526275e-05,
+      "loss": 0.0539,
+      "step": 4222
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2765253282185505e-05,
+      "loss": 0.0541,
+      "step": 4223
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2762094757162268e-05,
+      "loss": 0.0441,
+      "step": 4224
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.275893593379772e-05,
+      "loss": 0.0478,
+      "step": 4225
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.2755776812433056e-05,
+      "loss": 0.0538,
+      "step": 4226
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.27526173934095e-05,
+      "loss": 0.0534,
+      "step": 4227
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2749457677068312e-05,
+      "loss": 0.0537,
+      "step": 4228
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2746297663750778e-05,
+      "loss": 0.0532,
+      "step": 4229
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2743137353798218e-05,
+      "loss": 0.0602,
+      "step": 4230
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2739976747551991e-05,
+      "loss": 0.0632,
+      "step": 4231
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2736815845353483e-05,
+      "loss": 0.0602,
+      "step": 4232
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2733654647544107e-05,
+      "loss": 0.0557,
+      "step": 4233
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.2730493154465311e-05,
+      "loss": 0.0578,
+      "step": 4234
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2727331366458587e-05,
+      "loss": 0.0591,
+      "step": 4235
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2724169283865434e-05,
+      "loss": 0.0616,
+      "step": 4236
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2721006907027409e-05,
+      "loss": 0.0687,
+      "step": 4237
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.271784423628608e-05,
+      "loss": 0.057,
+      "step": 4238
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2714681271983062e-05,
+      "loss": 0.0495,
+      "step": 4239
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.2711518014459992e-05,
+      "loss": 0.0487,
+      "step": 4240
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2708354464058546e-05,
+      "loss": 0.0556,
+      "step": 4241
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.270519062112042e-05,
+      "loss": 0.052,
+      "step": 4242
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2702026485987358e-05,
+      "loss": 0.0649,
+      "step": 4243
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2698862059001117e-05,
+      "loss": 0.0533,
+      "step": 4244
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2695697340503506e-05,
+      "loss": 0.0577,
+      "step": 4245
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2692532330836346e-05,
+      "loss": 0.0514,
+      "step": 4246
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.2689367030341502e-05,
+      "loss": 0.0528,
+      "step": 4247
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.268620143936087e-05,
+      "loss": 0.062,
+      "step": 4248
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2683035558236367e-05,
+      "loss": 0.0604,
+      "step": 4249
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2679869387309956e-05,
+      "loss": 0.0569,
+      "step": 4250
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2676702926923615e-05,
+      "loss": 0.0597,
+      "step": 4251
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2673536177419367e-05,
+      "loss": 0.0536,
+      "step": 4252
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.267036913913926e-05,
+      "loss": 0.0549,
+      "step": 4253
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.2667201812425379e-05,
+      "loss": 0.0676,
+      "step": 4254
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2664034197619828e-05,
+      "loss": 0.0486,
+      "step": 4255
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2660866295064754e-05,
+      "loss": 0.0578,
+      "step": 4256
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2657698105102332e-05,
+      "loss": 0.068,
+      "step": 4257
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2654529628074761e-05,
+      "loss": 0.0544,
+      "step": 4258
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.2651360864324284e-05,
+      "loss": 0.0529,
+      "step": 4259
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.264819181419316e-05,
+      "loss": 0.0587,
+      "step": 4260
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2645022478023691e-05,
+      "loss": 0.0543,
+      "step": 4261
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2641852856158208e-05,
+      "loss": 0.0506,
+      "step": 4262
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2638682948939065e-05,
+      "loss": 0.0544,
+      "step": 4263
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.263551275670866e-05,
+      "loss": 0.0657,
+      "step": 4264
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2632342279809403e-05,
+      "loss": 0.0763,
+      "step": 4265
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2629171518583757e-05,
+      "loss": 0.0441,
+      "step": 4266
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.2626000473374194e-05,
+      "loss": 0.0595,
+      "step": 4267
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2622829144523231e-05,
+      "loss": 0.058,
+      "step": 4268
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2619657532373417e-05,
+      "loss": 0.0726,
+      "step": 4269
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2616485637267324e-05,
+      "loss": 0.0587,
+      "step": 4270
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2613313459547553e-05,
+      "loss": 0.0528,
+      "step": 4271
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2610140999556743e-05,
+      "loss": 0.0486,
+      "step": 4272
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.2606968257637557e-05,
+      "loss": 0.0552,
+      "step": 4273
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2603795234132694e-05,
+      "loss": 0.0587,
+      "step": 4274
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2600621929384882e-05,
+      "loss": 0.0553,
+      "step": 4275
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2597448343736875e-05,
+      "loss": 0.0435,
+      "step": 4276
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2594274477531461e-05,
+      "loss": 0.056,
+      "step": 4277
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.259110033111146e-05,
+      "loss": 0.0481,
+      "step": 4278
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.258792590481972e-05,
+      "loss": 0.0669,
+      "step": 4279
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.2584751198999115e-05,
+      "loss": 0.0683,
+      "step": 4280
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2581576213992558e-05,
+      "loss": 0.0622,
+      "step": 4281
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2578400950142988e-05,
+      "loss": 0.0655,
+      "step": 4282
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.257522540779337e-05,
+      "loss": 0.0629,
+      "step": 4283
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2572049587286706e-05,
+      "loss": 0.0561,
+      "step": 4284
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2568873488966022e-05,
+      "loss": 0.0623,
+      "step": 4285
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.2565697113174384e-05,
+      "loss": 0.0622,
+      "step": 4286
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2562520460254874e-05,
+      "loss": 0.065,
+      "step": 4287
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2559343530550612e-05,
+      "loss": 0.0565,
+      "step": 4288
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2556166324404747e-05,
+      "loss": 0.0612,
+      "step": 4289
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.255298884216046e-05,
+      "loss": 0.0642,
+      "step": 4290
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2549811084160956e-05,
+      "loss": 0.0544,
+      "step": 4291
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2546633050749477e-05,
+      "loss": 0.0693,
+      "step": 4292
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.2543454742269287e-05,
+      "loss": 0.0507,
+      "step": 4293
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2540276159063687e-05,
+      "loss": 0.0529,
+      "step": 4294
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2537097301476003e-05,
+      "loss": 0.0525,
+      "step": 4295
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2533918169849589e-05,
+      "loss": 0.0557,
+      "step": 4296
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2530738764527836e-05,
+      "loss": 0.0659,
+      "step": 4297
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.252755908585416e-05,
+      "loss": 0.0661,
+      "step": 4298
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2524379134172002e-05,
+      "loss": 0.0464,
+      "step": 4299
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.2521198909824844e-05,
+      "loss": 0.0638,
+      "step": 4300
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2518018413156185e-05,
+      "loss": 0.0652,
+      "step": 4301
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.251483764450956e-05,
+      "loss": 0.0643,
+      "step": 4302
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2511656604228538e-05,
+      "loss": 0.0612,
+      "step": 4303
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2508475292656702e-05,
+      "loss": 0.0608,
+      "step": 4304
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2505293710137682e-05,
+      "loss": 0.0536,
+      "step": 4305
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.2502111857015127e-05,
+      "loss": 0.064,
+      "step": 4306
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2498929733632714e-05,
+      "loss": 0.0539,
+      "step": 4307
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2495747340334159e-05,
+      "loss": 0.0615,
+      "step": 4308
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2492564677463198e-05,
+      "loss": 0.0496,
+      "step": 4309
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2489381745363599e-05,
+      "loss": 0.0569,
+      "step": 4310
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2486198544379156e-05,
+      "loss": 0.0674,
+      "step": 4311
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2483015074853699e-05,
+      "loss": 0.0616,
+      "step": 4312
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.2479831337131083e-05,
+      "loss": 0.0573,
+      "step": 4313
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2476647331555191e-05,
+      "loss": 0.0522,
+      "step": 4314
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2473463058469934e-05,
+      "loss": 0.0733,
+      "step": 4315
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2470278518219261e-05,
+      "loss": 0.06,
+      "step": 4316
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2467093711147133e-05,
+      "loss": 0.0587,
+      "step": 4317
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2463908637597557e-05,
+      "loss": 0.0565,
+      "step": 4318
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.2460723297914561e-05,
+      "loss": 0.0635,
+      "step": 4319
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2457537692442196e-05,
+      "loss": 0.0553,
+      "step": 4320
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2454351821524555e-05,
+      "loss": 0.0593,
+      "step": 4321
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2451165685505746e-05,
+      "loss": 0.0635,
+      "step": 4322
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2447979284729921e-05,
+      "loss": 0.0512,
+      "step": 4323
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2444792619541244e-05,
+      "loss": 0.0599,
+      "step": 4324
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2441605690283915e-05,
+      "loss": 0.0566,
+      "step": 4325
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.2438418497302168e-05,
+      "loss": 0.0548,
+      "step": 4326
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2435231040940257e-05,
+      "loss": 0.0651,
+      "step": 4327
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2432043321542468e-05,
+      "loss": 0.0597,
+      "step": 4328
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2428855339453116e-05,
+      "loss": 0.0525,
+      "step": 4329
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2425667095016544e-05,
+      "loss": 0.0694,
+      "step": 4330
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2422478588577125e-05,
+      "loss": 0.0568,
+      "step": 4331
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.2419289820479252e-05,
+      "loss": 0.0661,
+      "step": 4332
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2416100791067356e-05,
+      "loss": 0.0652,
+      "step": 4333
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.241291150068589e-05,
+      "loss": 0.0661,
+      "step": 4334
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2409721949679346e-05,
+      "loss": 0.0568,
+      "step": 4335
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2406532138392227e-05,
+      "loss": 0.0591,
+      "step": 4336
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2403342067169078e-05,
+      "loss": 0.0603,
+      "step": 4337
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2400151736354464e-05,
+      "loss": 0.0641,
+      "step": 4338
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.2396961146292985e-05,
+      "loss": 0.0577,
+      "step": 4339
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.239377029732926e-05,
+      "loss": 0.0498,
+      "step": 4340
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2390579189807948e-05,
+      "loss": 0.0481,
+      "step": 4341
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2387387824073725e-05,
+      "loss": 0.0698,
+      "step": 4342
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2384196200471297e-05,
+      "loss": 0.0647,
+      "step": 4343
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2381004319345402e-05,
+      "loss": 0.0761,
+      "step": 4344
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2377812181040807e-05,
+      "loss": 0.0589,
+      "step": 4345
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.2374619785902302e-05,
+      "loss": 0.0624,
+      "step": 4346
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2371427134274702e-05,
+      "loss": 0.0642,
+      "step": 4347
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2368234226502857e-05,
+      "loss": 0.064,
+      "step": 4348
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2365041062931643e-05,
+      "loss": 0.0686,
+      "step": 4349
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2361847643905955e-05,
+      "loss": 0.0694,
+      "step": 4350
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.2358653969770732e-05,
+      "loss": 0.0517,
+      "step": 4351
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.235546004087093e-05,
+      "loss": 0.0707,
+      "step": 4352
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2352265857551532e-05,
+      "loss": 0.0557,
+      "step": 4353
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2349071420157548e-05,
+      "loss": 0.0557,
+      "step": 4354
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2345876729034018e-05,
+      "loss": 0.0616,
+      "step": 4355
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2342681784526011e-05,
+      "loss": 0.0592,
+      "step": 4356
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2339486586978626e-05,
+      "loss": 0.0613,
+      "step": 4357
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2336291136736977e-05,
+      "loss": 0.0594,
+      "step": 4358
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.2333095434146216e-05,
+      "loss": 0.056,
+      "step": 4359
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2329899479551524e-05,
+      "loss": 0.0658,
+      "step": 4360
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2326703273298098e-05,
+      "loss": 0.0633,
+      "step": 4361
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2323506815731173e-05,
+      "loss": 0.0581,
+      "step": 4362
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2320310107196003e-05,
+      "loss": 0.0636,
+      "step": 4363
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2317113148037878e-05,
+      "loss": 0.0612,
+      "step": 4364
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.2313915938602105e-05,
+      "loss": 0.0636,
+      "step": 4365
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.231071847923403e-05,
+      "loss": 0.048,
+      "step": 4366
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2307520770279016e-05,
+      "loss": 0.0533,
+      "step": 4367
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2304322812082454e-05,
+      "loss": 0.0535,
+      "step": 4368
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2301124604989765e-05,
+      "loss": 0.0584,
+      "step": 4369
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2297926149346397e-05,
+      "loss": 0.0514,
+      "step": 4370
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2294727445497822e-05,
+      "loss": 0.0599,
+      "step": 4371
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.2291528493789545e-05,
+      "loss": 0.062,
+      "step": 4372
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.228832929456709e-05,
+      "loss": 0.0638,
+      "step": 4373
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.228512984817601e-05,
+      "loss": 0.0746,
+      "step": 4374
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.228193015496189e-05,
+      "loss": 0.0547,
+      "step": 4375
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.2278730215270334e-05,
+      "loss": 0.058,
+      "step": 4376
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.2275530029446975e-05,
+      "loss": 0.0679,
+      "step": 4377
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.2272329597837477e-05,
+      "loss": 0.061,
+      "step": 4378
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2269128920787527e-05,
+      "loss": 0.0584,
+      "step": 4379
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2265927998642839e-05,
+      "loss": 0.0674,
+      "step": 4380
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.226272683174915e-05,
+      "loss": 0.0627,
+      "step": 4381
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2259525420452232e-05,
+      "loss": 0.0513,
+      "step": 4382
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.225632376509787e-05,
+      "loss": 0.0499,
+      "step": 4383
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2253121866031893e-05,
+      "loss": 0.0575,
+      "step": 4384
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.2249919723600138e-05,
+      "loss": 0.0688,
+      "step": 4385
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2246717338148484e-05,
+      "loss": 0.0526,
+      "step": 4386
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2243514710022824e-05,
+      "loss": 0.0618,
+      "step": 4387
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2240311839569085e-05,
+      "loss": 0.0605,
+      "step": 4388
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2237108727133221e-05,
+      "loss": 0.0627,
+      "step": 4389
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2233905373061206e-05,
+      "loss": 0.0589,
+      "step": 4390
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2230701777699039e-05,
+      "loss": 0.0593,
+      "step": 4391
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.2227497941392753e-05,
+      "loss": 0.0521,
+      "step": 4392
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.22242938644884e-05,
+      "loss": 0.0527,
+      "step": 4393
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2221089547332066e-05,
+      "loss": 0.0663,
+      "step": 4394
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2217884990269857e-05,
+      "loss": 0.0526,
+      "step": 4395
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.22146801936479e-05,
+      "loss": 0.0514,
+      "step": 4396
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2211475157812362e-05,
+      "loss": 0.062,
+      "step": 4397
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.2208269883109422e-05,
+      "loss": 0.0683,
+      "step": 4398
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2205064369885291e-05,
+      "loss": 0.0697,
+      "step": 4399
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2201858618486205e-05,
+      "loss": 0.0578,
+      "step": 4400
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.219865262925843e-05,
+      "loss": 0.0691,
+      "step": 4401
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2195446402548249e-05,
+      "loss": 0.0591,
+      "step": 4402
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2192239938701973e-05,
+      "loss": 0.0549,
+      "step": 4403
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.218903323806595e-05,
+      "loss": 0.058,
+      "step": 4404
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.2185826300986536e-05,
+      "loss": 0.0641,
+      "step": 4405
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2182619127810124e-05,
+      "loss": 0.0533,
+      "step": 4406
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.217941171888313e-05,
+      "loss": 0.0653,
+      "step": 4407
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.217620407455199e-05,
+      "loss": 0.0639,
+      "step": 4408
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2172996195163178e-05,
+      "loss": 0.0578,
+      "step": 4409
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2169788081063181e-05,
+      "loss": 0.0542,
+      "step": 4410
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.2166579732598517e-05,
+      "loss": 0.0601,
+      "step": 4411
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2163371150115733e-05,
+      "loss": 0.0627,
+      "step": 4412
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2160162333961388e-05,
+      "loss": 0.0543,
+      "step": 4413
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2156953284482075e-05,
+      "loss": 0.0599,
+      "step": 4414
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2153744002024421e-05,
+      "loss": 0.0565,
+      "step": 4415
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2150534486935063e-05,
+      "loss": 0.0494,
+      "step": 4416
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.2147324739560674e-05,
+      "loss": 0.0582,
+      "step": 4417
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.214411476024794e-05,
+      "loss": 0.0689,
+      "step": 4418
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2140904549343589e-05,
+      "loss": 0.0661,
+      "step": 4419
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2137694107194353e-05,
+      "loss": 0.0637,
+      "step": 4420
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.213448343414701e-05,
+      "loss": 0.0545,
+      "step": 4421
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.213127253054835e-05,
+      "loss": 0.065,
+      "step": 4422
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2128061396745193e-05,
+      "loss": 0.063,
+      "step": 4423
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.2124850033084378e-05,
+      "loss": 0.0559,
+      "step": 4424
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2121638439912777e-05,
+      "loss": 0.0756,
+      "step": 4425
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2118426617577284e-05,
+      "loss": 0.0647,
+      "step": 4426
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2115214566424812e-05,
+      "loss": 0.0595,
+      "step": 4427
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2112002286802309e-05,
+      "loss": 0.0601,
+      "step": 4428
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2108789779056735e-05,
+      "loss": 0.0659,
+      "step": 4429
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2105577043535086e-05,
+      "loss": 0.064,
+      "step": 4430
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.2102364080584376e-05,
+      "loss": 0.0623,
+      "step": 4431
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.209915089055165e-05,
+      "loss": 0.0724,
+      "step": 4432
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.2095937473783973e-05,
+      "loss": 0.0554,
+      "step": 4433
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.209272383062843e-05,
+      "loss": 0.0606,
+      "step": 4434
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.2089509961432138e-05,
+      "loss": 0.0613,
+      "step": 4435
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.2086295866542233e-05,
+      "loss": 0.0557,
+      "step": 4436
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.208308154630588e-05,
+      "loss": 0.0634,
+      "step": 4437
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.2079867001070271e-05,
+      "loss": 0.0689,
+      "step": 4438
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2076652231182613e-05,
+      "loss": 0.0565,
+      "step": 4439
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2073437236990142e-05,
+      "loss": 0.0553,
+      "step": 4440
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.207022201884012e-05,
+      "loss": 0.0541,
+      "step": 4441
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.206700657707983e-05,
+      "loss": 0.0607,
+      "step": 4442
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2063790912056577e-05,
+      "loss": 0.0653,
+      "step": 4443
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.2060575024117702e-05,
+      "loss": 0.0581,
+      "step": 4444
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2057358913610554e-05,
+      "loss": 0.0731,
+      "step": 4445
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2054142580882517e-05,
+      "loss": 0.061,
+      "step": 4446
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2050926026280998e-05,
+      "loss": 0.0553,
+      "step": 4447
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2047709250153423e-05,
+      "loss": 0.0597,
+      "step": 4448
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2044492252847244e-05,
+      "loss": 0.0539,
+      "step": 4449
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.204127503470994e-05,
+      "loss": 0.0718,
+      "step": 4450
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.2038057596089008e-05,
+      "loss": 0.0589,
+      "step": 4451
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2034839937331977e-05,
+      "loss": 0.0628,
+      "step": 4452
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2031622058786389e-05,
+      "loss": 0.061,
+      "step": 4453
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2028403960799821e-05,
+      "loss": 0.0507,
+      "step": 4454
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2025185643719868e-05,
+      "loss": 0.0551,
+      "step": 4455
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.2021967107894148e-05,
+      "loss": 0.0586,
+      "step": 4456
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.20187483536703e-05,
+      "loss": 0.0518,
+      "step": 4457
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2015529381395993e-05,
+      "loss": 0.0625,
+      "step": 4458
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2012310191418922e-05,
+      "loss": 0.0602,
+      "step": 4459
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2009090784086792e-05,
+      "loss": 0.0629,
+      "step": 4460
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2005871159747348e-05,
+      "loss": 0.0525,
+      "step": 4461
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.2002651318748344e-05,
+      "loss": 0.0531,
+      "step": 4462
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.1999431261437567e-05,
+      "loss": 0.0636,
+      "step": 4463
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.1996210988162822e-05,
+      "loss": 0.0557,
+      "step": 4464
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1992990499271939e-05,
+      "loss": 0.0551,
+      "step": 4465
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1989769795112775e-05,
+      "loss": 0.0575,
+      "step": 4466
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1986548876033203e-05,
+      "loss": 0.061,
+      "step": 4467
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1983327742381129e-05,
+      "loss": 0.0568,
+      "step": 4468
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1980106394504469e-05,
+      "loss": 0.066,
+      "step": 4469
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.1976884832751174e-05,
+      "loss": 0.0676,
+      "step": 4470
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1973663057469216e-05,
+      "loss": 0.0493,
+      "step": 4471
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.197044106900658e-05,
+      "loss": 0.0492,
+      "step": 4472
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1967218867711285e-05,
+      "loss": 0.0667,
+      "step": 4473
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1963996453931372e-05,
+      "loss": 0.0578,
+      "step": 4474
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1960773828014903e-05,
+      "loss": 0.0547,
+      "step": 4475
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.1957550990309958e-05,
+      "loss": 0.0669,
+      "step": 4476
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.195432794116465e-05,
+      "loss": 0.0615,
+      "step": 4477
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1951104680927105e-05,
+      "loss": 0.0681,
+      "step": 4478
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1947881209945477e-05,
+      "loss": 0.0492,
+      "step": 4479
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1944657528567944e-05,
+      "loss": 0.0594,
+      "step": 4480
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.19414336371427e-05,
+      "loss": 0.0623,
+      "step": 4481
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1938209536017973e-05,
+      "loss": 0.0609,
+      "step": 4482
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1934985225541998e-05,
+      "loss": 0.0592,
+      "step": 4483
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.1931760706063047e-05,
+      "loss": 0.0735,
+      "step": 4484
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1928535977929413e-05,
+      "loss": 0.0635,
+      "step": 4485
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1925311041489398e-05,
+      "loss": 0.0555,
+      "step": 4486
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1922085897091342e-05,
+      "loss": 0.0573,
+      "step": 4487
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1918860545083602e-05,
+      "loss": 0.0636,
+      "step": 4488
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1915634985814555e-05,
+      "loss": 0.0698,
+      "step": 4489
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.1912409219632602e-05,
+      "loss": 0.0666,
+      "step": 4490
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1909183246886167e-05,
+      "loss": 0.0651,
+      "step": 4491
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1905957067923699e-05,
+      "loss": 0.0601,
+      "step": 4492
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1902730683093662e-05,
+      "loss": 0.0639,
+      "step": 4493
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1899504092744549e-05,
+      "loss": 0.062,
+      "step": 4494
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1896277297224871e-05,
+      "loss": 0.0585,
+      "step": 4495
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.1893050296883165e-05,
+      "loss": 0.0539,
+      "step": 4496
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.188982309206799e-05,
+      "loss": 0.0511,
+      "step": 4497
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1886595683127917e-05,
+      "loss": 0.0637,
+      "step": 4498
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1883368070411556e-05,
+      "loss": 0.0601,
+      "step": 4499
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1880140254267527e-05,
+      "loss": 0.0631,
+      "step": 4500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1876912235044473e-05,
+      "loss": 0.0564,
+      "step": 4501
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1873684013091063e-05,
+      "loss": 0.0603,
+      "step": 4502
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1870455588755984e-05,
+      "loss": 0.059,
+      "step": 4503
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1867226962387952e-05,
+      "loss": 0.0642,
+      "step": 4504
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1863998134335693e-05,
+      "loss": 0.049,
+      "step": 4505
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1860769104947965e-05,
+      "loss": 0.0724,
+      "step": 4506
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1857539874573547e-05,
+      "loss": 0.0624,
+      "step": 4507
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.185431044356123e-05,
+      "loss": 0.055,
+      "step": 4508
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1851080812259839e-05,
+      "loss": 0.0574,
+      "step": 4509
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.1847850981018213e-05,
+      "loss": 0.0606,
+      "step": 4510
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1844620950185216e-05,
+      "loss": 0.0614,
+      "step": 4511
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1841390720109728e-05,
+      "loss": 0.0731,
+      "step": 4512
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.183816029114066e-05,
+      "loss": 0.063,
+      "step": 4513
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1834929663626941e-05,
+      "loss": 0.0618,
+      "step": 4514
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1831698837917514e-05,
+      "loss": 0.0741,
+      "step": 4515
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.1828467814361353e-05,
+      "loss": 0.0556,
+      "step": 4516
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1825236593307448e-05,
+      "loss": 0.0619,
+      "step": 4517
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1822005175104809e-05,
+      "loss": 0.0665,
+      "step": 4518
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1818773560102477e-05,
+      "loss": 0.067,
+      "step": 4519
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1815541748649504e-05,
+      "loss": 0.0547,
+      "step": 4520
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1812309741094965e-05,
+      "loss": 0.0523,
+      "step": 4521
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1809077537787958e-05,
+      "loss": 0.058,
+      "step": 4522
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.1805845139077607e-05,
+      "loss": 0.0579,
+      "step": 4523
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1802612545313048e-05,
+      "loss": 0.0647,
+      "step": 4524
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.179937975684344e-05,
+      "loss": 0.0554,
+      "step": 4525
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1796146774017972e-05,
+      "loss": 0.0634,
+      "step": 4526
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.179291359718584e-05,
+      "loss": 0.0533,
+      "step": 4527
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1789680226696271e-05,
+      "loss": 0.0581,
+      "step": 4528
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1786446662898512e-05,
+      "loss": 0.064,
+      "step": 4529
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.1783212906141827e-05,
+      "loss": 0.0618,
+      "step": 4530
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.1779978956775507e-05,
+      "loss": 0.0541,
+      "step": 4531
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.177674481514885e-05,
+      "loss": 0.0539,
+      "step": 4532
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.177351048161119e-05,
+      "loss": 0.0668,
+      "step": 4533
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.177027595651188e-05,
+      "loss": 0.0632,
+      "step": 4534
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.1767041240200282e-05,
+      "loss": 0.0627,
+      "step": 4535
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.1763806333025793e-05,
+      "loss": 0.0533,
+      "step": 4536
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1760571235337821e-05,
+      "loss": 0.0628,
+      "step": 4537
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1757335947485799e-05,
+      "loss": 0.0588,
+      "step": 4538
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1754100469819175e-05,
+      "loss": 0.0617,
+      "step": 4539
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1750864802687428e-05,
+      "loss": 0.0562,
+      "step": 4540
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1747628946440047e-05,
+      "loss": 0.0671,
+      "step": 4541
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1744392901426549e-05,
+      "loss": 0.0612,
+      "step": 4542
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.1741156667996462e-05,
+      "loss": 0.0597,
+      "step": 4543
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1737920246499344e-05,
+      "loss": 0.0511,
+      "step": 4544
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1734683637284774e-05,
+      "loss": 0.0604,
+      "step": 4545
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1731446840702337e-05,
+      "loss": 0.0541,
+      "step": 4546
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1728209857101655e-05,
+      "loss": 0.0598,
+      "step": 4547
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1724972686832365e-05,
+      "loss": 0.0592,
+      "step": 4548
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.1721735330244117e-05,
+      "loss": 0.0578,
+      "step": 4549
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.171849778768659e-05,
+      "loss": 0.0625,
+      "step": 4550
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1715260059509474e-05,
+      "loss": 0.0703,
+      "step": 4551
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1712022146062496e-05,
+      "loss": 0.0573,
+      "step": 4552
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1708784047695385e-05,
+      "loss": 0.0635,
+      "step": 4553
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1705545764757892e-05,
+      "loss": 0.0605,
+      "step": 4554
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1702307297599802e-05,
+      "loss": 0.0619,
+      "step": 4555
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.1699068646570903e-05,
+      "loss": 0.0645,
+      "step": 4556
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1695829812021015e-05,
+      "loss": 0.0605,
+      "step": 4557
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1692590794299972e-05,
+      "loss": 0.0619,
+      "step": 4558
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1689351593757632e-05,
+      "loss": 0.0574,
+      "step": 4559
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1686112210743865e-05,
+      "loss": 0.0546,
+      "step": 4560
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1682872645608565e-05,
+      "loss": 0.0548,
+      "step": 4561
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.1679632898701649e-05,
+      "loss": 0.0551,
+      "step": 4562
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1676392970373052e-05,
+      "loss": 0.0603,
+      "step": 4563
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1673152860972725e-05,
+      "loss": 0.053,
+      "step": 4564
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.166991257085064e-05,
+      "loss": 0.0592,
+      "step": 4565
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1666672100356794e-05,
+      "loss": 0.0741,
+      "step": 4566
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1663431449841196e-05,
+      "loss": 0.0687,
+      "step": 4567
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1660190619653874e-05,
+      "loss": 0.0678,
+      "step": 4568
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.1656949610144887e-05,
+      "loss": 0.0757,
+      "step": 4569
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1653708421664299e-05,
+      "loss": 0.0668,
+      "step": 4570
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1650467054562198e-05,
+      "loss": 0.0677,
+      "step": 4571
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1647225509188697e-05,
+      "loss": 0.0588,
+      "step": 4572
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1643983785893924e-05,
+      "loss": 0.0602,
+      "step": 4573
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1640741885028027e-05,
+      "loss": 0.0673,
+      "step": 4574
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.163749980694117e-05,
+      "loss": 0.0696,
+      "step": 4575
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.1634257551983537e-05,
+      "loss": 0.0585,
+      "step": 4576
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1631015120505336e-05,
+      "loss": 0.0581,
+      "step": 4577
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1627772512856788e-05,
+      "loss": 0.0635,
+      "step": 4578
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.162452972938814e-05,
+      "loss": 0.0647,
+      "step": 4579
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1621286770449651e-05,
+      "loss": 0.0602,
+      "step": 4580
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.16180436363916e-05,
+      "loss": 0.0525,
+      "step": 4581
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.1614800327564293e-05,
+      "loss": 0.0678,
+      "step": 4582
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1611556844318041e-05,
+      "loss": 0.0705,
+      "step": 4583
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1608313187003185e-05,
+      "loss": 0.0658,
+      "step": 4584
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.160506935597008e-05,
+      "loss": 0.0702,
+      "step": 4585
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1601825351569105e-05,
+      "loss": 0.0686,
+      "step": 4586
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1598581174150647e-05,
+      "loss": 0.0587,
+      "step": 4587
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1595336824065123e-05,
+      "loss": 0.072,
+      "step": 4588
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.1592092301662963e-05,
+      "loss": 0.0769,
+      "step": 4589
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1588847607294617e-05,
+      "loss": 0.0623,
+      "step": 4590
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1585602741310554e-05,
+      "loss": 0.0719,
+      "step": 4591
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1582357704061255e-05,
+      "loss": 0.0653,
+      "step": 4592
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1579112495897231e-05,
+      "loss": 0.0657,
+      "step": 4593
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1575867117169006e-05,
+      "loss": 0.0633,
+      "step": 4594
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.1572621568227119e-05,
+      "loss": 0.0615,
+      "step": 4595
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1569375849422134e-05,
+      "loss": 0.0604,
+      "step": 4596
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1566129961104628e-05,
+      "loss": 0.069,
+      "step": 4597
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1562883903625195e-05,
+      "loss": 0.0693,
+      "step": 4598
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1559637677334455e-05,
+      "loss": 0.0515,
+      "step": 4599
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.155639128258304e-05,
+      "loss": 0.0381,
+      "step": 4600
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.1553144719721598e-05,
+      "loss": 0.0384,
+      "step": 4601
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.154989798910081e-05,
+      "loss": 0.0409,
+      "step": 4602
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1546651091071353e-05,
+      "loss": 0.0389,
+      "step": 4603
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.154340402598394e-05,
+      "loss": 0.0449,
+      "step": 4604
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1540156794189287e-05,
+      "loss": 0.0353,
+      "step": 4605
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1536909396038145e-05,
+      "loss": 0.0422,
+      "step": 4606
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.153366183188127e-05,
+      "loss": 0.0393,
+      "step": 4607
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.1530414102069444e-05,
+      "loss": 0.0453,
+      "step": 4608
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.152716620695346e-05,
+      "loss": 0.0401,
+      "step": 4609
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.152391814688413e-05,
+      "loss": 0.0343,
+      "step": 4610
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1520669922212288e-05,
+      "loss": 0.041,
+      "step": 4611
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1517421533288784e-05,
+      "loss": 0.0372,
+      "step": 4612
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1514172980464486e-05,
+      "loss": 0.0294,
+      "step": 4613
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1510924264090274e-05,
+      "loss": 0.0396,
+      "step": 4614
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.1507675384517059e-05,
+      "loss": 0.0441,
+      "step": 4615
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.150442634209575e-05,
+      "loss": 0.0404,
+      "step": 4616
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.1501177137177297e-05,
+      "loss": 0.0477,
+      "step": 4617
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.149792777011265e-05,
+      "loss": 0.0396,
+      "step": 4618
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.1494678241252781e-05,
+      "loss": 0.0463,
+      "step": 4619
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.149142855094868e-05,
+      "loss": 0.0412,
+      "step": 4620
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.1488178699551353e-05,
+      "loss": 0.0402,
+      "step": 4621
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1484928687411832e-05,
+      "loss": 0.0397,
+      "step": 4622
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1481678514881154e-05,
+      "loss": 0.0394,
+      "step": 4623
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1478428182310381e-05,
+      "loss": 0.0481,
+      "step": 4624
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.147517769005059e-05,
+      "loss": 0.0395,
+      "step": 4625
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1471927038452877e-05,
+      "loss": 0.0427,
+      "step": 4626
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1468676227868348e-05,
+      "loss": 0.0385,
+      "step": 4627
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.1465425258648138e-05,
+      "loss": 0.0385,
+      "step": 4628
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1462174131143391e-05,
+      "loss": 0.0424,
+      "step": 4629
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1458922845705267e-05,
+      "loss": 0.0325,
+      "step": 4630
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1455671402684952e-05,
+      "loss": 0.0379,
+      "step": 4631
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1452419802433636e-05,
+      "loss": 0.0433,
+      "step": 4632
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.144916804530254e-05,
+      "loss": 0.0255,
+      "step": 4633
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1445916131642893e-05,
+      "loss": 0.0415,
+      "step": 4634
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1442664061805942e-05,
+      "loss": 0.0354,
+      "step": 4635
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.143941183614295e-05,
+      "loss": 0.048,
+      "step": 4636
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.14361594550052e-05,
+      "loss": 0.0373,
+      "step": 4637
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1432906918743992e-05,
+      "loss": 0.0399,
+      "step": 4638
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.142965422771064e-05,
+      "loss": 0.0448,
+      "step": 4639
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.142640138225648e-05,
+      "loss": 0.0348,
+      "step": 4640
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.0423,
+      "step": 4641
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.141989522949113e-05,
+      "loss": 0.0462,
+      "step": 4642
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.141664192288269e-05,
+      "loss": 0.0379,
+      "step": 4643
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.141338846325893e-05,
+      "loss": 0.0347,
+      "step": 4644
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1410134850971268e-05,
+      "loss": 0.0423,
+      "step": 4645
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.140688108637114e-05,
+      "loss": 0.046,
+      "step": 4646
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1403627169809983e-05,
+      "loss": 0.0471,
+      "step": 4647
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1400373101639271e-05,
+      "loss": 0.039,
+      "step": 4648
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1397118882210479e-05,
+      "loss": 0.0327,
+      "step": 4649
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1393864511875106e-05,
+      "loss": 0.0372,
+      "step": 4650
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1390609990984664e-05,
+      "loss": 0.0523,
+      "step": 4651
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1387355319890685e-05,
+      "loss": 0.0344,
+      "step": 4652
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.138410049894471e-05,
+      "loss": 0.0474,
+      "step": 4653
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1380845528498307e-05,
+      "loss": 0.0374,
+      "step": 4654
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1377590408903055e-05,
+      "loss": 0.0464,
+      "step": 4655
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.137433514051054e-05,
+      "loss": 0.0399,
+      "step": 4656
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.137107972367238e-05,
+      "loss": 0.0411,
+      "step": 4657
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1367824158740195e-05,
+      "loss": 0.0329,
+      "step": 4658
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1364568446065633e-05,
+      "loss": 0.0445,
+      "step": 4659
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1361312586000351e-05,
+      "loss": 0.0448,
+      "step": 4660
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.135805657889602e-05,
+      "loss": 0.0369,
+      "step": 4661
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1354800425104335e-05,
+      "loss": 0.0503,
+      "step": 4662
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1351544124977001e-05,
+      "loss": 0.044,
+      "step": 4663
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1348287678865735e-05,
+      "loss": 0.045,
+      "step": 4664
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1345031087122277e-05,
+      "loss": 0.0451,
+      "step": 4665
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1341774350098383e-05,
+      "loss": 0.0354,
+      "step": 4666
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.133851746814582e-05,
+      "loss": 0.0411,
+      "step": 4667
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1335260441616372e-05,
+      "loss": 0.0445,
+      "step": 4668
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.133200327086184e-05,
+      "loss": 0.047,
+      "step": 4669
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.132874595623404e-05,
+      "loss": 0.0367,
+      "step": 4670
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1325488498084803e-05,
+      "loss": 0.0343,
+      "step": 4671
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1322230896765975e-05,
+      "loss": 0.0419,
+      "step": 4672
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.131897315262942e-05,
+      "loss": 0.0379,
+      "step": 4673
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1315715266027014e-05,
+      "loss": 0.0395,
+      "step": 4674
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1312457237310652e-05,
+      "loss": 0.037,
+      "step": 4675
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1309199066832239e-05,
+      "loss": 0.0421,
+      "step": 4676
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1305940754943706e-05,
+      "loss": 0.0386,
+      "step": 4677
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1302682301996985e-05,
+      "loss": 0.0382,
+      "step": 4678
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1299423708344033e-05,
+      "loss": 0.0358,
+      "step": 4679
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1296164974336816e-05,
+      "loss": 0.0429,
+      "step": 4680
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1292906100327325e-05,
+      "loss": 0.0369,
+      "step": 4681
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1289647086667554e-05,
+      "loss": 0.041,
+      "step": 4682
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1286387933709522e-05,
+      "loss": 0.0436,
+      "step": 4683
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1283128641805259e-05,
+      "loss": 0.0409,
+      "step": 4684
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.127986921130681e-05,
+      "loss": 0.0377,
+      "step": 4685
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1276609642566227e-05,
+      "loss": 0.0383,
+      "step": 4686
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.1273349935935594e-05,
+      "loss": 0.0387,
+      "step": 4687
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1270090091766996e-05,
+      "loss": 0.043,
+      "step": 4688
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1266830110412541e-05,
+      "loss": 0.0392,
+      "step": 4689
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1263569992224346e-05,
+      "loss": 0.0375,
+      "step": 4690
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1260309737554546e-05,
+      "loss": 0.0313,
+      "step": 4691
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1257049346755292e-05,
+      "loss": 0.0413,
+      "step": 4692
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1253788820178741e-05,
+      "loss": 0.035,
+      "step": 4693
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.1250528158177079e-05,
+      "loss": 0.0422,
+      "step": 4694
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1247267361102495e-05,
+      "loss": 0.0439,
+      "step": 4695
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1244006429307195e-05,
+      "loss": 0.0368,
+      "step": 4696
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1240745363143401e-05,
+      "loss": 0.0426,
+      "step": 4697
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1237484162963355e-05,
+      "loss": 0.0397,
+      "step": 4698
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1234222829119305e-05,
+      "loss": 0.042,
+      "step": 4699
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.1230961361963517e-05,
+      "loss": 0.0395,
+      "step": 4700
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1227699761848268e-05,
+      "loss": 0.0539,
+      "step": 4701
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1224438029125854e-05,
+      "loss": 0.0505,
+      "step": 4702
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1221176164148585e-05,
+      "loss": 0.0407,
+      "step": 4703
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.121791416726878e-05,
+      "loss": 0.0439,
+      "step": 4704
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1214652038838778e-05,
+      "loss": 0.042,
+      "step": 4705
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1211389779210935e-05,
+      "loss": 0.0332,
+      "step": 4706
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.1208127388737613e-05,
+      "loss": 0.0479,
+      "step": 4707
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1204864867771189e-05,
+      "loss": 0.0479,
+      "step": 4708
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1201602216664055e-05,
+      "loss": 0.0411,
+      "step": 4709
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1198339435768627e-05,
+      "loss": 0.0415,
+      "step": 4710
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.119507652543732e-05,
+      "loss": 0.0437,
+      "step": 4711
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1191813486022573e-05,
+      "loss": 0.0373,
+      "step": 4712
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.1188550317876833e-05,
+      "loss": 0.0468,
+      "step": 4713
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.118528702135257e-05,
+      "loss": 0.0287,
+      "step": 4714
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1182023596802252e-05,
+      "loss": 0.0386,
+      "step": 4715
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1178760044578376e-05,
+      "loss": 0.0379,
+      "step": 4716
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1175496365033449e-05,
+      "loss": 0.0433,
+      "step": 4717
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1172232558519983e-05,
+      "loss": 0.0412,
+      "step": 4718
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1168968625390515e-05,
+      "loss": 0.0352,
+      "step": 4719
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.1165704565997593e-05,
+      "loss": 0.046,
+      "step": 4720
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1162440380693775e-05,
+      "loss": 0.0378,
+      "step": 4721
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1159176069831635e-05,
+      "loss": 0.0387,
+      "step": 4722
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1155911633763754e-05,
+      "loss": 0.0389,
+      "step": 4723
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1152647072842739e-05,
+      "loss": 0.0399,
+      "step": 4724
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1149382387421204e-05,
+      "loss": 0.0366,
+      "step": 4725
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1146117577851773e-05,
+      "loss": 0.0372,
+      "step": 4726
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.1142852644487095e-05,
+      "loss": 0.0347,
+      "step": 4727
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1139587587679814e-05,
+      "loss": 0.0426,
+      "step": 4728
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1136322407782603e-05,
+      "loss": 0.0429,
+      "step": 4729
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1133057105148142e-05,
+      "loss": 0.0501,
+      "step": 4730
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1129791680129123e-05,
+      "loss": 0.0361,
+      "step": 4731
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1126526133078255e-05,
+      "loss": 0.0505,
+      "step": 4732
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.1123260464348262e-05,
+      "loss": 0.0469,
+      "step": 4733
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1119994674291873e-05,
+      "loss": 0.0502,
+      "step": 4734
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1116728763261834e-05,
+      "loss": 0.0435,
+      "step": 4735
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.111346273161091e-05,
+      "loss": 0.0446,
+      "step": 4736
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1110196579691871e-05,
+      "loss": 0.0407,
+      "step": 4737
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1106930307857503e-05,
+      "loss": 0.0363,
+      "step": 4738
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1103663916460605e-05,
+      "loss": 0.044,
+      "step": 4739
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.1100397405853988e-05,
+      "loss": 0.0429,
+      "step": 4740
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1097130776390477e-05,
+      "loss": 0.0379,
+      "step": 4741
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.109386402842291e-05,
+      "loss": 0.0434,
+      "step": 4742
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1090597162304137e-05,
+      "loss": 0.038,
+      "step": 4743
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1087330178387024e-05,
+      "loss": 0.0403,
+      "step": 4744
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1084063077024438e-05,
+      "loss": 0.0438,
+      "step": 4745
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.1080795858569275e-05,
+      "loss": 0.033,
+      "step": 4746
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.1077528523374436e-05,
+      "loss": 0.0433,
+      "step": 4747
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.107426107179283e-05,
+      "loss": 0.035,
+      "step": 4748
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.1070993504177392e-05,
+      "loss": 0.0381,
+      "step": 4749
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.106772582088105e-05,
+      "loss": 0.0481,
+      "step": 4750
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.1064458022256764e-05,
+      "loss": 0.0421,
+      "step": 4751
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.106119010865749e-05,
+      "loss": 0.035,
+      "step": 4752
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.105792208043621e-05,
+      "loss": 0.0438,
+      "step": 4753
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1054653937945909e-05,
+      "loss": 0.0466,
+      "step": 4754
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1051385681539592e-05,
+      "loss": 0.0305,
+      "step": 4755
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.104811731157027e-05,
+      "loss": 0.0341,
+      "step": 4756
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1044848828390966e-05,
+      "loss": 0.0346,
+      "step": 4757
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1041580232354724e-05,
+      "loss": 0.037,
+      "step": 4758
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.1038311523814584e-05,
+      "loss": 0.042,
+      "step": 4759
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.103504270312362e-05,
+      "loss": 0.0417,
+      "step": 4760
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1031773770634896e-05,
+      "loss": 0.0542,
+      "step": 4761
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1028504726701502e-05,
+      "loss": 0.0401,
+      "step": 4762
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1025235571676537e-05,
+      "loss": 0.0425,
+      "step": 4763
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1021966305913112e-05,
+      "loss": 0.0426,
+      "step": 4764
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.101869692976435e-05,
+      "loss": 0.0391,
+      "step": 4765
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.1015427443583384e-05,
+      "loss": 0.0419,
+      "step": 4766
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.1012157847723355e-05,
+      "loss": 0.0391,
+      "step": 4767
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.1008888142537426e-05,
+      "loss": 0.0332,
+      "step": 4768
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.1005618328378764e-05,
+      "loss": 0.0414,
+      "step": 4769
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.1002348405600554e-05,
+      "loss": 0.0448,
+      "step": 4770
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.099907837455599e-05,
+      "loss": 0.0397,
+      "step": 4771
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.0995808235598271e-05,
+      "loss": 0.043,
+      "step": 4772
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.0992537989080618e-05,
+      "loss": 0.0322,
+      "step": 4773
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0989267635356257e-05,
+      "loss": 0.0387,
+      "step": 4774
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.098599717477843e-05,
+      "loss": 0.0326,
+      "step": 4775
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0982726607700387e-05,
+      "loss": 0.0516,
+      "step": 4776
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.097945593447539e-05,
+      "loss": 0.0381,
+      "step": 4777
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0976185155456713e-05,
+      "loss": 0.052,
+      "step": 4778
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.0972914270997644e-05,
+      "loss": 0.0374,
+      "step": 4779
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.096964328145148e-05,
+      "loss": 0.0438,
+      "step": 4780
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0966372187171524e-05,
+      "loss": 0.0436,
+      "step": 4781
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0963100988511104e-05,
+      "loss": 0.0523,
+      "step": 4782
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0959829685823544e-05,
+      "loss": 0.0378,
+      "step": 4783
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0956558279462188e-05,
+      "loss": 0.0364,
+      "step": 4784
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.0953286769780394e-05,
+      "loss": 0.0448,
+      "step": 4785
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.095001515713152e-05,
+      "loss": 0.042,
+      "step": 4786
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0946743441868947e-05,
+      "loss": 0.0405,
+      "step": 4787
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0943471624346058e-05,
+      "loss": 0.0433,
+      "step": 4788
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0940199704916256e-05,
+      "loss": 0.0392,
+      "step": 4789
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0936927683932942e-05,
+      "loss": 0.0386,
+      "step": 4790
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.093365556174954e-05,
+      "loss": 0.0408,
+      "step": 4791
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.0930383338719483e-05,
+      "loss": 0.0419,
+      "step": 4792
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0927111015196212e-05,
+      "loss": 0.0522,
+      "step": 4793
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0923838591533176e-05,
+      "loss": 0.0363,
+      "step": 4794
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.092056606808384e-05,
+      "loss": 0.0441,
+      "step": 4795
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.091729344520168e-05,
+      "loss": 0.0464,
+      "step": 4796
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0914020723240177e-05,
+      "loss": 0.0429,
+      "step": 4797
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0910747902552832e-05,
+      "loss": 0.0418,
+      "step": 4798
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.0907474983493144e-05,
+      "loss": 0.0422,
+      "step": 4799
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0904201966414636e-05,
+      "loss": 0.0401,
+      "step": 4800
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0900928851670831e-05,
+      "loss": 0.0433,
+      "step": 4801
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0897655639615272e-05,
+      "loss": 0.0462,
+      "step": 4802
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0894382330601502e-05,
+      "loss": 0.036,
+      "step": 4803
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.0891108924983087e-05,
+      "loss": 0.0449,
+      "step": 4804
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.088783542311359e-05,
+      "loss": 0.0504,
+      "step": 4805
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0884561825346589e-05,
+      "loss": 0.0451,
+      "step": 4806
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.088128813203568e-05,
+      "loss": 0.0423,
+      "step": 4807
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0878014343534459e-05,
+      "loss": 0.036,
+      "step": 4808
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0874740460196545e-05,
+      "loss": 0.0404,
+      "step": 4809
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0871466482375547e-05,
+      "loss": 0.0296,
+      "step": 4810
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.0868192410425107e-05,
+      "loss": 0.0414,
+      "step": 4811
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.086491824469886e-05,
+      "loss": 0.0469,
+      "step": 4812
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0861643985550456e-05,
+      "loss": 0.039,
+      "step": 4813
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0858369633333561e-05,
+      "loss": 0.0497,
+      "step": 4814
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0855095188401847e-05,
+      "loss": 0.0419,
+      "step": 4815
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0851820651108993e-05,
+      "loss": 0.0448,
+      "step": 4816
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.084854602180869e-05,
+      "loss": 0.0374,
+      "step": 4817
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0845271300854643e-05,
+      "loss": 0.044,
+      "step": 4818
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.0841996488600559e-05,
+      "loss": 0.0502,
+      "step": 4819
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0838721585400165e-05,
+      "loss": 0.0373,
+      "step": 4820
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0835446591607186e-05,
+      "loss": 0.0377,
+      "step": 4821
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0832171507575366e-05,
+      "loss": 0.0431,
+      "step": 4822
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0828896333658456e-05,
+      "loss": 0.046,
+      "step": 4823
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0825621070210215e-05,
+      "loss": 0.0475,
+      "step": 4824
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.0822345717584417e-05,
+      "loss": 0.0345,
+      "step": 4825
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0819070276134837e-05,
+      "loss": 0.0447,
+      "step": 4826
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0815794746215265e-05,
+      "loss": 0.0368,
+      "step": 4827
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0812519128179501e-05,
+      "loss": 0.0433,
+      "step": 4828
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0809243422381352e-05,
+      "loss": 0.0521,
+      "step": 4829
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0805967629174636e-05,
+      "loss": 0.0441,
+      "step": 4830
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0802691748913185e-05,
+      "loss": 0.0465,
+      "step": 4831
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.0799415781950828e-05,
+      "loss": 0.0419,
+      "step": 4832
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0796139728641419e-05,
+      "loss": 0.0376,
+      "step": 4833
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0792863589338804e-05,
+      "loss": 0.0402,
+      "step": 4834
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0789587364396856e-05,
+      "loss": 0.0496,
+      "step": 4835
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0786311054169446e-05,
+      "loss": 0.0349,
+      "step": 4836
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0783034659010454e-05,
+      "loss": 0.0434,
+      "step": 4837
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.0779758179273776e-05,
+      "loss": 0.0431,
+      "step": 4838
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0776481615313314e-05,
+      "loss": 0.0517,
+      "step": 4839
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0773204967482977e-05,
+      "loss": 0.0411,
+      "step": 4840
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0769928236136682e-05,
+      "loss": 0.0413,
+      "step": 4841
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0766651421628365e-05,
+      "loss": 0.0397,
+      "step": 4842
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0763374524311954e-05,
+      "loss": 0.0316,
+      "step": 4843
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.07600975445414e-05,
+      "loss": 0.0472,
+      "step": 4844
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.0756820482670663e-05,
+      "loss": 0.052,
+      "step": 4845
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.0753543339053698e-05,
+      "loss": 0.044,
+      "step": 4846
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.0750266114044487e-05,
+      "loss": 0.0473,
+      "step": 4847
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.074698880799701e-05,
+      "loss": 0.0336,
+      "step": 4848
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.074371142126525e-05,
+      "loss": 0.0452,
+      "step": 4849
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.0740433954203215e-05,
+      "loss": 0.0434,
+      "step": 4850
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.073715640716491e-05,
+      "loss": 0.0404,
+      "step": 4851
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.073387878050435e-05,
+      "loss": 0.0382,
+      "step": 4852
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0730601074575567e-05,
+      "loss": 0.044,
+      "step": 4853
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0727323289732586e-05,
+      "loss": 0.0428,
+      "step": 4854
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0724045426329454e-05,
+      "loss": 0.042,
+      "step": 4855
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0720767484720222e-05,
+      "loss": 0.0408,
+      "step": 4856
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0717489465258949e-05,
+      "loss": 0.0523,
+      "step": 4857
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.0714211368299704e-05,
+      "loss": 0.0362,
+      "step": 4858
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0710933194196558e-05,
+      "loss": 0.0464,
+      "step": 4859
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0707654943303598e-05,
+      "loss": 0.0356,
+      "step": 4860
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.070437661597492e-05,
+      "loss": 0.0461,
+      "step": 4861
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0701098212564625e-05,
+      "loss": 0.0553,
+      "step": 4862
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0697819733426815e-05,
+      "loss": 0.0398,
+      "step": 4863
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0694541178915616e-05,
+      "loss": 0.0389,
+      "step": 4864
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.0691262549385146e-05,
+      "loss": 0.0376,
+      "step": 4865
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0687983845189545e-05,
+      "loss": 0.0436,
+      "step": 4866
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.068470506668295e-05,
+      "loss": 0.0451,
+      "step": 4867
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0681426214219515e-05,
+      "loss": 0.045,
+      "step": 4868
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0678147288153396e-05,
+      "loss": 0.0431,
+      "step": 4869
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0674868288838758e-05,
+      "loss": 0.0485,
+      "step": 4870
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.0671589216629772e-05,
+      "loss": 0.039,
+      "step": 4871
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.066831007188062e-05,
+      "loss": 0.0413,
+      "step": 4872
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.0665030854945498e-05,
+      "loss": 0.0455,
+      "step": 4873
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.06617515661786e-05,
+      "loss": 0.0372,
+      "step": 4874
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.0658472205934124e-05,
+      "loss": 0.045,
+      "step": 4875
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.065519277456629e-05,
+      "loss": 0.0433,
+      "step": 4876
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.0651913272429317e-05,
+      "loss": 0.0474,
+      "step": 4877
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.064863369987743e-05,
+      "loss": 0.0395,
+      "step": 4878
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0645354057264867e-05,
+      "loss": 0.0416,
+      "step": 4879
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0642074344945873e-05,
+      "loss": 0.0459,
+      "step": 4880
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0638794563274695e-05,
+      "loss": 0.0457,
+      "step": 4881
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.063551471260559e-05,
+      "loss": 0.0463,
+      "step": 4882
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0632234793292826e-05,
+      "loss": 0.0507,
+      "step": 4883
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.0628954805690679e-05,
+      "loss": 0.0392,
+      "step": 4884
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0625674750153423e-05,
+      "loss": 0.0437,
+      "step": 4885
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0622394627035349e-05,
+      "loss": 0.0421,
+      "step": 4886
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0619114436690751e-05,
+      "loss": 0.0389,
+      "step": 4887
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0615834179473932e-05,
+      "loss": 0.0436,
+      "step": 4888
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0612553855739201e-05,
+      "loss": 0.0497,
+      "step": 4889
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0609273465840873e-05,
+      "loss": 0.0468,
+      "step": 4890
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.0605993010133278e-05,
+      "loss": 0.0574,
+      "step": 4891
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.0602712488970742e-05,
+      "loss": 0.0466,
+      "step": 4892
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.05994319027076e-05,
+      "loss": 0.0452,
+      "step": 4893
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.05961512516982e-05,
+      "loss": 0.0469,
+      "step": 4894
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.0592870536296896e-05,
+      "loss": 0.0502,
+      "step": 4895
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.0589589756858046e-05,
+      "loss": 0.035,
+      "step": 4896
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.0586308913736013e-05,
+      "loss": 0.0429,
+      "step": 4897
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0583028007285173e-05,
+      "loss": 0.0419,
+      "step": 4898
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0579747037859906e-05,
+      "loss": 0.0353,
+      "step": 4899
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0576466005814595e-05,
+      "loss": 0.0424,
+      "step": 4900
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0573184911503634e-05,
+      "loss": 0.0398,
+      "step": 4901
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0569903755281425e-05,
+      "loss": 0.0524,
+      "step": 4902
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.0566622537502372e-05,
+      "loss": 0.0471,
+      "step": 4903
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.056334125852089e-05,
+      "loss": 0.0446,
+      "step": 4904
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0560059918691396e-05,
+      "loss": 0.0382,
+      "step": 4905
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0556778518368324e-05,
+      "loss": 0.0381,
+      "step": 4906
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.05534970579061e-05,
+      "loss": 0.0642,
+      "step": 4907
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0550215537659163e-05,
+      "loss": 0.042,
+      "step": 4908
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.054693395798196e-05,
+      "loss": 0.0453,
+      "step": 4909
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0543652319228946e-05,
+      "loss": 0.0438,
+      "step": 4910
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.0540370621754575e-05,
+      "loss": 0.0418,
+      "step": 4911
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.0537088865913313e-05,
+      "loss": 0.0382,
+      "step": 4912
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.053380705205964e-05,
+      "loss": 0.0454,
+      "step": 4913
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.0530525180548024e-05,
+      "loss": 0.038,
+      "step": 4914
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.052724325173295e-05,
+      "loss": 0.0392,
+      "step": 4915
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.0523961265968909e-05,
+      "loss": 0.0373,
+      "step": 4916
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.0520679223610398e-05,
+      "loss": 0.048,
+      "step": 4917
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0517397125011921e-05,
+      "loss": 0.0454,
+      "step": 4918
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0514114970527979e-05,
+      "loss": 0.0407,
+      "step": 4919
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0510832760513097e-05,
+      "loss": 0.046,
+      "step": 4920
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0507550495321788e-05,
+      "loss": 0.0472,
+      "step": 4921
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0504268175308579e-05,
+      "loss": 0.039,
+      "step": 4922
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.0500985800828005e-05,
+      "loss": 0.0374,
+      "step": 4923
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.04977033722346e-05,
+      "loss": 0.0534,
+      "step": 4924
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.049442088988291e-05,
+      "loss": 0.0404,
+      "step": 4925
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0491138354127488e-05,
+      "loss": 0.0461,
+      "step": 4926
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0487855765322886e-05,
+      "loss": 0.0445,
+      "step": 4927
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0484573123823666e-05,
+      "loss": 0.0403,
+      "step": 4928
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0481290429984397e-05,
+      "loss": 0.0421,
+      "step": 4929
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.0478007684159646e-05,
+      "loss": 0.0458,
+      "step": 4930
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0474724886703995e-05,
+      "loss": 0.047,
+      "step": 4931
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0471442037972029e-05,
+      "loss": 0.044,
+      "step": 4932
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0468159138318336e-05,
+      "loss": 0.0514,
+      "step": 4933
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0464876188097513e-05,
+      "loss": 0.0454,
+      "step": 4934
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0461593187664153e-05,
+      "loss": 0.0386,
+      "step": 4935
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0458310137372874e-05,
+      "loss": 0.0408,
+      "step": 4936
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.0455027037578276e-05,
+      "loss": 0.0471,
+      "step": 4937
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.045174388863498e-05,
+      "loss": 0.0531,
+      "step": 4938
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0448460690897606e-05,
+      "loss": 0.0383,
+      "step": 4939
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0445177444720785e-05,
+      "loss": 0.0399,
+      "step": 4940
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0441894150459144e-05,
+      "loss": 0.0465,
+      "step": 4941
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.0438610808467325e-05,
+      "loss": 0.0439,
+      "step": 4942
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.043532741909997e-05,
+      "loss": 0.0533,
+      "step": 4943
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0432043982711722e-05,
+      "loss": 0.042,
+      "step": 4944
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.042876049965724e-05,
+      "loss": 0.0435,
+      "step": 4945
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0425476970291177e-05,
+      "loss": 0.05,
+      "step": 4946
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0422193394968198e-05,
+      "loss": 0.0371,
+      "step": 4947
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0418909774042974e-05,
+      "loss": 0.036,
+      "step": 4948
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0415626107870171e-05,
+      "loss": 0.047,
+      "step": 4949
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.0412342396804473e-05,
+      "loss": 0.0453,
+      "step": 4950
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0409058641200562e-05,
+      "loss": 0.039,
+      "step": 4951
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.040577484141312e-05,
+      "loss": 0.0433,
+      "step": 4952
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0402490997796841e-05,
+      "loss": 0.042,
+      "step": 4953
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0399207110706425e-05,
+      "loss": 0.0505,
+      "step": 4954
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0395923180496568e-05,
+      "loss": 0.0435,
+      "step": 4955
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0392639207521984e-05,
+      "loss": 0.0526,
+      "step": 4956
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.0389355192137379e-05,
+      "loss": 0.0482,
+      "step": 4957
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0386071134697469e-05,
+      "loss": 0.0372,
+      "step": 4958
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0382787035556972e-05,
+      "loss": 0.0458,
+      "step": 4959
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0379502895070615e-05,
+      "loss": 0.046,
+      "step": 4960
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0376218713593125e-05,
+      "loss": 0.0451,
+      "step": 4961
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0372934491479238e-05,
+      "loss": 0.0418,
+      "step": 4962
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.0369650229083687e-05,
+      "loss": 0.042,
+      "step": 4963
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.036636592676122e-05,
+      "loss": 0.0358,
+      "step": 4964
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.036308158486658e-05,
+      "loss": 0.0494,
+      "step": 4965
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0359797203754518e-05,
+      "loss": 0.0372,
+      "step": 4966
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.035651278377979e-05,
+      "loss": 0.0437,
+      "step": 4967
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0353228325297151e-05,
+      "loss": 0.0422,
+      "step": 4968
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0349943828661368e-05,
+      "loss": 0.0432,
+      "step": 4969
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.0346659294227208e-05,
+      "loss": 0.0466,
+      "step": 4970
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0343374722349442e-05,
+      "loss": 0.0381,
+      "step": 4971
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0340090113382848e-05,
+      "loss": 0.0467,
+      "step": 4972
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0336805467682202e-05,
+      "loss": 0.0423,
+      "step": 4973
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0333520785602288e-05,
+      "loss": 0.052,
+      "step": 4974
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0330236067497892e-05,
+      "loss": 0.0453,
+      "step": 4975
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.0326951313723806e-05,
+      "loss": 0.0382,
+      "step": 4976
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0323666524634828e-05,
+      "loss": 0.0432,
+      "step": 4977
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0320381700585758e-05,
+      "loss": 0.0392,
+      "step": 4978
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.031709684193139e-05,
+      "loss": 0.0419,
+      "step": 4979
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0313811949026542e-05,
+      "loss": 0.04,
+      "step": 4980
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0310527022226015e-05,
+      "loss": 0.0424,
+      "step": 4981
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0307242061884625e-05,
+      "loss": 0.0394,
+      "step": 4982
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.0303957068357196e-05,
+      "loss": 0.0381,
+      "step": 4983
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.030067204199854e-05,
+      "loss": 0.0529,
+      "step": 4984
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0297386983163484e-05,
+      "loss": 0.0394,
+      "step": 4985
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0294101892206859e-05,
+      "loss": 0.0497,
+      "step": 4986
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0290816769483497e-05,
+      "loss": 0.0445,
+      "step": 4987
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0287531615348226e-05,
+      "loss": 0.0447,
+      "step": 4988
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.0284246430155894e-05,
+      "loss": 0.0484,
+      "step": 4989
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0280961214261333e-05,
+      "loss": 0.0351,
+      "step": 4990
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0277675968019396e-05,
+      "loss": 0.0397,
+      "step": 4991
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0274390691784927e-05,
+      "loss": 0.0373,
+      "step": 4992
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0271105385912779e-05,
+      "loss": 0.0368,
+      "step": 4993
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0267820050757809e-05,
+      "loss": 0.0423,
+      "step": 4994
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0264534686674871e-05,
+      "loss": 0.0466,
+      "step": 4995
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.0261249294018827e-05,
+      "loss": 0.0405,
+      "step": 4996
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0257963873144542e-05,
+      "loss": 0.0384,
+      "step": 4997
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0254678424406882e-05,
+      "loss": 0.0402,
+      "step": 4998
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0251392948160717e-05,
+      "loss": 0.0574,
+      "step": 4999
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0248107444760924e-05,
+      "loss": 0.0485,
+      "step": 5000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0244821914562376e-05,
+      "loss": 0.0366,
+      "step": 5001
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0241536357919952e-05,
+      "loss": 0.0407,
+      "step": 5002
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.0238250775188532e-05,
+      "loss": 0.043,
+      "step": 5003
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0234965166723003e-05,
+      "loss": 0.0424,
+      "step": 5004
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0231679532878256e-05,
+      "loss": 0.0376,
+      "step": 5005
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0228393874009172e-05,
+      "loss": 0.0415,
+      "step": 5006
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0225108190470652e-05,
+      "loss": 0.0484,
+      "step": 5007
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0221822482617587e-05,
+      "loss": 0.0492,
+      "step": 5008
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.0218536750804882e-05,
+      "loss": 0.0428,
+      "step": 5009
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0215250995387427e-05,
+      "loss": 0.0414,
+      "step": 5010
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0211965216720136e-05,
+      "loss": 0.0389,
+      "step": 5011
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0208679415157907e-05,
+      "loss": 0.0501,
+      "step": 5012
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.020539359105565e-05,
+      "loss": 0.047,
+      "step": 5013
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0202107744768277e-05,
+      "loss": 0.047,
+      "step": 5014
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.0198821876650702e-05,
+      "loss": 0.0405,
+      "step": 5015
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.019553598705784e-05,
+      "loss": 0.0411,
+      "step": 5016
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.019225007634461e-05,
+      "loss": 0.0523,
+      "step": 5017
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0188964144865927e-05,
+      "loss": 0.0485,
+      "step": 5018
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0185678192976717e-05,
+      "loss": 0.0473,
+      "step": 5019
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0182392221031905e-05,
+      "loss": 0.0444,
+      "step": 5020
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0179106229386417e-05,
+      "loss": 0.0444,
+      "step": 5021
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.0175820218395182e-05,
+      "loss": 0.0505,
+      "step": 5022
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0172534188413132e-05,
+      "loss": 0.0499,
+      "step": 5023
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0169248139795197e-05,
+      "loss": 0.0421,
+      "step": 5024
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0165962072896314e-05,
+      "loss": 0.0465,
+      "step": 5025
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0162675988071419e-05,
+      "loss": 0.0468,
+      "step": 5026
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0159389885675452e-05,
+      "loss": 0.0513,
+      "step": 5027
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0156103766063354e-05,
+      "loss": 0.0438,
+      "step": 5028
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.0152817629590066e-05,
+      "loss": 0.0433,
+      "step": 5029
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0149531476610533e-05,
+      "loss": 0.0401,
+      "step": 5030
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0146245307479703e-05,
+      "loss": 0.049,
+      "step": 5031
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0142959122552522e-05,
+      "loss": 0.0495,
+      "step": 5032
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.013967292218394e-05,
+      "loss": 0.0507,
+      "step": 5033
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0136386706728909e-05,
+      "loss": 0.0357,
+      "step": 5034
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.0133100476542379e-05,
+      "loss": 0.0459,
+      "step": 5035
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.012981423197931e-05,
+      "loss": 0.0461,
+      "step": 5036
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0126527973394653e-05,
+      "loss": 0.0393,
+      "step": 5037
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0123241701143374e-05,
+      "loss": 0.0396,
+      "step": 5038
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0119955415580422e-05,
+      "loss": 0.044,
+      "step": 5039
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0116669117060762e-05,
+      "loss": 0.0537,
+      "step": 5040
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0113382805939355e-05,
+      "loss": 0.0471,
+      "step": 5041
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.0110096482571166e-05,
+      "loss": 0.0364,
+      "step": 5042
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0106810147311161e-05,
+      "loss": 0.0411,
+      "step": 5043
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0103523800514301e-05,
+      "loss": 0.0479,
+      "step": 5044
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0100237442535555e-05,
+      "loss": 0.035,
+      "step": 5045
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0096951073729894e-05,
+      "loss": 0.0482,
+      "step": 5046
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0093664694452286e-05,
+      "loss": 0.0416,
+      "step": 5047
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0090378305057702e-05,
+      "loss": 0.0393,
+      "step": 5048
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.0087091905901114e-05,
+      "loss": 0.0449,
+      "step": 5049
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.008380549733749e-05,
+      "loss": 0.0406,
+      "step": 5050
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.008051907972181e-05,
+      "loss": 0.0424,
+      "step": 5051
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0077232653409049e-05,
+      "loss": 0.0368,
+      "step": 5052
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0073946218754181e-05,
+      "loss": 0.0351,
+      "step": 5053
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0070659776112179e-05,
+      "loss": 0.0476,
+      "step": 5054
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.0067373325838026e-05,
+      "loss": 0.043,
+      "step": 5055
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0064086868286697e-05,
+      "loss": 0.0381,
+      "step": 5056
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0060800403813173e-05,
+      "loss": 0.0409,
+      "step": 5057
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0057513932772433e-05,
+      "loss": 0.0414,
+      "step": 5058
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0054227455519459e-05,
+      "loss": 0.0466,
+      "step": 5059
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0050940972409231e-05,
+      "loss": 0.0469,
+      "step": 5060
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.0047654483796732e-05,
+      "loss": 0.048,
+      "step": 5061
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.004436799003694e-05,
+      "loss": 0.0544,
+      "step": 5062
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0041081491484843e-05,
+      "loss": 0.0396,
+      "step": 5063
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0037794988495424e-05,
+      "loss": 0.045,
+      "step": 5064
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0034508481423668e-05,
+      "loss": 0.0373,
+      "step": 5065
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0031221970624557e-05,
+      "loss": 0.0437,
+      "step": 5066
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.0027935456453072e-05,
+      "loss": 0.0446,
+      "step": 5067
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.002464893926421e-05,
+      "loss": 0.043,
+      "step": 5068
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0021362419412946e-05,
+      "loss": 0.0459,
+      "step": 5069
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0018075897254269e-05,
+      "loss": 0.0417,
+      "step": 5070
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0014789373143168e-05,
+      "loss": 0.0407,
+      "step": 5071
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0011502847434625e-05,
+      "loss": 0.0469,
+      "step": 5072
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0008216320483625e-05,
+      "loss": 0.0407,
+      "step": 5073
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.0004929792645162e-05,
+      "loss": 0.0481,
+      "step": 5074
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.000164326427422e-05,
+      "loss": 0.0463,
+      "step": 5075
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.998356735725783e-06,
+      "loss": 0.044,
+      "step": 5076
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.995070207354838e-06,
+      "loss": 0.0503,
+      "step": 5077
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.991783679516376e-06,
+      "loss": 0.0532,
+      "step": 5078
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.988497152565382e-06,
+      "loss": 0.0389,
+      "step": 5079
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.985210626856836e-06,
+      "loss": 0.0438,
+      "step": 5080
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.981924102745734e-06,
+      "loss": 0.0433,
+      "step": 5081
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.978637580587057e-06,
+      "loss": 0.0427,
+      "step": 5082
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.975351060735793e-06,
+      "loss": 0.0542,
+      "step": 5083
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.97206454354693e-06,
+      "loss": 0.0362,
+      "step": 5084
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.968778029375448e-06,
+      "loss": 0.0462,
+      "step": 5085
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.965491518576337e-06,
+      "loss": 0.0431,
+      "step": 5086
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.962205011504577e-06,
+      "loss": 0.0439,
+      "step": 5087
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.958918508515159e-06,
+      "loss": 0.0472,
+      "step": 5088
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.955632009963062e-06,
+      "loss": 0.0442,
+      "step": 5089
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.952345516203272e-06,
+      "loss": 0.0425,
+      "step": 5090
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.949059027590772e-06,
+      "loss": 0.0476,
+      "step": 5091
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.945772544480543e-06,
+      "loss": 0.0504,
+      "step": 5092
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.94248606722757e-06,
+      "loss": 0.0473,
+      "step": 5093
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.939199596186827e-06,
+      "loss": 0.051,
+      "step": 5094
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.935913131713305e-06,
+      "loss": 0.0499,
+      "step": 5095
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.932626674161978e-06,
+      "loss": 0.0533,
+      "step": 5096
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.929340223887824e-06,
+      "loss": 0.0487,
+      "step": 5097
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.926053781245824e-06,
+      "loss": 0.0363,
+      "step": 5098
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.922767346590953e-06,
+      "loss": 0.0401,
+      "step": 5099
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.919480920278191e-06,
+      "loss": 0.0433,
+      "step": 5100
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.916194502662514e-06,
+      "loss": 0.0535,
+      "step": 5101
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.91290809409889e-06,
+      "loss": 0.039,
+      "step": 5102
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.909621694942301e-06,
+      "loss": 0.0449,
+      "step": 5103
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.906335305547716e-06,
+      "loss": 0.0465,
+      "step": 5104
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.903048926270107e-06,
+      "loss": 0.0352,
+      "step": 5105
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.899762557464447e-06,
+      "loss": 0.0414,
+      "step": 5106
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.896476199485702e-06,
+      "loss": 0.039,
+      "step": 5107
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.893189852688844e-06,
+      "loss": 0.0406,
+      "step": 5108
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.889903517428836e-06,
+      "loss": 0.0396,
+      "step": 5109
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.886617194060647e-06,
+      "loss": 0.0431,
+      "step": 5110
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.883330882939243e-06,
+      "loss": 0.0533,
+      "step": 5111
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.88004458441958e-06,
+      "loss": 0.0418,
+      "step": 5112
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.87675829885663e-06,
+      "loss": 0.0366,
+      "step": 5113
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.873472026605347e-06,
+      "loss": 0.0442,
+      "step": 5114
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.870185768020694e-06,
+      "loss": 0.0508,
+      "step": 5115
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.866899523457621e-06,
+      "loss": 0.0434,
+      "step": 5116
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.863613293271094e-06,
+      "loss": 0.0478,
+      "step": 5117
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.860327077816065e-06,
+      "loss": 0.0416,
+      "step": 5118
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.85704087744748e-06,
+      "loss": 0.0457,
+      "step": 5119
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.8537546925203e-06,
+      "loss": 0.0467,
+      "step": 5120
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.850468523389469e-06,
+      "loss": 0.0526,
+      "step": 5121
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.847182370409937e-06,
+      "loss": 0.0466,
+      "step": 5122
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.843896233936653e-06,
+      "loss": 0.0523,
+      "step": 5123
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.84061011432455e-06,
+      "loss": 0.0398,
+      "step": 5124
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.837324011928584e-06,
+      "loss": 0.0435,
+      "step": 5125
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.834037927103689e-06,
+      "loss": 0.0378,
+      "step": 5126
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.830751860204804e-06,
+      "loss": 0.0442,
+      "step": 5127
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.827465811586871e-06,
+      "loss": 0.0423,
+      "step": 5128
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.82417978160482e-06,
+      "loss": 0.0398,
+      "step": 5129
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.820893770613586e-06,
+      "loss": 0.0506,
+      "step": 5130
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.817607778968095e-06,
+      "loss": 0.0503,
+      "step": 5131
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.814321807023284e-06,
+      "loss": 0.0421,
+      "step": 5132
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.811035855134078e-06,
+      "loss": 0.0438,
+      "step": 5133
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.807749923655394e-06,
+      "loss": 0.0437,
+      "step": 5134
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.804464012942163e-06,
+      "loss": 0.0564,
+      "step": 5135
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.801178123349298e-06,
+      "loss": 0.0421,
+      "step": 5136
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.797892255231726e-06,
+      "loss": 0.0497,
+      "step": 5137
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.79460640894435e-06,
+      "loss": 0.0469,
+      "step": 5138
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.791320584842096e-06,
+      "loss": 0.0437,
+      "step": 5139
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.788034783279869e-06,
+      "loss": 0.0424,
+      "step": 5140
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.784749004612573e-06,
+      "loss": 0.0403,
+      "step": 5141
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.781463249195121e-06,
+      "loss": 0.0449,
+      "step": 5142
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.778177517382413e-06,
+      "loss": 0.039,
+      "step": 5143
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.774891809529351e-06,
+      "loss": 0.0477,
+      "step": 5144
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.771606125990833e-06,
+      "loss": 0.0461,
+      "step": 5145
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.768320467121747e-06,
+      "loss": 0.0392,
+      "step": 5146
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.765034833277e-06,
+      "loss": 0.0506,
+      "step": 5147
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.76174922481147e-06,
+      "loss": 0.0503,
+      "step": 5148
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.758463642080051e-06,
+      "loss": 0.0406,
+      "step": 5149
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.755178085437627e-06,
+      "loss": 0.05,
+      "step": 5150
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.75189255523908e-06,
+      "loss": 0.0498,
+      "step": 5151
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.748607051839285e-06,
+      "loss": 0.0422,
+      "step": 5152
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.74532157559312e-06,
+      "loss": 0.0521,
+      "step": 5153
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.74203612685546e-06,
+      "loss": 0.0386,
+      "step": 5154
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.738750705981178e-06,
+      "loss": 0.0461,
+      "step": 5155
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.73546531332513e-06,
+      "loss": 0.0426,
+      "step": 5156
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.732179949242194e-06,
+      "loss": 0.0458,
+      "step": 5157
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.728894614087221e-06,
+      "loss": 0.0437,
+      "step": 5158
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.725609308215076e-06,
+      "loss": 0.0519,
+      "step": 5159
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.722324031980609e-06,
+      "loss": 0.0423,
+      "step": 5160
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.719038785738669e-06,
+      "loss": 0.045,
+      "step": 5161
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.715753569844111e-06,
+      "loss": 0.0451,
+      "step": 5162
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.712468384651774e-06,
+      "loss": 0.043,
+      "step": 5163
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.709183230516508e-06,
+      "loss": 0.049,
+      "step": 5164
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.705898107793143e-06,
+      "loss": 0.0418,
+      "step": 5165
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.702613016836518e-06,
+      "loss": 0.0403,
+      "step": 5166
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.699327958001466e-06,
+      "loss": 0.0403,
+      "step": 5167
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.696042931642807e-06,
+      "loss": 0.039,
+      "step": 5168
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.692757938115376e-06,
+      "loss": 0.0447,
+      "step": 5169
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.689472977773986e-06,
+      "loss": 0.0417,
+      "step": 5170
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.686188050973461e-06,
+      "loss": 0.0467,
+      "step": 5171
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.682903158068611e-06,
+      "loss": 0.0531,
+      "step": 5172
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.679618299414247e-06,
+      "loss": 0.0513,
+      "step": 5173
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.676333475365176e-06,
+      "loss": 0.0439,
+      "step": 5174
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.673048686276194e-06,
+      "loss": 0.0389,
+      "step": 5175
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.669763932502111e-06,
+      "loss": 0.0465,
+      "step": 5176
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.666479214397717e-06,
+      "loss": 0.0413,
+      "step": 5177
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.663194532317801e-06,
+      "loss": 0.0485,
+      "step": 5178
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.659909886617155e-06,
+      "loss": 0.0509,
+      "step": 5179
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.656625277650558e-06,
+      "loss": 0.0439,
+      "step": 5180
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.653340705772794e-06,
+      "loss": 0.047,
+      "step": 5181
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.650056171338637e-06,
+      "loss": 0.0403,
+      "step": 5182
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.646771674702852e-06,
+      "loss": 0.0476,
+      "step": 5183
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.643487216220216e-06,
+      "loss": 0.0454,
+      "step": 5184
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.640202796245484e-06,
+      "loss": 0.044,
+      "step": 5185
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.636918415133423e-06,
+      "loss": 0.046,
+      "step": 5186
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.633634073238782e-06,
+      "loss": 0.0521,
+      "step": 5187
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.630349770916316e-06,
+      "loss": 0.0514,
+      "step": 5188
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.627065508520767e-06,
+      "loss": 0.0469,
+      "step": 5189
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.623781286406877e-06,
+      "loss": 0.0419,
+      "step": 5190
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.62049710492939e-06,
+      "loss": 0.0393,
+      "step": 5191
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.61721296444303e-06,
+      "loss": 0.0391,
+      "step": 5192
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.613928865302534e-06,
+      "loss": 0.0491,
+      "step": 5193
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.610644807862625e-06,
+      "loss": 0.0493,
+      "step": 5194
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.607360792478018e-06,
+      "loss": 0.048,
+      "step": 5195
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.604076819503434e-06,
+      "loss": 0.0526,
+      "step": 5196
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.600792889293577e-06,
+      "loss": 0.0449,
+      "step": 5197
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.597509002203162e-06,
+      "loss": 0.0505,
+      "step": 5198
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.594225158586886e-06,
+      "loss": 0.0407,
+      "step": 5199
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.590941358799442e-06,
+      "loss": 0.0476,
+      "step": 5200
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.587657603195528e-06,
+      "loss": 0.0454,
+      "step": 5201
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.584373892129829e-06,
+      "loss": 0.0488,
+      "step": 5202
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.58109022595703e-06,
+      "loss": 0.0512,
+      "step": 5203
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.577806605031805e-06,
+      "loss": 0.0433,
+      "step": 5204
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.574523029708826e-06,
+      "loss": 0.0477,
+      "step": 5205
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.571239500342764e-06,
+      "loss": 0.0458,
+      "step": 5206
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.56795601728828e-06,
+      "loss": 0.0438,
+      "step": 5207
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.564672580900034e-06,
+      "loss": 0.0411,
+      "step": 5208
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.561389191532679e-06,
+      "loss": 0.0439,
+      "step": 5209
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.558105849540859e-06,
+      "loss": 0.0495,
+      "step": 5210
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.55482255527922e-06,
+      "loss": 0.0427,
+      "step": 5211
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.551539309102395e-06,
+      "loss": 0.0429,
+      "step": 5212
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.548256111365025e-06,
+      "loss": 0.0546,
+      "step": 5213
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.544972962421724e-06,
+      "loss": 0.0421,
+      "step": 5214
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.541689862627129e-06,
+      "loss": 0.0404,
+      "step": 5215
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.538406812335848e-06,
+      "loss": 0.0448,
+      "step": 5216
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.53512381190249e-06,
+      "loss": 0.046,
+      "step": 5217
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.531840861681667e-06,
+      "loss": 0.0529,
+      "step": 5218
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.528557962027971e-06,
+      "loss": 0.0475,
+      "step": 5219
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.525275113296007e-06,
+      "loss": 0.0483,
+      "step": 5220
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.521992315840359e-06,
+      "loss": 0.0404,
+      "step": 5221
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.518709570015606e-06,
+      "loss": 0.0502,
+      "step": 5222
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.515426876176337e-06,
+      "loss": 0.0513,
+      "step": 5223
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.512144234677116e-06,
+      "loss": 0.0437,
+      "step": 5224
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.508861645872515e-06,
+      "loss": 0.0523,
+      "step": 5225
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.505579110117094e-06,
+      "loss": 0.0523,
+      "step": 5226
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.502296627765402e-06,
+      "loss": 0.0507,
+      "step": 5227
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.499014199172001e-06,
+      "loss": 0.0449,
+      "step": 5228
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.495731824691423e-06,
+      "loss": 0.0448,
+      "step": 5229
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.492449504678215e-06,
+      "loss": 0.0567,
+      "step": 5230
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.489167239486908e-06,
+      "loss": 0.0429,
+      "step": 5231
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.485885029472023e-06,
+      "loss": 0.0506,
+      "step": 5232
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.482602874988086e-06,
+      "loss": 0.0416,
+      "step": 5233
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.479320776389602e-06,
+      "loss": 0.0474,
+      "step": 5234
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.476038734031093e-06,
+      "loss": 0.0437,
+      "step": 5235
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.47275674826705e-06,
+      "loss": 0.0505,
+      "step": 5236
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.469474819451978e-06,
+      "loss": 0.0431,
+      "step": 5237
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.466192947940362e-06,
+      "loss": 0.0488,
+      "step": 5238
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.462911134086685e-06,
+      "loss": 0.0558,
+      "step": 5239
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.459629378245428e-06,
+      "loss": 0.0421,
+      "step": 5240
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.456347680771056e-06,
+      "loss": 0.0436,
+      "step": 5241
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.453066042018043e-06,
+      "loss": 0.0417,
+      "step": 5242
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.449784462340844e-06,
+      "loss": 0.0562,
+      "step": 5243
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.446502942093902e-06,
+      "loss": 0.0457,
+      "step": 5244
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.443221481631679e-06,
+      "loss": 0.0516,
+      "step": 5245
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 9.439940081308602e-06,
+      "loss": 0.0488,
+      "step": 5246
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.436658741479114e-06,
+      "loss": 0.0446,
+      "step": 5247
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.433377462497632e-06,
+      "loss": 0.0465,
+      "step": 5248
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.430096244718578e-06,
+      "loss": 0.0497,
+      "step": 5249
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.42681508849637e-06,
+      "loss": 0.0508,
+      "step": 5250
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.423533994185407e-06,
+      "loss": 0.0494,
+      "step": 5251
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.420252962140097e-06,
+      "loss": 0.0434,
+      "step": 5252
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.416971992714829e-06,
+      "loss": 0.0488,
+      "step": 5253
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.41369108626399e-06,
+      "loss": 0.0407,
+      "step": 5254
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.410410243141959e-06,
+      "loss": 0.0506,
+      "step": 5255
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.407129463703106e-06,
+      "loss": 0.0453,
+      "step": 5256
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.403848748301802e-06,
+      "loss": 0.0294,
+      "step": 5257
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.400568097292407e-06,
+      "loss": 0.0275,
+      "step": 5258
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.397287511029263e-06,
+      "loss": 0.0362,
+      "step": 5259
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.394006989866726e-06,
+      "loss": 0.0354,
+      "step": 5260
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.390726534159127e-06,
+      "loss": 0.0299,
+      "step": 5261
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.387446144260804e-06,
+      "loss": 0.029,
+      "step": 5262
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.38416582052607e-06,
+      "loss": 0.0379,
+      "step": 5263
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.380885563309252e-06,
+      "loss": 0.0285,
+      "step": 5264
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.377605372964655e-06,
+      "loss": 0.0316,
+      "step": 5265
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.374325249846579e-06,
+      "loss": 0.0283,
+      "step": 5266
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.371045194309325e-06,
+      "loss": 0.0375,
+      "step": 5267
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.367765206707174e-06,
+      "loss": 0.0307,
+      "step": 5268
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.364485287394414e-06,
+      "loss": 0.025,
+      "step": 5269
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.361205436725312e-06,
+      "loss": 0.028,
+      "step": 5270
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.357925655054128e-06,
+      "loss": 0.0317,
+      "step": 5271
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.354645942735134e-06,
+      "loss": 0.0403,
+      "step": 5272
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.351366300122569e-06,
+      "loss": 0.0341,
+      "step": 5273
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.348086727570686e-06,
+      "loss": 0.0276,
+      "step": 5274
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.344807225433712e-06,
+      "loss": 0.024,
+      "step": 5275
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.341527794065879e-06,
+      "loss": 0.0275,
+      "step": 5276
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.338248433821406e-06,
+      "loss": 0.0265,
+      "step": 5277
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.334969145054502e-06,
+      "loss": 0.0282,
+      "step": 5278
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.331689928119382e-06,
+      "loss": 0.0297,
+      "step": 5279
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.328410783370235e-06,
+      "loss": 0.0355,
+      "step": 5280
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.325131711161246e-06,
+      "loss": 0.0291,
+      "step": 5281
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.32185271184661e-06,
+      "loss": 0.0317,
+      "step": 5282
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.318573785780487e-06,
+      "loss": 0.0358,
+      "step": 5283
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.315294933317053e-06,
+      "loss": 0.0303,
+      "step": 5284
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.312016154810455e-06,
+      "loss": 0.0285,
+      "step": 5285
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.308737450614856e-06,
+      "loss": 0.0306,
+      "step": 5286
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.30545882108439e-06,
+      "loss": 0.026,
+      "step": 5287
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.302180266573185e-06,
+      "loss": 0.0319,
+      "step": 5288
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.29890178743538e-06,
+      "loss": 0.0306,
+      "step": 5289
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.295623384025081e-06,
+      "loss": 0.0384,
+      "step": 5290
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.292345056696403e-06,
+      "loss": 0.0287,
+      "step": 5291
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.289066805803448e-06,
+      "loss": 0.0286,
+      "step": 5292
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.2857886317003e-06,
+      "loss": 0.0255,
+      "step": 5293
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.282510534741053e-06,
+      "loss": 0.0324,
+      "step": 5294
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.279232515279778e-06,
+      "loss": 0.0301,
+      "step": 5295
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.275954573670548e-06,
+      "loss": 0.0302,
+      "step": 5296
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.272676710267416e-06,
+      "loss": 0.0289,
+      "step": 5297
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.269398925424436e-06,
+      "loss": 0.0288,
+      "step": 5298
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.266121219495652e-06,
+      "loss": 0.0294,
+      "step": 5299
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.262843592835092e-06,
+      "loss": 0.0275,
+      "step": 5300
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.259566045796787e-06,
+      "loss": 0.028,
+      "step": 5301
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.256288578734753e-06,
+      "loss": 0.0267,
+      "step": 5302
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.253011192002994e-06,
+      "loss": 0.0294,
+      "step": 5303
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.249733885955515e-06,
+      "loss": 0.0318,
+      "step": 5304
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.246456660946302e-06,
+      "loss": 0.026,
+      "step": 5305
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.24317951732934e-06,
+      "loss": 0.0265,
+      "step": 5306
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.239902455458604e-06,
+      "loss": 0.0301,
+      "step": 5307
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.236625475688047e-06,
+      "loss": 0.0272,
+      "step": 5308
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.23334857837164e-06,
+      "loss": 0.0305,
+      "step": 5309
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.230071763863318e-06,
+      "loss": 0.032,
+      "step": 5310
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.226795032517026e-06,
+      "loss": 0.0294,
+      "step": 5311
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.223518384686688e-06,
+      "loss": 0.031,
+      "step": 5312
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.220241820726226e-06,
+      "loss": 0.0312,
+      "step": 5313
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.21696534098955e-06,
+      "loss": 0.0285,
+      "step": 5314
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.213688945830557e-06,
+      "loss": 0.0308,
+      "step": 5315
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.210412635603148e-06,
+      "loss": 0.0385,
+      "step": 5316
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.207136410661196e-06,
+      "loss": 0.0289,
+      "step": 5317
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.203860271358584e-06,
+      "loss": 0.0292,
+      "step": 5318
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.200584218049175e-06,
+      "loss": 0.027,
+      "step": 5319
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.197308251086818e-06,
+      "loss": 0.0294,
+      "step": 5320
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.194032370825366e-06,
+      "loss": 0.0298,
+      "step": 5321
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.19075657761865e-06,
+      "loss": 0.0277,
+      "step": 5322
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.187480871820502e-06,
+      "loss": 0.0284,
+      "step": 5323
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.18420525378474e-06,
+      "loss": 0.0296,
+      "step": 5324
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.180929723865165e-06,
+      "loss": 0.0279,
+      "step": 5325
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.177654282415585e-06,
+      "loss": 0.0269,
+      "step": 5326
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.174378929789785e-06,
+      "loss": 0.0275,
+      "step": 5327
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.171103666341547e-06,
+      "loss": 0.0305,
+      "step": 5328
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.167828492424639e-06,
+      "loss": 0.0361,
+      "step": 5329
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.164553408392816e-06,
+      "loss": 0.0427,
+      "step": 5330
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.16127841459984e-06,
+      "loss": 0.0285,
+      "step": 5331
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.158003511399441e-06,
+      "loss": 0.0284,
+      "step": 5332
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.154728699145359e-06,
+      "loss": 0.0299,
+      "step": 5333
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.151453978191312e-06,
+      "loss": 0.0311,
+      "step": 5334
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.148179348891012e-06,
+      "loss": 0.0298,
+      "step": 5335
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.144904811598158e-06,
+      "loss": 0.0324,
+      "step": 5336
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.141630366666439e-06,
+      "loss": 0.0235,
+      "step": 5337
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.138356014449547e-06,
+      "loss": 0.0278,
+      "step": 5338
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.135081755301142e-06,
+      "loss": 0.0232,
+      "step": 5339
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.131807589574896e-06,
+      "loss": 0.0347,
+      "step": 5340
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.128533517624454e-06,
+      "loss": 0.0304,
+      "step": 5341
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.125259539803459e-06,
+      "loss": 0.0305,
+      "step": 5342
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.121985656465543e-06,
+      "loss": 0.0351,
+      "step": 5343
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.118711867964322e-06,
+      "loss": 0.0263,
+      "step": 5344
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.115438174653415e-06,
+      "loss": 0.0343,
+      "step": 5345
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.112164576886417e-06,
+      "loss": 0.0297,
+      "step": 5346
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.108891075016916e-06,
+      "loss": 0.0266,
+      "step": 5347
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.1056176693985e-06,
+      "loss": 0.0305,
+      "step": 5348
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.102344360384731e-06,
+      "loss": 0.0274,
+      "step": 5349
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.099071148329172e-06,
+      "loss": 0.0269,
+      "step": 5350
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.09579803358537e-06,
+      "loss": 0.0346,
+      "step": 5351
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.092525016506858e-06,
+      "loss": 0.0436,
+      "step": 5352
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.089252097447173e-06,
+      "loss": 0.0362,
+      "step": 5353
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.085979276759823e-06,
+      "loss": 0.0292,
+      "step": 5354
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.082706554798324e-06,
+      "loss": 0.0293,
+      "step": 5355
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.07943393191616e-06,
+      "loss": 0.0289,
+      "step": 5356
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.076161408466827e-06,
+      "loss": 0.0276,
+      "step": 5357
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.072888984803793e-06,
+      "loss": 0.0378,
+      "step": 5358
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.069616661280517e-06,
+      "loss": 0.0268,
+      "step": 5359
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.066344438250461e-06,
+      "loss": 0.0273,
+      "step": 5360
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.063072316067058e-06,
+      "loss": 0.0349,
+      "step": 5361
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.059800295083747e-06,
+      "loss": 0.0264,
+      "step": 5362
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.056528375653944e-06,
+      "loss": 0.0381,
+      "step": 5363
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.053256558131054e-06,
+      "loss": 0.0351,
+      "step": 5364
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.049984842868484e-06,
+      "loss": 0.0341,
+      "step": 5365
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.046713230219608e-06,
+      "loss": 0.0298,
+      "step": 5366
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.043441720537814e-06,
+      "loss": 0.0316,
+      "step": 5367
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.040170314176461e-06,
+      "loss": 0.0285,
+      "step": 5368
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.0368990114889e-06,
+      "loss": 0.0283,
+      "step": 5369
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.03362781282848e-06,
+      "loss": 0.0323,
+      "step": 5370
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.030356718548526e-06,
+      "loss": 0.0333,
+      "step": 5371
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.027085729002361e-06,
+      "loss": 0.0293,
+      "step": 5372
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.023814844543292e-06,
+      "loss": 0.0374,
+      "step": 5373
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.020544065524614e-06,
+      "loss": 0.0287,
+      "step": 5374
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.017273392299619e-06,
+      "loss": 0.0251,
+      "step": 5375
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.014002825221572e-06,
+      "loss": 0.0262,
+      "step": 5376
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.010732364643744e-06,
+      "loss": 0.0314,
+      "step": 5377
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.007462010919387e-06,
+      "loss": 0.0229,
+      "step": 5378
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.004191764401734e-06,
+      "loss": 0.0424,
+      "step": 5379
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.000921625444015e-06,
+      "loss": 0.0344,
+      "step": 5380
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 8.997651594399446e-06,
+      "loss": 0.0296,
+      "step": 5381
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 8.994381671621238e-06,
+      "loss": 0.0298,
+      "step": 5382
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 8.991111857462575e-06,
+      "loss": 0.0383,
+      "step": 5383
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 8.987842152276648e-06,
+      "loss": 0.0277,
+      "step": 5384
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.984572556416623e-06,
+      "loss": 0.0296,
+      "step": 5385
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.981303070235652e-06,
+      "loss": 0.0329,
+      "step": 5386
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.97803369408689e-06,
+      "loss": 0.0322,
+      "step": 5387
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.974764428323461e-06,
+      "loss": 0.031,
+      "step": 5388
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.9714952732985e-06,
+      "loss": 0.036,
+      "step": 5389
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 8.968226229365109e-06,
+      "loss": 0.0297,
+      "step": 5390
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.964957296876383e-06,
+      "loss": 0.0274,
+      "step": 5391
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.961688476185417e-06,
+      "loss": 0.0275,
+      "step": 5392
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.95841976764528e-06,
+      "loss": 0.0359,
+      "step": 5393
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.955151171609037e-06,
+      "loss": 0.0246,
+      "step": 5394
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.951882688429737e-06,
+      "loss": 0.0278,
+      "step": 5395
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.94861431846041e-06,
+      "loss": 0.0293,
+      "step": 5396
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.945346062054094e-06,
+      "loss": 0.0403,
+      "step": 5397
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.942077919563791e-06,
+      "loss": 0.0295,
+      "step": 5398
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.938809891342512e-06,
+      "loss": 0.0374,
+      "step": 5399
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.935541977743243e-06,
+      "loss": 0.0377,
+      "step": 5400
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.932274179118952e-06,
+      "loss": 0.0344,
+      "step": 5401
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.929006495822615e-06,
+      "loss": 0.0398,
+      "step": 5402
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.92573892820717e-06,
+      "loss": 0.032,
+      "step": 5403
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.922471476625567e-06,
+      "loss": 0.0313,
+      "step": 5404
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.919204141430724e-06,
+      "loss": 0.0353,
+      "step": 5405
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.915936922975564e-06,
+      "loss": 0.031,
+      "step": 5406
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.91266982161298e-06,
+      "loss": 0.027,
+      "step": 5407
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.909402837695864e-06,
+      "loss": 0.0329,
+      "step": 5408
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.906135971577094e-06,
+      "loss": 0.0328,
+      "step": 5409
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.902869223609523e-06,
+      "loss": 0.0344,
+      "step": 5410
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.899602594146013e-06,
+      "loss": 0.0355,
+      "step": 5411
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.896336083539399e-06,
+      "loss": 0.0257,
+      "step": 5412
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.893069692142498e-06,
+      "loss": 0.0283,
+      "step": 5413
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.88980342030813e-06,
+      "loss": 0.0318,
+      "step": 5414
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.886537268389091e-06,
+      "loss": 0.0337,
+      "step": 5415
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.883271236738168e-06,
+      "loss": 0.0412,
+      "step": 5416
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.880005325708133e-06,
+      "loss": 0.035,
+      "step": 5417
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.876739535651741e-06,
+      "loss": 0.0322,
+      "step": 5418
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.873473866921749e-06,
+      "loss": 0.0388,
+      "step": 5419
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.870208319870878e-06,
+      "loss": 0.0351,
+      "step": 5420
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.866942894851862e-06,
+      "loss": 0.0304,
+      "step": 5421
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.863677592217402e-06,
+      "loss": 0.0394,
+      "step": 5422
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.860412412320189e-06,
+      "loss": 0.0302,
+      "step": 5423
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.85714735551291e-06,
+      "loss": 0.0325,
+      "step": 5424
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.853882422148225e-06,
+      "loss": 0.0291,
+      "step": 5425
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.850617612578797e-06,
+      "loss": 0.0348,
+      "step": 5426
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.847352927157265e-06,
+      "loss": 0.0313,
+      "step": 5427
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.84408836623625e-06,
+      "loss": 0.0301,
+      "step": 5428
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.840823930168372e-06,
+      "loss": 0.0264,
+      "step": 5429
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.837559619306228e-06,
+      "loss": 0.0266,
+      "step": 5430
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.83429543400241e-06,
+      "loss": 0.0355,
+      "step": 5431
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.831031374609485e-06,
+      "loss": 0.0362,
+      "step": 5432
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.82776744148002e-06,
+      "loss": 0.0283,
+      "step": 5433
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.824503634966556e-06,
+      "loss": 0.0355,
+      "step": 5434
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.821239955421624e-06,
+      "loss": 0.0341,
+      "step": 5435
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.81797640319775e-06,
+      "loss": 0.0336,
+      "step": 5436
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.814712978647432e-06,
+      "loss": 0.0277,
+      "step": 5437
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.811449682123168e-06,
+      "loss": 0.0286,
+      "step": 5438
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.808186513977432e-06,
+      "loss": 0.0394,
+      "step": 5439
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.804923474562683e-06,
+      "loss": 0.0275,
+      "step": 5440
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.801660564231379e-06,
+      "loss": 0.0336,
+      "step": 5441
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.798397783335946e-06,
+      "loss": 0.0331,
+      "step": 5442
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.795135132228816e-06,
+      "loss": 0.028,
+      "step": 5443
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.791872611262393e-06,
+      "loss": 0.0311,
+      "step": 5444
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.788610220789068e-06,
+      "loss": 0.0369,
+      "step": 5445
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.785347961161225e-06,
+      "loss": 0.0331,
+      "step": 5446
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.782085832731222e-06,
+      "loss": 0.0369,
+      "step": 5447
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.77882383585142e-06,
+      "loss": 0.0309,
+      "step": 5448
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.775561970874151e-06,
+      "loss": 0.0288,
+      "step": 5449
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.772300238151735e-06,
+      "loss": 0.0343,
+      "step": 5450
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.769038638036488e-06,
+      "loss": 0.034,
+      "step": 5451
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.765777170880697e-06,
+      "loss": 0.0331,
+      "step": 5452
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.762515837036646e-06,
+      "loss": 0.0285,
+      "step": 5453
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.759254636856597e-06,
+      "loss": 0.0332,
+      "step": 5454
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.755993570692808e-06,
+      "loss": 0.0373,
+      "step": 5455
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.75273263889751e-06,
+      "loss": 0.0275,
+      "step": 5456
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.749471841822923e-06,
+      "loss": 0.0278,
+      "step": 5457
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.74621117982126e-06,
+      "loss": 0.0306,
+      "step": 5458
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.742950653244712e-06,
+      "loss": 0.0382,
+      "step": 5459
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.739690262445457e-06,
+      "loss": 0.0304,
+      "step": 5460
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.736430007775658e-06,
+      "loss": 0.0381,
+      "step": 5461
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.73316988958746e-06,
+      "loss": 0.0303,
+      "step": 5462
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.729909908233006e-06,
+      "loss": 0.027,
+      "step": 5463
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.726650064064407e-06,
+      "loss": 0.0399,
+      "step": 5464
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.723390357433775e-06,
+      "loss": 0.0321,
+      "step": 5465
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.720130788693197e-06,
+      "loss": 0.0285,
+      "step": 5466
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.716871358194745e-06,
+      "loss": 0.0332,
+      "step": 5467
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.713612066290481e-06,
+      "loss": 0.038,
+      "step": 5468
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.710352913332446e-06,
+      "loss": 0.035,
+      "step": 5469
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.707093899672678e-06,
+      "loss": 0.0346,
+      "step": 5470
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.703835025663188e-06,
+      "loss": 0.0331,
+      "step": 5471
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.700576291655972e-06,
+      "loss": 0.0305,
+      "step": 5472
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.69731769800302e-06,
+      "loss": 0.0271,
+      "step": 5473
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.694059245056298e-06,
+      "loss": 0.031,
+      "step": 5474
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.690800933167763e-06,
+      "loss": 0.034,
+      "step": 5475
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.687542762689355e-06,
+      "loss": 0.034,
+      "step": 5476
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.684284733972989e-06,
+      "loss": 0.0316,
+      "step": 5477
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.681026847370585e-06,
+      "loss": 0.0363,
+      "step": 5478
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.677769103234026e-06,
+      "loss": 0.0403,
+      "step": 5479
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.6745115019152e-06,
+      "loss": 0.0306,
+      "step": 5480
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.671254043765962e-06,
+      "loss": 0.0392,
+      "step": 5481
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.667996729138163e-06,
+      "loss": 0.0273,
+      "step": 5482
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.664739558383633e-06,
+      "loss": 0.0316,
+      "step": 5483
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.661482531854182e-06,
+      "loss": 0.037,
+      "step": 5484
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.65822564990162e-06,
+      "loss": 0.0306,
+      "step": 5485
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.654968912877724e-06,
+      "loss": 0.0284,
+      "step": 5486
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.65171232113427e-06,
+      "loss": 0.0298,
+      "step": 5487
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.648455875023006e-06,
+      "loss": 0.0367,
+      "step": 5488
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.645199574895668e-06,
+      "loss": 0.03,
+      "step": 5489
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.641943421103984e-06,
+      "loss": 0.0328,
+      "step": 5490
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.63868741399965e-06,
+      "loss": 0.03,
+      "step": 5491
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.63543155393437e-06,
+      "loss": 0.0332,
+      "step": 5492
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.63217584125981e-06,
+      "loss": 0.0296,
+      "step": 5493
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.628920276327624e-06,
+      "loss": 0.0302,
+      "step": 5494
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.625664859489464e-06,
+      "loss": 0.0297,
+      "step": 5495
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.62240959109695e-06,
+      "loss": 0.0338,
+      "step": 5496
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.619154471501694e-06,
+      "loss": 0.0333,
+      "step": 5497
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.615899501055293e-06,
+      "loss": 0.0305,
+      "step": 5498
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.61264468010932e-06,
+      "loss": 0.0303,
+      "step": 5499
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.609390009015341e-06,
+      "loss": 0.0271,
+      "step": 5500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.606135488124895e-06,
+      "loss": 0.0347,
+      "step": 5501
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.602881117789525e-06,
+      "loss": 0.0355,
+      "step": 5502
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.59962689836073e-06,
+      "loss": 0.0294,
+      "step": 5503
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.596372830190019e-06,
+      "loss": 0.0289,
+      "step": 5504
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.593118913628866e-06,
+      "loss": 0.0365,
+      "step": 5505
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.589865149028732e-06,
+      "loss": 0.0322,
+      "step": 5506
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.586611536741072e-06,
+      "loss": 0.0313,
+      "step": 5507
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.583358077117312e-06,
+      "loss": 0.0267,
+      "step": 5508
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.580104770508874e-06,
+      "loss": 0.0292,
+      "step": 5509
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.0279,
+      "step": 5510
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.573598617743524e-06,
+      "loss": 0.0334,
+      "step": 5511
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.570345772289362e-06,
+      "loss": 0.034,
+      "step": 5512
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.567093081256008e-06,
+      "loss": 0.037,
+      "step": 5513
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.563840544994802e-06,
+      "loss": 0.0331,
+      "step": 5514
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.560588163857056e-06,
+      "loss": 0.0344,
+      "step": 5515
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.557335938194061e-06,
+      "loss": 0.0374,
+      "step": 5516
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.55408386835711e-06,
+      "loss": 0.0322,
+      "step": 5517
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.550831954697461e-06,
+      "loss": 0.0321,
+      "step": 5518
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.547580197566367e-06,
+      "loss": 0.0289,
+      "step": 5519
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.544328597315055e-06,
+      "loss": 0.0276,
+      "step": 5520
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.541077154294735e-06,
+      "loss": 0.0332,
+      "step": 5521
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.537825868856615e-06,
+      "loss": 0.0292,
+      "step": 5522
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.534574741351864e-06,
+      "loss": 0.0386,
+      "step": 5523
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.531323772131654e-06,
+      "loss": 0.0397,
+      "step": 5524
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.528072961547128e-06,
+      "loss": 0.0378,
+      "step": 5525
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.524822309949413e-06,
+      "loss": 0.0357,
+      "step": 5526
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.521571817689624e-06,
+      "loss": 0.0256,
+      "step": 5527
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.518321485118847e-06,
+      "loss": 0.0339,
+      "step": 5528
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.515071312588171e-06,
+      "loss": 0.0317,
+      "step": 5529
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.511821300448646e-06,
+      "loss": 0.0392,
+      "step": 5530
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.508571449051325e-06,
+      "loss": 0.0311,
+      "step": 5531
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.505321758747224e-06,
+      "loss": 0.0346,
+      "step": 5532
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.502072229887352e-06,
+      "loss": 0.029,
+      "step": 5533
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.498822862822706e-06,
+      "loss": 0.0358,
+      "step": 5534
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.495573657904248e-06,
+      "loss": 0.0313,
+      "step": 5535
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.492324615482946e-06,
+      "loss": 0.0382,
+      "step": 5536
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.489075735909729e-06,
+      "loss": 0.0275,
+      "step": 5537
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.485827019535517e-06,
+      "loss": 0.0357,
+      "step": 5538
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.48257846671122e-06,
+      "loss": 0.0373,
+      "step": 5539
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.479330077787714e-06,
+      "loss": 0.036,
+      "step": 5540
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 8.476081853115873e-06,
+      "loss": 0.0386,
+      "step": 5541
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.472833793046547e-06,
+      "loss": 0.0387,
+      "step": 5542
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.469585897930557e-06,
+      "loss": 0.034,
+      "step": 5543
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.466338168118731e-06,
+      "loss": 0.0302,
+      "step": 5544
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.463090603961854e-06,
+      "loss": 0.0329,
+      "step": 5545
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.459843205810715e-06,
+      "loss": 0.0311,
+      "step": 5546
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.456595974016067e-06,
+      "loss": 0.038,
+      "step": 5547
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.45334890892865e-06,
+      "loss": 0.0274,
+      "step": 5548
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.450102010899194e-06,
+      "loss": 0.0315,
+      "step": 5549
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.4468552802784e-06,
+      "loss": 0.035,
+      "step": 5550
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.443608717416964e-06,
+      "loss": 0.0456,
+      "step": 5551
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.440362322665547e-06,
+      "loss": 0.0323,
+      "step": 5552
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.437116096374807e-06,
+      "loss": 0.0337,
+      "step": 5553
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.433870038895377e-06,
+      "loss": 0.0414,
+      "step": 5554
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.430624150577868e-06,
+      "loss": 0.0345,
+      "step": 5555
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.427378431772883e-06,
+      "loss": 0.0367,
+      "step": 5556
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.424132882830996e-06,
+      "loss": 0.0354,
+      "step": 5557
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.42088750410277e-06,
+      "loss": 0.0272,
+      "step": 5558
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.41764229593875e-06,
+      "loss": 0.0344,
+      "step": 5559
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.41439725868945e-06,
+      "loss": 0.0345,
+      "step": 5560
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.411152392705386e-06,
+      "loss": 0.0313,
+      "step": 5561
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.407907698337039e-06,
+      "loss": 0.0333,
+      "step": 5562
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.40466317593488e-06,
+      "loss": 0.03,
+      "step": 5563
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.40141882584936e-06,
+      "loss": 0.035,
+      "step": 5564
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.398174648430898e-06,
+      "loss": 0.0352,
+      "step": 5565
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.394930644029924e-06,
+      "loss": 0.0363,
+      "step": 5566
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.391686812996817e-06,
+      "loss": 0.0429,
+      "step": 5567
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.388443155681962e-06,
+      "loss": 0.0392,
+      "step": 5568
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.385199672435712e-06,
+      "loss": 0.0349,
+      "step": 5569
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.381956363608401e-06,
+      "loss": 0.0327,
+      "step": 5570
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.378713229550354e-06,
+      "loss": 0.0341,
+      "step": 5571
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.375470270611862e-06,
+      "loss": 0.0334,
+      "step": 5572
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.372227487143213e-06,
+      "loss": 0.0299,
+      "step": 5573
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 8.368984879494669e-06,
+      "loss": 0.035,
+      "step": 5574
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.365742448016465e-06,
+      "loss": 0.0376,
+      "step": 5575
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.362500193058836e-06,
+      "loss": 0.0426,
+      "step": 5576
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.359258114971976e-06,
+      "loss": 0.0343,
+      "step": 5577
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.356016214106078e-06,
+      "loss": 0.0298,
+      "step": 5578
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.352774490811303e-06,
+      "loss": 0.0303,
+      "step": 5579
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.349532945437804e-06,
+      "loss": 0.0345,
+      "step": 5580
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.346291578335706e-06,
+      "loss": 0.0316,
+      "step": 5581
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.343050389855115e-06,
+      "loss": 0.0357,
+      "step": 5582
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.339809380346127e-06,
+      "loss": 0.0445,
+      "step": 5583
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.336568550158807e-06,
+      "loss": 0.0355,
+      "step": 5584
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.333327899643208e-06,
+      "loss": 0.0284,
+      "step": 5585
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.33008742914936e-06,
+      "loss": 0.0348,
+      "step": 5586
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.326847139027278e-06,
+      "loss": 0.0264,
+      "step": 5587
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.323607029626953e-06,
+      "loss": 0.042,
+      "step": 5588
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.320367101298351e-06,
+      "loss": 0.0349,
+      "step": 5589
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.317127354391437e-06,
+      "loss": 0.0312,
+      "step": 5590
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.313887789256142e-06,
+      "loss": 0.0334,
+      "step": 5591
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.310648406242373e-06,
+      "loss": 0.0327,
+      "step": 5592
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.307409205700031e-06,
+      "loss": 0.0294,
+      "step": 5593
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.304170187978987e-06,
+      "loss": 0.0354,
+      "step": 5594
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.300931353429099e-06,
+      "loss": 0.0334,
+      "step": 5595
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.297692702400205e-06,
+      "loss": 0.037,
+      "step": 5596
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.29445423524211e-06,
+      "loss": 0.0269,
+      "step": 5597
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.291215952304622e-06,
+      "loss": 0.0351,
+      "step": 5598
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.287977853937507e-06,
+      "loss": 0.0349,
+      "step": 5599
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.284739940490528e-06,
+      "loss": 0.0301,
+      "step": 5600
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.281502212313412e-06,
+      "loss": 0.0387,
+      "step": 5601
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.278264669755887e-06,
+      "loss": 0.0364,
+      "step": 5602
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.27502731316764e-06,
+      "loss": 0.0247,
+      "step": 5603
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.271790142898345e-06,
+      "loss": 0.0297,
+      "step": 5604
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.268553159297665e-06,
+      "loss": 0.035,
+      "step": 5605
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.26531636271523e-06,
+      "loss": 0.0271,
+      "step": 5606
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.262079753500658e-06,
+      "loss": 0.0355,
+      "step": 5607
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.258843332003541e-06,
+      "loss": 0.04,
+      "step": 5608
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.255607098573454e-06,
+      "loss": 0.0386,
+      "step": 5609
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.252371053559956e-06,
+      "loss": 0.0336,
+      "step": 5610
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.249135197312572e-06,
+      "loss": 0.032,
+      "step": 5611
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.245899530180826e-06,
+      "loss": 0.0314,
+      "step": 5612
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.242664052514206e-06,
+      "loss": 0.0335,
+      "step": 5613
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.239428764662182e-06,
+      "loss": 0.0367,
+      "step": 5614
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.236193666974209e-06,
+      "loss": 0.0328,
+      "step": 5615
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.23295875979972e-06,
+      "loss": 0.03,
+      "step": 5616
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.229724043488126e-06,
+      "loss": 0.0323,
+      "step": 5617
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.226489518388814e-06,
+      "loss": 0.0351,
+      "step": 5618
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.223255184851153e-06,
+      "loss": 0.0311,
+      "step": 5619
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.2200210432245e-06,
+      "loss": 0.0351,
+      "step": 5620
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.216787093858174e-06,
+      "loss": 0.0384,
+      "step": 5621
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.213553337101492e-06,
+      "loss": 0.0341,
+      "step": 5622
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.21031977330373e-06,
+      "loss": 0.0345,
+      "step": 5623
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.207086402814163e-06,
+      "loss": 0.0306,
+      "step": 5624
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.203853225982032e-06,
+      "loss": 0.0354,
+      "step": 5625
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.20062024315656e-06,
+      "loss": 0.0338,
+      "step": 5626
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.197387454686954e-06,
+      "loss": 0.0369,
+      "step": 5627
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.194154860922395e-06,
+      "loss": 0.0279,
+      "step": 5628
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.190922462212044e-06,
+      "loss": 0.0319,
+      "step": 5629
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.187690258905038e-06,
+      "loss": 0.0404,
+      "step": 5630
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.1844582513505e-06,
+      "loss": 0.0313,
+      "step": 5631
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.181226439897527e-06,
+      "loss": 0.0319,
+      "step": 5632
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 8.17799482489519e-06,
+      "loss": 0.0348,
+      "step": 5633
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.174763406692557e-06,
+      "loss": 0.0364,
+      "step": 5634
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.171532185638652e-06,
+      "loss": 0.0294,
+      "step": 5635
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.168301162082489e-06,
+      "loss": 0.0369,
+      "step": 5636
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.165070336373062e-06,
+      "loss": 0.0332,
+      "step": 5637
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.161839708859341e-06,
+      "loss": 0.0359,
+      "step": 5638
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.158609279890275e-06,
+      "loss": 0.0301,
+      "step": 5639
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.155379049814791e-06,
+      "loss": 0.0391,
+      "step": 5640
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.15214901898179e-06,
+      "loss": 0.0333,
+      "step": 5641
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.148919187740165e-06,
+      "loss": 0.0346,
+      "step": 5642
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.145689556438773e-06,
+      "loss": 0.033,
+      "step": 5643
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.142460125426456e-06,
+      "loss": 0.0399,
+      "step": 5644
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.139230895052036e-06,
+      "loss": 0.0358,
+      "step": 5645
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.136001865664308e-06,
+      "loss": 0.0311,
+      "step": 5646
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.132773037612053e-06,
+      "loss": 0.0362,
+      "step": 5647
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.129544411244018e-06,
+      "loss": 0.0308,
+      "step": 5648
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.12631598690894e-06,
+      "loss": 0.0297,
+      "step": 5649
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.123087764955529e-06,
+      "loss": 0.0354,
+      "step": 5650
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.119859745732475e-06,
+      "loss": 0.0365,
+      "step": 5651
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.116631929588445e-06,
+      "loss": 0.0348,
+      "step": 5652
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.113404316872083e-06,
+      "loss": 0.0268,
+      "step": 5653
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.110176907932014e-06,
+      "loss": 0.039,
+      "step": 5654
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.106949703116834e-06,
+      "loss": 0.0359,
+      "step": 5655
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.10372270277513e-06,
+      "loss": 0.0279,
+      "step": 5656
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.100495907255454e-06,
+      "loss": 0.0346,
+      "step": 5657
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.097269316906342e-06,
+      "loss": 0.0373,
+      "step": 5658
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.094042932076304e-06,
+      "loss": 0.0366,
+      "step": 5659
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 8.090816753113835e-06,
+      "loss": 0.0409,
+      "step": 5660
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.087590780367402e-06,
+      "loss": 0.0369,
+      "step": 5661
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.084365014185451e-06,
+      "loss": 0.0385,
+      "step": 5662
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.0811394549164e-06,
+      "loss": 0.0368,
+      "step": 5663
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.077914102908661e-06,
+      "loss": 0.0318,
+      "step": 5664
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.074688958510604e-06,
+      "loss": 0.0333,
+      "step": 5665
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.07146402207059e-06,
+      "loss": 0.037,
+      "step": 5666
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.068239293936955e-06,
+      "loss": 0.0296,
+      "step": 5667
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.065014774458004e-06,
+      "loss": 0.0367,
+      "step": 5668
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.061790463982033e-06,
+      "loss": 0.0275,
+      "step": 5669
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.058566362857301e-06,
+      "loss": 0.0361,
+      "step": 5670
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.055342471432059e-06,
+      "loss": 0.0392,
+      "step": 5671
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.052118790054526e-06,
+      "loss": 0.0368,
+      "step": 5672
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.048895319072896e-06,
+      "loss": 0.0337,
+      "step": 5673
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.045672058835352e-06,
+      "loss": 0.0307,
+      "step": 5674
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.042449009690044e-06,
+      "loss": 0.0349,
+      "step": 5675
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.0392261719851e-06,
+      "loss": 0.0316,
+      "step": 5676
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.036003546068626e-06,
+      "loss": 0.039,
+      "step": 5677
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.032781132288716e-06,
+      "loss": 0.0348,
+      "step": 5678
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.029558930993425e-06,
+      "loss": 0.0312,
+      "step": 5679
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.026336942530787e-06,
+      "loss": 0.0335,
+      "step": 5680
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.023115167248828e-06,
+      "loss": 0.0314,
+      "step": 5681
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.019893605495533e-06,
+      "loss": 0.0362,
+      "step": 5682
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.016672257618875e-06,
+      "loss": 0.0308,
+      "step": 5683
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.0134511239668e-06,
+      "loss": 0.0399,
+      "step": 5684
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.010230204887228e-06,
+      "loss": 0.0326,
+      "step": 5685
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.007009500728065e-06,
+      "loss": 0.0384,
+      "step": 5686
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 8.003789011837182e-06,
+      "loss": 0.0337,
+      "step": 5687
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 8.000568738562437e-06,
+      "loss": 0.0254,
+      "step": 5688
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.99734868125166e-06,
+      "loss": 0.0347,
+      "step": 5689
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.994128840252655e-06,
+      "loss": 0.0376,
+      "step": 5690
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.99090921591321e-06,
+      "loss": 0.0327,
+      "step": 5691
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.98768980858108e-06,
+      "loss": 0.0397,
+      "step": 5692
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.984470618604008e-06,
+      "loss": 0.0393,
+      "step": 5693
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.981251646329704e-06,
+      "loss": 0.0336,
+      "step": 5694
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.978032892105855e-06,
+      "loss": 0.0362,
+      "step": 5695
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.974814356280135e-06,
+      "loss": 0.0383,
+      "step": 5696
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.97159603920018e-06,
+      "loss": 0.0313,
+      "step": 5697
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.968377941213614e-06,
+      "loss": 0.0352,
+      "step": 5698
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 7.965160062668024e-06,
+      "loss": 0.0419,
+      "step": 5699
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.961942403910994e-06,
+      "loss": 0.034,
+      "step": 5700
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.958724965290064e-06,
+      "loss": 0.0358,
+      "step": 5701
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.955507747152758e-06,
+      "loss": 0.0339,
+      "step": 5702
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.95229074984658e-06,
+      "loss": 0.0386,
+      "step": 5703
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.949073973719004e-06,
+      "loss": 0.0354,
+      "step": 5704
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.945857419117485e-06,
+      "loss": 0.0302,
+      "step": 5705
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.94264108638945e-06,
+      "loss": 0.034,
+      "step": 5706
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.939424975882301e-06,
+      "loss": 0.0311,
+      "step": 5707
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.936209087943424e-06,
+      "loss": 0.0364,
+      "step": 5708
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.932993422920172e-06,
+      "loss": 0.0288,
+      "step": 5709
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.929777981159883e-06,
+      "loss": 0.032,
+      "step": 5710
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.926562763009861e-06,
+      "loss": 0.0337,
+      "step": 5711
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 7.923347768817389e-06,
+      "loss": 0.033,
+      "step": 5712
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.92013299892973e-06,
+      "loss": 0.0387,
+      "step": 5713
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.916918453694118e-06,
+      "loss": 0.034,
+      "step": 5714
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.913704133457769e-06,
+      "loss": 0.0339,
+      "step": 5715
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.910490038567869e-06,
+      "loss": 0.0331,
+      "step": 5716
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.907276169371572e-06,
+      "loss": 0.0372,
+      "step": 5717
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.90406252621603e-06,
+      "loss": 0.0463,
+      "step": 5718
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.90084910944835e-06,
+      "loss": 0.0349,
+      "step": 5719
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.897635919415625e-06,
+      "loss": 0.048,
+      "step": 5720
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.89442295646492e-06,
+      "loss": 0.0397,
+      "step": 5721
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.891210220943268e-06,
+      "loss": 0.0342,
+      "step": 5722
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.887997713197698e-06,
+      "loss": 0.0292,
+      "step": 5723
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.884785433575188e-06,
+      "loss": 0.0401,
+      "step": 5724
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 7.881573382422718e-06,
+      "loss": 0.0339,
+      "step": 5725
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.878361560087223e-06,
+      "loss": 0.0348,
+      "step": 5726
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.875149966915625e-06,
+      "loss": 0.0315,
+      "step": 5727
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.871938603254812e-06,
+      "loss": 0.0387,
+      "step": 5728
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.868727469451652e-06,
+      "loss": 0.0353,
+      "step": 5729
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.865516565852993e-06,
+      "loss": 0.0365,
+      "step": 5730
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.862305892805647e-06,
+      "loss": 0.0342,
+      "step": 5731
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 7.859095450656416e-06,
+      "loss": 0.0313,
+      "step": 5732
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.855885239752062e-06,
+      "loss": 0.036,
+      "step": 5733
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.852675260439328e-06,
+      "loss": 0.0278,
+      "step": 5734
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.849465513064938e-06,
+      "loss": 0.0333,
+      "step": 5735
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.846255997975579e-06,
+      "loss": 0.0338,
+      "step": 5736
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.843046715517926e-06,
+      "loss": 0.0286,
+      "step": 5737
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.839837666038619e-06,
+      "loss": 0.0402,
+      "step": 5738
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.83662884988427e-06,
+      "loss": 0.0349,
+      "step": 5739
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.833420267401484e-06,
+      "loss": 0.0372,
+      "step": 5740
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.83021191893682e-06,
+      "loss": 0.0357,
+      "step": 5741
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.827003804836825e-06,
+      "loss": 0.0338,
+      "step": 5742
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.823795925448013e-06,
+      "loss": 0.039,
+      "step": 5743
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.820588281116874e-06,
+      "loss": 0.028,
+      "step": 5744
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 7.817380872189881e-06,
+      "loss": 0.0416,
+      "step": 5745
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.814173699013465e-06,
+      "loss": 0.0334,
+      "step": 5746
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.810966761934053e-06,
+      "loss": 0.0339,
+      "step": 5747
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.807760061298027e-06,
+      "loss": 0.0313,
+      "step": 5748
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.804553597451755e-06,
+      "loss": 0.0307,
+      "step": 5749
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.801347370741575e-06,
+      "loss": 0.0339,
+      "step": 5750
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.798141381513795e-06,
+      "loss": 0.0397,
+      "step": 5751
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 7.794935630114712e-06,
+      "loss": 0.0408,
+      "step": 5752
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.79173011689058e-06,
+      "loss": 0.0316,
+      "step": 5753
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.78852484218764e-06,
+      "loss": 0.0326,
+      "step": 5754
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.785319806352101e-06,
+      "loss": 0.037,
+      "step": 5755
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.782115009730146e-06,
+      "loss": 0.0364,
+      "step": 5756
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.778910452667936e-06,
+      "loss": 0.0317,
+      "step": 5757
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.7757061355116e-06,
+      "loss": 0.0347,
+      "step": 5758
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.77250205860725e-06,
+      "loss": 0.0385,
+      "step": 5759
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.769298222300966e-06,
+      "loss": 0.0295,
+      "step": 5760
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.766094626938797e-06,
+      "loss": 0.0362,
+      "step": 5761
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.762891272866782e-06,
+      "loss": 0.0382,
+      "step": 5762
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.759688160430913e-06,
+      "loss": 0.0341,
+      "step": 5763
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.756485289977178e-06,
+      "loss": 0.0354,
+      "step": 5764
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.753282661851521e-06,
+      "loss": 0.0348,
+      "step": 5765
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.750080276399864e-06,
+      "loss": 0.0326,
+      "step": 5766
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.746878133968112e-06,
+      "loss": 0.0342,
+      "step": 5767
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.743676234902129e-06,
+      "loss": 0.0362,
+      "step": 5768
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.740474579547773e-06,
+      "loss": 0.0394,
+      "step": 5769
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.737273168250853e-06,
+      "loss": 0.034,
+      "step": 5770
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.734072001357165e-06,
+      "loss": 0.0366,
+      "step": 5771
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.730871079212476e-06,
+      "loss": 0.0313,
+      "step": 5772
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.727670402162524e-06,
+      "loss": 0.0331,
+      "step": 5773
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.724469970553028e-06,
+      "loss": 0.0363,
+      "step": 5774
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.721269784729668e-06,
+      "loss": 0.0348,
+      "step": 5775
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.718069845038114e-06,
+      "loss": 0.0304,
+      "step": 5776
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.714870151823992e-06,
+      "loss": 0.0377,
+      "step": 5777
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.711670705432914e-06,
+      "loss": 0.046,
+      "step": 5778
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.708471506210459e-06,
+      "loss": 0.0338,
+      "step": 5779
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.705272554502178e-06,
+      "loss": 0.0295,
+      "step": 5780
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.702073850653606e-06,
+      "loss": 0.0408,
+      "step": 5781
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.69887539501024e-06,
+      "loss": 0.0424,
+      "step": 5782
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.695677187917548e-06,
+      "loss": 0.0336,
+      "step": 5783
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.692479229720989e-06,
+      "loss": 0.0367,
+      "step": 5784
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.68928152076597e-06,
+      "loss": 0.0472,
+      "step": 5785
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.686084061397897e-06,
+      "loss": 0.0377,
+      "step": 5786
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.682886851962127e-06,
+      "loss": 0.0345,
+      "step": 5787
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.679689892803999e-06,
+      "loss": 0.0366,
+      "step": 5788
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.676493184268832e-06,
+      "loss": 0.0266,
+      "step": 5789
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.673296726701902e-06,
+      "loss": 0.0353,
+      "step": 5790
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.67010052044848e-06,
+      "loss": 0.0339,
+      "step": 5791
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.666904565853786e-06,
+      "loss": 0.031,
+      "step": 5792
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.663708863263026e-06,
+      "loss": 0.0349,
+      "step": 5793
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.660513413021379e-06,
+      "loss": 0.0382,
+      "step": 5794
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.657318215473987e-06,
+      "loss": 0.039,
+      "step": 5795
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.654123270965984e-06,
+      "loss": 0.0378,
+      "step": 5796
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.650928579842453e-06,
+      "loss": 0.0378,
+      "step": 5797
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.647734142448471e-06,
+      "loss": 0.0346,
+      "step": 5798
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.644539959129072e-06,
+      "loss": 0.037,
+      "step": 5799
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.641346030229268e-06,
+      "loss": 0.0348,
+      "step": 5800
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.638152356094046e-06,
+      "loss": 0.0357,
+      "step": 5801
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.634958937068359e-06,
+      "loss": 0.0351,
+      "step": 5802
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.631765773497146e-06,
+      "loss": 0.0289,
+      "step": 5803
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.6285728657253036e-06,
+      "loss": 0.0352,
+      "step": 5804
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.6253802140977005e-06,
+      "loss": 0.0307,
+      "step": 5805
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.622187818959195e-06,
+      "loss": 0.0385,
+      "step": 5806
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.618995680654599e-06,
+      "loss": 0.0319,
+      "step": 5807
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.615803799528707e-06,
+      "loss": 0.0342,
+      "step": 5808
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.612612175926282e-06,
+      "loss": 0.0321,
+      "step": 5809
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.609420810192054e-06,
+      "loss": 0.0342,
+      "step": 5810
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 7.606229702670742e-06,
+      "loss": 0.0342,
+      "step": 5811
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.603038853707017e-06,
+      "loss": 0.0316,
+      "step": 5812
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.599848263645538e-06,
+      "loss": 0.0359,
+      "step": 5813
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.5966579328309265e-06,
+      "loss": 0.0356,
+      "step": 5814
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.593467861607776e-06,
+      "loss": 0.0372,
+      "step": 5815
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.5902780503206585e-06,
+      "loss": 0.0374,
+      "step": 5816
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 7.587088499314109e-06,
+      "loss": 0.0347,
+      "step": 5817
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.583899208932648e-06,
+      "loss": 0.0395,
+      "step": 5818
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.580710179520754e-06,
+      "loss": 0.0356,
+      "step": 5819
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.5775214114228775e-06,
+      "loss": 0.0371,
+      "step": 5820
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.574332904983458e-06,
+      "loss": 0.0322,
+      "step": 5821
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.571144660546885e-06,
+      "loss": 0.0344,
+      "step": 5822
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.567956678457535e-06,
+      "loss": 0.0411,
+      "step": 5823
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 7.564768959059744e-06,
+      "loss": 0.0383,
+      "step": 5824
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.561581502697835e-06,
+      "loss": 0.0355,
+      "step": 5825
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.558394309716088e-06,
+      "loss": 0.0298,
+      "step": 5826
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.555207380458759e-06,
+      "loss": 0.0331,
+      "step": 5827
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.552020715270083e-06,
+      "loss": 0.0386,
+      "step": 5828
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.548834314494254e-06,
+      "loss": 0.0311,
+      "step": 5829
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 7.545648178475448e-06,
+      "loss": 0.0339,
+      "step": 5830
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.5424623075578076e-06,
+      "loss": 0.0376,
+      "step": 5831
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.5392767020854414e-06,
+      "loss": 0.0384,
+      "step": 5832
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.536091362402446e-06,
+      "loss": 0.0356,
+      "step": 5833
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.532906288852866e-06,
+      "loss": 0.0325,
+      "step": 5834
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.529721481780742e-06,
+      "loss": 0.042,
+      "step": 5835
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.526536941530067e-06,
+      "loss": 0.0301,
+      "step": 5836
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 7.523352668444812e-06,
+      "loss": 0.0412,
+      "step": 5837
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.520168662868921e-06,
+      "loss": 0.0311,
+      "step": 5838
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.5169849251463025e-06,
+      "loss": 0.034,
+      "step": 5839
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.513801455620847e-06,
+      "loss": 0.0321,
+      "step": 5840
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.510618254636407e-06,
+      "loss": 0.0317,
+      "step": 5841
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.507435322536803e-06,
+      "loss": 0.0368,
+      "step": 5842
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.504252659665843e-06,
+      "loss": 0.0354,
+      "step": 5843
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 7.501070266367287e-06,
+      "loss": 0.0347,
+      "step": 5844
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.497888142984877e-06,
+      "loss": 0.0283,
+      "step": 5845
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.494706289862319e-06,
+      "loss": 0.0402,
+      "step": 5846
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.491524707343299e-06,
+      "loss": 0.0397,
+      "step": 5847
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.4883433957714665e-06,
+      "loss": 0.0365,
+      "step": 5848
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.4851623554904395e-06,
+      "loss": 0.0405,
+      "step": 5849
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 7.481981586843817e-06,
+      "loss": 0.0372,
+      "step": 5850
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.478801090175159e-06,
+      "loss": 0.031,
+      "step": 5851
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.475620865828001e-06,
+      "loss": 0.0347,
+      "step": 5852
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.472440914145846e-06,
+      "loss": 0.0329,
+      "step": 5853
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.469261235472165e-06,
+      "loss": 0.0378,
+      "step": 5854
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.466081830150414e-06,
+      "loss": 0.0333,
+      "step": 5855
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.4629026985239996e-06,
+      "loss": 0.0381,
+      "step": 5856
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 7.4597238409363145e-06,
+      "loss": 0.0288,
+      "step": 5857
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.4565452577307165e-06,
+      "loss": 0.039,
+      "step": 5858
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.453366949250526e-06,
+      "loss": 0.039,
+      "step": 5859
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.450188915839047e-06,
+      "loss": 0.0361,
+      "step": 5860
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.447011157839541e-06,
+      "loss": 0.0363,
+      "step": 5861
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.443833675595254e-06,
+      "loss": 0.0402,
+      "step": 5862
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 7.440656469449392e-06,
+      "loss": 0.0384,
+      "step": 5863
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.437479539745127e-06,
+      "loss": 0.0315,
+      "step": 5864
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.434302886825618e-06,
+      "loss": 0.0392,
+      "step": 5865
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.431126511033977e-06,
+      "loss": 0.0301,
+      "step": 5866
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.4279504127132974e-06,
+      "loss": 0.036,
+      "step": 5867
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.424774592206635e-06,
+      "loss": 0.0541,
+      "step": 5868
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.421599049857016e-06,
+      "loss": 0.0334,
+      "step": 5869
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 7.418423786007445e-06,
+      "loss": 0.037,
+      "step": 5870
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.415248801000886e-06,
+      "loss": 0.0309,
+      "step": 5871
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.412074095180284e-06,
+      "loss": 0.037,
+      "step": 5872
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.408899668888541e-06,
+      "loss": 0.0391,
+      "step": 5873
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.405725522468542e-06,
+      "loss": 0.0311,
+      "step": 5874
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.402551656263131e-06,
+      "loss": 0.0344,
+      "step": 5875
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 7.39937807061512e-06,
+      "loss": 0.0342,
+      "step": 5876
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.396204765867307e-06,
+      "loss": 0.031,
+      "step": 5877
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.393031742362442e-06,
+      "loss": 0.0386,
+      "step": 5878
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.389859000443259e-06,
+      "loss": 0.0301,
+      "step": 5879
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.38668654045245e-06,
+      "loss": 0.0431,
+      "step": 5880
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.383514362732679e-06,
+      "loss": 0.0343,
+      "step": 5881
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.3803424676265844e-06,
+      "loss": 0.0309,
+      "step": 5882
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 7.377170855476767e-06,
+      "loss": 0.0355,
+      "step": 5883
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.37399952662581e-06,
+      "loss": 0.0331,
+      "step": 5884
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.37082848141625e-06,
+      "loss": 0.0328,
+      "step": 5885
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.367657720190597e-06,
+      "loss": 0.0359,
+      "step": 5886
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.364487243291345e-06,
+      "loss": 0.0365,
+      "step": 5887
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.361317051060936e-06,
+      "loss": 0.043,
+      "step": 5888
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.358147143841796e-06,
+      "loss": 0.0295,
+      "step": 5889
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 7.354977521976312e-06,
+      "loss": 0.0333,
+      "step": 5890
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.351808185806842e-06,
+      "loss": 0.0423,
+      "step": 5891
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.348639135675722e-06,
+      "loss": 0.038,
+      "step": 5892
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.34547037192524e-06,
+      "loss": 0.0376,
+      "step": 5893
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.342301894897672e-06,
+      "loss": 0.0343,
+      "step": 5894
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.339133704935246e-06,
+      "loss": 0.0394,
+      "step": 5895
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 7.335965802380174e-06,
+      "loss": 0.0345,
+      "step": 5896
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.332798187574626e-06,
+      "loss": 0.045,
+      "step": 5897
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.3296308608607394e-06,
+      "loss": 0.0418,
+      "step": 5898
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.326463822580636e-06,
+      "loss": 0.0303,
+      "step": 5899
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.323297073076385e-06,
+      "loss": 0.0364,
+      "step": 5900
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.320130612690048e-06,
+      "loss": 0.031,
+      "step": 5901
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.3169644417636354e-06,
+      "loss": 0.0316,
+      "step": 5902
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 7.313798560639134e-06,
+      "loss": 0.0314,
+      "step": 5903
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.3106329696585e-06,
+      "loss": 0.0389,
+      "step": 5904
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.307467669163655e-06,
+      "loss": 0.0314,
+      "step": 5905
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.304302659496499e-06,
+      "loss": 0.0383,
+      "step": 5906
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.301137940998886e-06,
+      "loss": 0.0366,
+      "step": 5907
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.297973514012645e-06,
+      "loss": 0.0346,
+      "step": 5908
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 7.294809378879583e-06,
+      "loss": 0.0428,
+      "step": 5909
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.291645535941457e-06,
+      "loss": 0.0355,
+      "step": 5910
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.2884819855400106e-06,
+      "loss": 0.0365,
+      "step": 5911
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.285318728016942e-06,
+      "loss": 0.0337,
+      "step": 5912
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 7.282155763713921e-06,
+      "loss": 0.0319,
+      "step": 5913
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 9855,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 15,
+  "save_steps": 3.0,
+  "total_flos": 116657414369280.0,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}