diff --git "a/checkpoint-11212/trainer_state.json" "b/checkpoint-11212/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11212/trainer_state.json"
@@ -0,0 +1,67324 @@
+{
+  "best_metric": 0.34804627299308777,
+  "best_model_checkpoint": "./medium-article-titles-engagement/checkpoint-8409",
+  "epoch": 4.0,
+  "global_step": 11212,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.7159,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.7009,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 0.7291,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.7201,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 0.7007,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-06,
+      "loss": 0.7034,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 0.74,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6726,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-06,
+      "loss": 0.7118,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 0.7025,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.7034,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.6956,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.6789,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.6891,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-05,
+      "loss": 0.6971,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.7049,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.6963,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-05,
+      "loss": 0.6857,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9e-05,
+      "loss": 0.7105,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 0.6904,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-05,
+      "loss": 0.694,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.7006,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.6932,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4e-05,
+      "loss": 0.6898,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-05,
+      "loss": 0.6834,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.694,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.6915,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.6742,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-05,
+      "loss": 0.6847,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-05,
+      "loss": 0.6809,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1e-05,
+      "loss": 0.7081,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.6871,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3e-05,
+      "loss": 0.6796,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.6808,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-05,
+      "loss": 0.706,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-05,
+      "loss": 0.6854,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-05,
+      "loss": 0.691,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8e-05,
+      "loss": 0.6791,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.6871,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.7009,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1e-05,
+      "loss": 0.6879,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2e-05,
+      "loss": 0.6903,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-05,
+      "loss": 0.7009,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.6622,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-05,
+      "loss": 0.6796,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.6778,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-05,
+      "loss": 0.7009,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8e-05,
+      "loss": 0.6946,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9e-05,
+      "loss": 0.6546,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-05,
+      "loss": 0.6836,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9996419620479776e-05,
+      "loss": 0.6608,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.999283924095954e-05,
+      "loss": 0.702,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9989258861439315e-05,
+      "loss": 0.7556,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.998567848191908e-05,
+      "loss": 0.6576,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9982098102398855e-05,
+      "loss": 0.6991,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997851772287863e-05,
+      "loss": 0.6654,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9974937343358394e-05,
+      "loss": 0.7093,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997135696383817e-05,
+      "loss": 0.7182,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996777658431794e-05,
+      "loss": 0.6627,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996419620479771e-05,
+      "loss": 0.6947,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996061582527748e-05,
+      "loss": 0.677,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9957035445757254e-05,
+      "loss": 0.652,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995345506623703e-05,
+      "loss": 0.631,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99498746867168e-05,
+      "loss": 0.6458,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9946294307196566e-05,
+      "loss": 0.6994,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994271392767634e-05,
+      "loss": 0.6767,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9939133548156106e-05,
+      "loss": 0.6782,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993555316863588e-05,
+      "loss": 0.6532,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993197278911565e-05,
+      "loss": 0.721,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992839240959542e-05,
+      "loss": 0.6353,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992481203007519e-05,
+      "loss": 0.6671,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992123165055496e-05,
+      "loss": 0.693,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991765127103473e-05,
+      "loss": 0.6639,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9914070891514505e-05,
+      "loss": 0.6729,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991049051199427e-05,
+      "loss": 0.6678,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906910132474044e-05,
+      "loss": 0.6627,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990332975295382e-05,
+      "loss": 0.7264,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9899749373433584e-05,
+      "loss": 0.7034,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989616899391336e-05,
+      "loss": 0.7133,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9892588614393123e-05,
+      "loss": 0.6636,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98890082348729e-05,
+      "loss": 0.6578,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988542785535267e-05,
+      "loss": 0.6693,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9881847475832436e-05,
+      "loss": 0.6566,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987826709631221e-05,
+      "loss": 0.7203,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987468671679198e-05,
+      "loss": 0.6635,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9871106337271756e-05,
+      "loss": 0.6184,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986752595775153e-05,
+      "loss": 0.6922,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9863945578231295e-05,
+      "loss": 0.6691,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986036519871107e-05,
+      "loss": 0.6873,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9856784819190835e-05,
+      "loss": 0.6924,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985320443967061e-05,
+      "loss": 0.6754,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984962406015038e-05,
+      "loss": 0.658,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984604368063015e-05,
+      "loss": 0.6457,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984246330110992e-05,
+      "loss": 0.6697,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9838882921589694e-05,
+      "loss": 0.6244,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.983530254206946e-05,
+      "loss": 0.6171,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9831722162549234e-05,
+      "loss": 0.7583,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9828141783029e-05,
+      "loss": 0.7512,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9824561403508773e-05,
+      "loss": 0.6467,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9820981023988547e-05,
+      "loss": 0.7834,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981740064446831e-05,
+      "loss": 0.6662,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9813820264948086e-05,
+      "loss": 0.693,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981023988542786e-05,
+      "loss": 0.6854,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9806659505907626e-05,
+      "loss": 0.6908,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98030791263874e-05,
+      "loss": 0.6454,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9799498746867165e-05,
+      "loss": 0.7064,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979591836734694e-05,
+      "loss": 0.6956,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979233798782671e-05,
+      "loss": 0.6842,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9788757608306485e-05,
+      "loss": 0.6716,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978517722878626e-05,
+      "loss": 0.6738,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9781596849266025e-05,
+      "loss": 0.6509,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97780164697458e-05,
+      "loss": 0.6746,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977443609022557e-05,
+      "loss": 0.6497,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977085571070534e-05,
+      "loss": 0.652,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976727533118511e-05,
+      "loss": 0.6651,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976369495166488e-05,
+      "loss": 0.6341,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976011457214465e-05,
+      "loss": 0.698,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975653419262442e-05,
+      "loss": 0.6979,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975295381310419e-05,
+      "loss": 0.671,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.974937343358396e-05,
+      "loss": 0.6906,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9745793054063736e-05,
+      "loss": 0.6825,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97422126745435e-05,
+      "loss": 0.6585,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9738632295023276e-05,
+      "loss": 0.6658,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.973505191550304e-05,
+      "loss": 0.6184,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9731471535982815e-05,
+      "loss": 0.6479,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.972789115646259e-05,
+      "loss": 0.7496,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9724310776942355e-05,
+      "loss": 0.604,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972073039742213e-05,
+      "loss": 0.7436,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9717150017901894e-05,
+      "loss": 0.6447,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971356963838167e-05,
+      "loss": 0.7154,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970998925886144e-05,
+      "loss": 0.6345,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9706408879341214e-05,
+      "loss": 0.6996,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970282849982099e-05,
+      "loss": 0.6765,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9699248120300754e-05,
+      "loss": 0.6099,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969566774078053e-05,
+      "loss": 0.6562,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96920873612603e-05,
+      "loss": 0.6532,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9688506981740066e-05,
+      "loss": 0.6839,
+      "step": 137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968492660221984e-05,
+      "loss": 0.6499,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968134622269961e-05,
+      "loss": 0.6182,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967776584317938e-05,
+      "loss": 0.682,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967418546365915e-05,
+      "loss": 0.7618,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967060508413892e-05,
+      "loss": 0.7105,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966702470461869e-05,
+      "loss": 0.6784,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9663444325098465e-05,
+      "loss": 0.7058,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965986394557823e-05,
+      "loss": 0.6746,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9656283566058005e-05,
+      "loss": 0.6963,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965270318653777e-05,
+      "loss": 0.6886,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9649122807017544e-05,
+      "loss": 0.6601,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964554242749732e-05,
+      "loss": 0.6388,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9641962047977084e-05,
+      "loss": 0.7187,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963838166845686e-05,
+      "loss": 0.6441,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963480128893663e-05,
+      "loss": 0.635,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96312209094164e-05,
+      "loss": 0.6421,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962764052989617e-05,
+      "loss": 0.6535,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9624060150375936e-05,
+      "loss": 0.6718,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9620479770855716e-05,
+      "loss": 0.6475,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961689939133549e-05,
+      "loss": 0.6324,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9613319011815256e-05,
+      "loss": 0.6297,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960973863229503e-05,
+      "loss": 0.6823,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9606158252774796e-05,
+      "loss": 0.7251,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960257787325457e-05,
+      "loss": 0.7437,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959899749373434e-05,
+      "loss": 0.6722,
+      "step": 162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959541711421411e-05,
+      "loss": 0.6506,
+      "step": 163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959183673469388e-05,
+      "loss": 0.7161,
+      "step": 164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958825635517365e-05,
+      "loss": 0.6613,
+      "step": 165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958467597565342e-05,
+      "loss": 0.6492,
+      "step": 166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9581095596133194e-05,
+      "loss": 0.6708,
+      "step": 167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957751521661296e-05,
+      "loss": 0.75,
+      "step": 168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9573934837092734e-05,
+      "loss": 0.6443,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957035445757251e-05,
+      "loss": 0.6741,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9566774078052273e-05,
+      "loss": 0.6885,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956319369853205e-05,
+      "loss": 0.6739,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955961331901181e-05,
+      "loss": 0.6247,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9556032939491586e-05,
+      "loss": 0.704,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955245255997136e-05,
+      "loss": 0.6588,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9548872180451126e-05,
+      "loss": 0.7004,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95452918009309e-05,
+      "loss": 0.6415,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954171142141067e-05,
+      "loss": 0.6693,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9538131041890445e-05,
+      "loss": 0.672,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953455066237022e-05,
+      "loss": 0.635,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9530970282849985e-05,
+      "loss": 0.6926,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952738990332976e-05,
+      "loss": 0.6558,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9523809523809525e-05,
+      "loss": 0.6828,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.95202291442893e-05,
+      "loss": 0.7209,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951664876476907e-05,
+      "loss": 0.6179,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.951306838524884e-05,
+      "loss": 0.7027,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950948800572861e-05,
+      "loss": 0.6311,
+      "step": 187
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9505907626208384e-05,
+      "loss": 0.6565,
+      "step": 188
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.950232724668815e-05,
+      "loss": 0.7102,
+      "step": 189
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9498746867167923e-05,
+      "loss": 0.6111,
+      "step": 190
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949516648764769e-05,
+      "loss": 0.6164,
+      "step": 191
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.949158610812746e-05,
+      "loss": 0.6465,
+      "step": 192
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9488005728607236e-05,
+      "loss": 0.6344,
+      "step": 193
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9484425349087e-05,
+      "loss": 0.6496,
+      "step": 194
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9480844969566776e-05,
+      "loss": 0.6155,
+      "step": 195
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947726459004655e-05,
+      "loss": 0.6761,
+      "step": 196
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9473684210526315e-05,
+      "loss": 0.6198,
+      "step": 197
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947010383100609e-05,
+      "loss": 0.7867,
+      "step": 198
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9466523451485855e-05,
+      "loss": 0.6036,
+      "step": 199
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946294307196563e-05,
+      "loss": 0.5905,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94593626924454e-05,
+      "loss": 0.7537,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9455782312925175e-05,
+      "loss": 0.6287,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945220193340495e-05,
+      "loss": 0.6808,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9448621553884714e-05,
+      "loss": 0.7303,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944504117436449e-05,
+      "loss": 0.6118,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944146079484426e-05,
+      "loss": 0.6998,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943788041532403e-05,
+      "loss": 0.6566,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94343000358038e-05,
+      "loss": 0.6808,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9430719656283567e-05,
+      "loss": 0.613,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942713927676334e-05,
+      "loss": 0.6322,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.942355889724311e-05,
+      "loss": 0.6834,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941997851772288e-05,
+      "loss": 0.6896,
+      "step": 212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.941639813820265e-05,
+      "loss": 0.6805,
+      "step": 213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9412817758682426e-05,
+      "loss": 0.6569,
+      "step": 214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940923737916219e-05,
+      "loss": 0.6865,
+      "step": 215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9405656999641965e-05,
+      "loss": 0.6639,
+      "step": 216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.940207662012173e-05,
+      "loss": 0.6121,
+      "step": 217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9398496240601505e-05,
+      "loss": 0.6661,
+      "step": 218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.939491586108128e-05,
+      "loss": 0.6537,
+      "step": 219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9391335481561044e-05,
+      "loss": 0.6549,
+      "step": 220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938775510204082e-05,
+      "loss": 0.6223,
+      "step": 221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938417472252059e-05,
+      "loss": 0.6496,
+      "step": 222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.938059434300036e-05,
+      "loss": 0.6532,
+      "step": 223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.937701396348013e-05,
+      "loss": 0.6497,
+      "step": 224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93734335839599e-05,
+      "loss": 0.6836,
+      "step": 225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936985320443968e-05,
+      "loss": 0.6646,
+      "step": 226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.936627282491944e-05,
+      "loss": 0.6123,
+      "step": 227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9362692445399216e-05,
+      "loss": 0.695,
+      "step": 228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935911206587899e-05,
+      "loss": 0.576,
+      "step": 229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9355531686358756e-05,
+      "loss": 0.624,
+      "step": 230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.935195130683853e-05,
+      "loss": 0.7198,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.93483709273183e-05,
+      "loss": 0.6552,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934479054779807e-05,
+      "loss": 0.688,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.934121016827784e-05,
+      "loss": 0.6205,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933762978875761e-05,
+      "loss": 0.6495,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.933404940923738e-05,
+      "loss": 0.6722,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9330469029717155e-05,
+      "loss": 0.6437,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.932688865019692e-05,
+      "loss": 0.6648,
+      "step": 238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9323308270676694e-05,
+      "loss": 0.629,
+      "step": 239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931972789115647e-05,
+      "loss": 0.7004,
+      "step": 240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9316147511636234e-05,
+      "loss": 0.6584,
+      "step": 241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.931256713211601e-05,
+      "loss": 0.6705,
+      "step": 242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9308986752595774e-05,
+      "loss": 0.6621,
+      "step": 243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930540637307555e-05,
+      "loss": 0.6223,
+      "step": 244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.930182599355532e-05,
+      "loss": 0.688,
+      "step": 245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9298245614035086e-05,
+      "loss": 0.5823,
+      "step": 246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.929466523451486e-05,
+      "loss": 0.6132,
+      "step": 247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9291084854994626e-05,
+      "loss": 0.683,
+      "step": 248
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9287504475474406e-05,
+      "loss": 0.6604,
+      "step": 249
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.928392409595418e-05,
+      "loss": 0.6685,
+      "step": 250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9280343716433946e-05,
+      "loss": 0.6335,
+      "step": 251
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.927676333691372e-05,
+      "loss": 0.7343,
+      "step": 252
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9273182957393485e-05,
+      "loss": 0.6402,
+      "step": 253
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926960257787326e-05,
+      "loss": 0.6632,
+      "step": 254
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.926602219835303e-05,
+      "loss": 0.6252,
+      "step": 255
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92624418188328e-05,
+      "loss": 0.6656,
+      "step": 256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925886143931257e-05,
+      "loss": 0.6923,
+      "step": 257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925528105979234e-05,
+      "loss": 0.657,
+      "step": 258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.925170068027211e-05,
+      "loss": 0.672,
+      "step": 259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9248120300751884e-05,
+      "loss": 0.6562,
+      "step": 260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.924453992123165e-05,
+      "loss": 0.6776,
+      "step": 261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9240959541711423e-05,
+      "loss": 0.6536,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92373791621912e-05,
+      "loss": 0.6506,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.923379878267096e-05,
+      "loss": 0.6681,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9230218403150736e-05,
+      "loss": 0.6001,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.92266380236305e-05,
+      "loss": 0.6295,
+      "step": 266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9223057644110276e-05,
+      "loss": 0.6362,
+      "step": 267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921947726459005e-05,
+      "loss": 0.6425,
+      "step": 268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9215896885069815e-05,
+      "loss": 0.6809,
+      "step": 269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.921231650554959e-05,
+      "loss": 0.6225,
+      "step": 270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920873612602936e-05,
+      "loss": 0.6077,
+      "step": 271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9205155746509135e-05,
+      "loss": 0.6758,
+      "step": 272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.920157536698891e-05,
+      "loss": 0.7356,
+      "step": 273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9197994987468675e-05,
+      "loss": 0.6248,
+      "step": 274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.919441460794845e-05,
+      "loss": 0.61,
+      "step": 275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9190834228428214e-05,
+      "loss": 0.6932,
+      "step": 276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918725384890799e-05,
+      "loss": 0.5455,
+      "step": 277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918367346938776e-05,
+      "loss": 0.7632,
+      "step": 278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.918009308986753e-05,
+      "loss": 0.7385,
+      "step": 279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.91765127103473e-05,
+      "loss": 0.6461,
+      "step": 280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9172932330827073e-05,
+      "loss": 0.6154,
+      "step": 281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916935195130684e-05,
+      "loss": 0.6747,
+      "step": 282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916577157178661e-05,
+      "loss": 0.6463,
+      "step": 283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.916219119226638e-05,
+      "loss": 0.5522,
+      "step": 284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915861081274615e-05,
+      "loss": 0.7031,
+      "step": 285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9155030433225926e-05,
+      "loss": 0.6883,
+      "step": 286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.915145005370569e-05,
+      "loss": 0.6796,
+      "step": 287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9147869674185465e-05,
+      "loss": 0.6623,
+      "step": 288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.914428929466524e-05,
+      "loss": 0.6417,
+      "step": 289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9140708915145005e-05,
+      "loss": 0.6574,
+      "step": 290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.913712853562478e-05,
+      "loss": 0.6741,
+      "step": 291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9133548156104545e-05,
+      "loss": 0.6445,
+      "step": 292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912996777658432e-05,
+      "loss": 0.6071,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.912638739706409e-05,
+      "loss": 0.6877,
+      "step": 294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 0.6561,
+      "step": 295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911922663802364e-05,
+      "loss": 0.6447,
+      "step": 296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9115646258503404e-05,
+      "loss": 0.6917,
+      "step": 297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.911206587898318e-05,
+      "loss": 0.6759,
+      "step": 298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910848549946295e-05,
+      "loss": 0.6906,
+      "step": 299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9104905119942717e-05,
+      "loss": 0.671,
+      "step": 300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.910132474042249e-05,
+      "loss": 0.6472,
+      "step": 301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9097744360902256e-05,
+      "loss": 0.6306,
+      "step": 302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.909416398138203e-05,
+      "loss": 0.6318,
+      "step": 303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.90905836018618e-05,
+      "loss": 0.7244,
+      "step": 304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908700322234157e-05,
+      "loss": 0.6813,
+      "step": 305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.908342284282134e-05,
+      "loss": 0.7184,
+      "step": 306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9079842463301115e-05,
+      "loss": 0.6831,
+      "step": 307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.907626208378088e-05,
+      "loss": 0.6572,
+      "step": 308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9072681704260655e-05,
+      "loss": 0.6687,
+      "step": 309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.906910132474042e-05,
+      "loss": 0.6577,
+      "step": 310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9065520945220194e-05,
+      "loss": 0.7001,
+      "step": 311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.906194056569997e-05,
+      "loss": 0.6641,
+      "step": 312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9058360186179734e-05,
+      "loss": 0.6668,
+      "step": 313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.905477980665951e-05,
+      "loss": 0.6482,
+      "step": 314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.905119942713928e-05,
+      "loss": 0.6939,
+      "step": 315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.904761904761905e-05,
+      "loss": 0.636,
+      "step": 316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.904403866809882e-05,
+      "loss": 0.685,
+      "step": 317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9040458288578586e-05,
+      "loss": 0.6946,
+      "step": 318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9036877909058366e-05,
+      "loss": 0.6496,
+      "step": 319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.903329752953813e-05,
+      "loss": 0.6315,
+      "step": 320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.9029717150017906e-05,
+      "loss": 0.7018,
+      "step": 321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.902613677049768e-05,
+      "loss": 0.6714,
+      "step": 322
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.9022556390977446e-05,
+      "loss": 0.6552,
+      "step": 323
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901897601145722e-05,
+      "loss": 0.6131,
+      "step": 324
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901539563193699e-05,
+      "loss": 0.61,
+      "step": 325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.901181525241676e-05,
+      "loss": 0.6432,
+      "step": 326
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900823487289653e-05,
+      "loss": 0.687,
+      "step": 327
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.90046544933763e-05,
+      "loss": 0.6628,
+      "step": 328
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.900107411385607e-05,
+      "loss": 0.5912,
+      "step": 329
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8997493734335844e-05,
+      "loss": 0.7056,
+      "step": 330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.899391335481561e-05,
+      "loss": 0.6474,
+      "step": 331
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8990332975295384e-05,
+      "loss": 0.6698,
+      "step": 332
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.898675259577516e-05,
+      "loss": 0.7126,
+      "step": 333
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8983172216254924e-05,
+      "loss": 0.6747,
+      "step": 334
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.89795918367347e-05,
+      "loss": 0.6181,
+      "step": 335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.897601145721446e-05,
+      "loss": 0.6507,
+      "step": 336
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8972431077694236e-05,
+      "loss": 0.6724,
+      "step": 337
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896885069817401e-05,
+      "loss": 0.6737,
+      "step": 338
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8965270318653776e-05,
+      "loss": 0.6476,
+      "step": 339
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.896168993913355e-05,
+      "loss": 0.6422,
+      "step": 340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8958109559613316e-05,
+      "loss": 0.6096,
+      "step": 341
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8954529180093096e-05,
+      "loss": 0.5987,
+      "step": 342
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.895094880057287e-05,
+      "loss": 0.6941,
+      "step": 343
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8947368421052635e-05,
+      "loss": 0.6435,
+      "step": 344
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.894378804153241e-05,
+      "loss": 0.6444,
+      "step": 345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8940207662012175e-05,
+      "loss": 0.6439,
+      "step": 346
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893662728249195e-05,
+      "loss": 0.6361,
+      "step": 347
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.893304690297172e-05,
+      "loss": 0.6791,
+      "step": 348
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892946652345149e-05,
+      "loss": 0.6774,
+      "step": 349
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.892588614393126e-05,
+      "loss": 0.6584,
+      "step": 350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8922305764411034e-05,
+      "loss": 0.5579,
+      "step": 351
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.89187253848908e-05,
+      "loss": 0.68,
+      "step": 352
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8915145005370573e-05,
+      "loss": 0.6111,
+      "step": 353
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.891156462585034e-05,
+      "loss": 0.7083,
+      "step": 354
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.890798424633011e-05,
+      "loss": 0.5948,
+      "step": 355
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8904403866809886e-05,
+      "loss": 0.6101,
+      "step": 356
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.890082348728965e-05,
+      "loss": 0.659,
+      "step": 357
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8897243107769426e-05,
+      "loss": 0.5919,
+      "step": 358
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.889366272824919e-05,
+      "loss": 0.6367,
+      "step": 359
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8890082348728965e-05,
+      "loss": 0.5942,
+      "step": 360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.888650196920874e-05,
+      "loss": 0.7403,
+      "step": 361
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8882921589688505e-05,
+      "loss": 0.6494,
+      "step": 362
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.887934121016828e-05,
+      "loss": 0.6883,
+      "step": 363
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.887576083064805e-05,
+      "loss": 0.7156,
+      "step": 364
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.887218045112782e-05,
+      "loss": 0.6999,
+      "step": 365
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.88686000716076e-05,
+      "loss": 0.621,
+      "step": 366
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8865019692087364e-05,
+      "loss": 0.6363,
+      "step": 367
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.886143931256714e-05,
+      "loss": 0.736,
+      "step": 368
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8857858933046904e-05,
+      "loss": 0.6141,
+      "step": 369
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.885427855352668e-05,
+      "loss": 0.634,
+      "step": 370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.885069817400645e-05,
+      "loss": 0.6934,
+      "step": 371
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.884711779448622e-05,
+      "loss": 0.6729,
+      "step": 372
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.884353741496599e-05,
+      "loss": 0.6844,
+      "step": 373
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883995703544576e-05,
+      "loss": 0.6387,
+      "step": 374
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.883637665592553e-05,
+      "loss": 0.6455,
+      "step": 375
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.88327962764053e-05,
+      "loss": 0.6749,
+      "step": 376
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882921589688507e-05,
+      "loss": 0.6524,
+      "step": 377
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.882563551736484e-05,
+      "loss": 0.6688,
+      "step": 378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8822055137844615e-05,
+      "loss": 0.6294,
+      "step": 379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.881847475832438e-05,
+      "loss": 0.6947,
+      "step": 380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8814894378804155e-05,
+      "loss": 0.6572,
+      "step": 381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.881131399928393e-05,
+      "loss": 0.6082,
+      "step": 382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8807733619763695e-05,
+      "loss": 0.6239,
+      "step": 383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.880415324024347e-05,
+      "loss": 0.6248,
+      "step": 384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8800572860723234e-05,
+      "loss": 0.6568,
+      "step": 385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.879699248120301e-05,
+      "loss": 0.6226,
+      "step": 386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.879341210168278e-05,
+      "loss": 0.6218,
+      "step": 387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.878983172216255e-05,
+      "loss": 0.7277,
+      "step": 388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.878625134264233e-05,
+      "loss": 0.6453,
+      "step": 389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.878267096312209e-05,
+      "loss": 0.6878,
+      "step": 390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8779090583601867e-05,
+      "loss": 0.6557,
+      "step": 391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.877551020408164e-05,
+      "loss": 0.6571,
+      "step": 392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8771929824561406e-05,
+      "loss": 0.7255,
+      "step": 393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.876834944504118e-05,
+      "loss": 0.6093,
+      "step": 394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8764769065520946e-05,
+      "loss": 0.6427,
+      "step": 395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.876118868600072e-05,
+      "loss": 0.6775,
+      "step": 396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.875760830648049e-05,
+      "loss": 0.6224,
+      "step": 397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.875402792696026e-05,
+      "loss": 0.6647,
+      "step": 398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.875044754744003e-05,
+      "loss": 0.6915,
+      "step": 399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8746867167919805e-05,
+      "loss": 0.6612,
+      "step": 400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.874328678839957e-05,
+      "loss": 0.6733,
+      "step": 401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8739706408879344e-05,
+      "loss": 0.704,
+      "step": 402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.873612602935911e-05,
+      "loss": 0.6456,
+      "step": 403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8732545649838884e-05,
+      "loss": 0.6645,
+      "step": 404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.872896527031866e-05,
+      "loss": 0.6658,
+      "step": 405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8725384890798424e-05,
+      "loss": 0.5848,
+      "step": 406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.87218045112782e-05,
+      "loss": 0.6454,
+      "step": 407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.871822413175797e-05,
+      "loss": 0.6447,
+      "step": 408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8714643752237736e-05,
+      "loss": 0.6263,
+      "step": 409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.871106337271751e-05,
+      "loss": 0.6417,
+      "step": 410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8707482993197276e-05,
+      "loss": 0.6794,
+      "step": 411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8703902613677056e-05,
+      "loss": 0.6687,
+      "step": 412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.870032223415682e-05,
+      "loss": 0.6588,
+      "step": 413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8696741854636596e-05,
+      "loss": 0.6687,
+      "step": 414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.869316147511637e-05,
+      "loss": 0.6341,
+      "step": 415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8689581095596135e-05,
+      "loss": 0.6101,
+      "step": 416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868600071607591e-05,
+      "loss": 0.5921,
+      "step": 417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.868242033655568e-05,
+      "loss": 0.6727,
+      "step": 418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.867883995703545e-05,
+      "loss": 0.6331,
+      "step": 419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.867525957751522e-05,
+      "loss": 0.7366,
+      "step": 420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.867167919799499e-05,
+      "loss": 0.6447,
+      "step": 421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.866809881847476e-05,
+      "loss": 0.6373,
+      "step": 422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8664518438954534e-05,
+      "loss": 0.716,
+      "step": 423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86609380594343e-05,
+      "loss": 0.6327,
+      "step": 424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8657357679914074e-05,
+      "loss": 0.6562,
+      "step": 425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.865377730039385e-05,
+      "loss": 0.6641,
+      "step": 426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.865019692087361e-05,
+      "loss": 0.6815,
+      "step": 427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8646616541353386e-05,
+      "loss": 0.6255,
+      "step": 428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.864303616183315e-05,
+      "loss": 0.6015,
+      "step": 429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8639455782312926e-05,
+      "loss": 0.6325,
+      "step": 430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.86358754027927e-05,
+      "loss": 0.7066,
+      "step": 431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8632295023272466e-05,
+      "loss": 0.6477,
+      "step": 432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.862871464375224e-05,
+      "loss": 0.6505,
+      "step": 433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8625134264232005e-05,
+      "loss": 0.6186,
+      "step": 434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.862155388471178e-05,
+      "loss": 0.6616,
+      "step": 435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.861797350519156e-05,
+      "loss": 0.5488,
+      "step": 436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8614393125671325e-05,
+      "loss": 0.6712,
+      "step": 437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.86108127461511e-05,
+      "loss": 0.6749,
+      "step": 438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8607232366630864e-05,
+      "loss": 0.5795,
+      "step": 439
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.860365198711064e-05,
+      "loss": 0.653,
+      "step": 440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.860007160759041e-05,
+      "loss": 0.7009,
+      "step": 441
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.859649122807018e-05,
+      "loss": 0.6194,
+      "step": 442
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.859291084854995e-05,
+      "loss": 0.6678,
+      "step": 443
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8589330469029724e-05,
+      "loss": 0.647,
+      "step": 444
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.858575008950949e-05,
+      "loss": 0.6413,
+      "step": 445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.858216970998926e-05,
+      "loss": 0.6327,
+      "step": 446
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.857858933046903e-05,
+      "loss": 0.6109,
+      "step": 447
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.85750089509488e-05,
+      "loss": 0.6762,
+      "step": 448
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8571428571428576e-05,
+      "loss": 0.7022,
+      "step": 449
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.856784819190834e-05,
+      "loss": 0.6419,
+      "step": 450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8564267812388115e-05,
+      "loss": 0.6153,
+      "step": 451
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.856068743286788e-05,
+      "loss": 0.6892,
+      "step": 452
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8557107053347655e-05,
+      "loss": 0.6952,
+      "step": 453
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.855352667382743e-05,
+      "loss": 0.583,
+      "step": 454
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8549946294307195e-05,
+      "loss": 0.6318,
+      "step": 455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.854636591478697e-05,
+      "loss": 0.6183,
+      "step": 456
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.854278553526674e-05,
+      "loss": 0.5793,
+      "step": 457
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853920515574651e-05,
+      "loss": 0.6385,
+      "step": 458
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.853562477622629e-05,
+      "loss": 0.664,
+      "step": 459
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.8532044396706054e-05,
+      "loss": 0.632,
+      "step": 460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.852846401718583e-05,
+      "loss": 0.6475,
+      "step": 461
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.85248836376656e-05,
+      "loss": 0.6592,
+      "step": 462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.852130325814537e-05,
+      "loss": 0.6733,
+      "step": 463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.851772287862514e-05,
+      "loss": 0.6162,
+      "step": 464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8514142499104906e-05,
+      "loss": 0.618,
+      "step": 465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.851056211958468e-05,
+      "loss": 0.6883,
+      "step": 466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.850698174006445e-05,
+      "loss": 0.6399,
+      "step": 467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.850340136054422e-05,
+      "loss": 0.6409,
+      "step": 468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.849982098102399e-05,
+      "loss": 0.6838,
+      "step": 469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.849624060150376e-05,
+      "loss": 0.6646,
+      "step": 470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.849266022198353e-05,
+      "loss": 0.5508,
+      "step": 471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8489079842463305e-05,
+      "loss": 0.5931,
+      "step": 472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.848549946294307e-05,
+      "loss": 0.66,
+      "step": 473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8481919083422845e-05,
+      "loss": 0.6226,
+      "step": 474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.847833870390262e-05,
+      "loss": 0.6174,
+      "step": 475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8474758324382384e-05,
+      "loss": 0.6376,
+      "step": 476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.847117794486216e-05,
+      "loss": 0.5678,
+      "step": 477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8467597565341924e-05,
+      "loss": 0.687,
+      "step": 478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.84640171858217e-05,
+      "loss": 0.6057,
+      "step": 479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.846043680630147e-05,
+      "loss": 0.6648,
+      "step": 480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8456856426781237e-05,
+      "loss": 0.6361,
+      "step": 481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.845327604726101e-05,
+      "loss": 0.5769,
+      "step": 482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.844969566774078e-05,
+      "loss": 0.5918,
+      "step": 483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8446115288220556e-05,
+      "loss": 0.6066,
+      "step": 484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.844253490870033e-05,
+      "loss": 0.6267,
+      "step": 485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8438954529180096e-05,
+      "loss": 0.6467,
+      "step": 486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.843537414965987e-05,
+      "loss": 0.6491,
+      "step": 487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.8431793770139635e-05,
+      "loss": 0.8145,
+      "step": 488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842821339061941e-05,
+      "loss": 0.742,
+      "step": 489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.842463301109918e-05,
+      "loss": 0.6979,
+      "step": 490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.842105263157895e-05,
+      "loss": 0.6173,
+      "step": 491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.841747225205872e-05,
+      "loss": 0.5638,
+      "step": 492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8413891872538494e-05,
+      "loss": 0.7286,
+      "step": 493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.841031149301826e-05,
+      "loss": 0.687,
+      "step": 494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8406731113498034e-05,
+      "loss": 0.6512,
+      "step": 495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.84031507339778e-05,
+      "loss": 0.6249,
+      "step": 496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8399570354457574e-05,
+      "loss": 0.6255,
+      "step": 497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.839598997493735e-05,
+      "loss": 0.6909,
+      "step": 498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.839240959541711e-05,
+      "loss": 0.6348,
+      "step": 499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8388829215896886e-05,
+      "loss": 0.6658,
+      "step": 500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.838524883637666e-05,
+      "loss": 0.644,
+      "step": 501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8381668456856426e-05,
+      "loss": 0.6356,
+      "step": 502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83780880773362e-05,
+      "loss": 0.6128,
+      "step": 503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8374507697815966e-05,
+      "loss": 0.6952,
+      "step": 504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.837092731829574e-05,
+      "loss": 0.6279,
+      "step": 505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.836734693877551e-05,
+      "loss": 0.6199,
+      "step": 506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8363766559255285e-05,
+      "loss": 0.624,
+      "step": 507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.836018617973506e-05,
+      "loss": 0.5719,
+      "step": 508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8356605800214825e-05,
+      "loss": 0.6383,
+      "step": 509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83530254206946e-05,
+      "loss": 0.7009,
+      "step": 510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.834944504117437e-05,
+      "loss": 0.6177,
+      "step": 511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.834586466165414e-05,
+      "loss": 0.6602,
+      "step": 512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.834228428213391e-05,
+      "loss": 0.6886,
+      "step": 513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.833870390261368e-05,
+      "loss": 0.6373,
+      "step": 514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.833512352309345e-05,
+      "loss": 0.6096,
+      "step": 515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.8331543143573224e-05,
+      "loss": 0.6428,
+      "step": 516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832796276405299e-05,
+      "loss": 0.624,
+      "step": 517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.832438238453276e-05,
+      "loss": 0.5812,
+      "step": 518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8320802005012536e-05,
+      "loss": 0.6127,
+      "step": 519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.83172216254923e-05,
+      "loss": 0.5992,
+      "step": 520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8313641245972076e-05,
+      "loss": 0.6678,
+      "step": 521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.831006086645184e-05,
+      "loss": 0.6842,
+      "step": 522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8306480486931616e-05,
+      "loss": 0.7032,
+      "step": 523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.830290010741139e-05,
+      "loss": 0.6641,
+      "step": 524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8299319727891155e-05,
+      "loss": 0.7118,
+      "step": 525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.829573934837093e-05,
+      "loss": 0.7305,
+      "step": 526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8292158968850695e-05,
+      "loss": 0.6732,
+      "step": 527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.828857858933047e-05,
+      "loss": 0.6677,
+      "step": 528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.828499820981025e-05,
+      "loss": 0.6551,
+      "step": 529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8281417830290014e-05,
+      "loss": 0.6346,
+      "step": 530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.827783745076979e-05,
+      "loss": 0.6542,
+      "step": 531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8274257071249554e-05,
+      "loss": 0.6681,
+      "step": 532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.827067669172933e-05,
+      "loss": 0.7053,
+      "step": 533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.82670963122091e-05,
+      "loss": 0.6796,
+      "step": 534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.826351593268887e-05,
+      "loss": 0.6098,
+      "step": 535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.825993555316864e-05,
+      "loss": 0.6273,
+      "step": 536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.825635517364841e-05,
+      "loss": 0.645,
+      "step": 537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.825277479412818e-05,
+      "loss": 0.6251,
+      "step": 538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.824919441460795e-05,
+      "loss": 0.6119,
+      "step": 539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.824561403508772e-05,
+      "loss": 0.6612,
+      "step": 540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.824203365556749e-05,
+      "loss": 0.6999,
+      "step": 541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8238453276047265e-05,
+      "loss": 0.5999,
+      "step": 542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.823487289652703e-05,
+      "loss": 0.6121,
+      "step": 543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8231292517006805e-05,
+      "loss": 0.7019,
+      "step": 544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.822771213748657e-05,
+      "loss": 0.6551,
+      "step": 545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.8224131757966345e-05,
+      "loss": 0.6495,
+      "step": 546
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.822055137844612e-05,
+      "loss": 0.6392,
+      "step": 547
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8216970998925884e-05,
+      "loss": 0.6866,
+      "step": 548
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.821339061940566e-05,
+      "loss": 0.5955,
+      "step": 549
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.820981023988543e-05,
+      "loss": 0.6209,
+      "step": 550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.82062298603652e-05,
+      "loss": 0.6364,
+      "step": 551
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.820264948084497e-05,
+      "loss": 0.6199,
+      "step": 552
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8199069101324743e-05,
+      "loss": 0.7362,
+      "step": 553
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.819548872180452e-05,
+      "loss": 0.6252,
+      "step": 554
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.819190834228429e-05,
+      "loss": 0.6603,
+      "step": 555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8188327962764056e-05,
+      "loss": 0.6274,
+      "step": 556
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.818474758324383e-05,
+      "loss": 0.6248,
+      "step": 557
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8181167203723596e-05,
+      "loss": 0.6941,
+      "step": 558
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.817758682420337e-05,
+      "loss": 0.6473,
+      "step": 559
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.817400644468314e-05,
+      "loss": 0.6254,
+      "step": 560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.817042606516291e-05,
+      "loss": 0.616,
+      "step": 561
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.816684568564268e-05,
+      "loss": 0.6774,
+      "step": 562
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.816326530612245e-05,
+      "loss": 0.6736,
+      "step": 563
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.815968492660222e-05,
+      "loss": 0.6353,
+      "step": 564
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8156104547081995e-05,
+      "loss": 0.6226,
+      "step": 565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.815252416756176e-05,
+      "loss": 0.7304,
+      "step": 566
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8148943788041534e-05,
+      "loss": 0.608,
+      "step": 567
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.814536340852131e-05,
+      "loss": 0.5938,
+      "step": 568
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8141783029001074e-05,
+      "loss": 0.6176,
+      "step": 569
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.813820264948085e-05,
+      "loss": 0.6421,
+      "step": 570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8134622269960613e-05,
+      "loss": 0.621,
+      "step": 571
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8131041890440387e-05,
+      "loss": 0.6653,
+      "step": 572
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.812746151092016e-05,
+      "loss": 0.6567,
+      "step": 573
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.8123881131399926e-05,
+      "loss": 0.6493,
+      "step": 574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.81203007518797e-05,
+      "loss": 0.6662,
+      "step": 575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.811672037235947e-05,
+      "loss": 0.6493,
+      "step": 576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8113139992839246e-05,
+      "loss": 0.6751,
+      "step": 577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810955961331902e-05,
+      "loss": 0.6376,
+      "step": 578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8105979233798785e-05,
+      "loss": 0.6608,
+      "step": 579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.810239885427856e-05,
+      "loss": 0.6628,
+      "step": 580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8098818474758325e-05,
+      "loss": 0.5823,
+      "step": 581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.80952380952381e-05,
+      "loss": 0.6506,
+      "step": 582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.809165771571787e-05,
+      "loss": 0.6328,
+      "step": 583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.808807733619764e-05,
+      "loss": 0.6182,
+      "step": 584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.808449695667741e-05,
+      "loss": 0.6496,
+      "step": 585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8080916577157184e-05,
+      "loss": 0.6087,
+      "step": 586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807733619763695e-05,
+      "loss": 0.6368,
+      "step": 587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8073755818116724e-05,
+      "loss": 0.737,
+      "step": 588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.807017543859649e-05,
+      "loss": 0.6798,
+      "step": 589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.806659505907626e-05,
+      "loss": 0.635,
+      "step": 590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8063014679556036e-05,
+      "loss": 0.7139,
+      "step": 591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.80594343000358e-05,
+      "loss": 0.644,
+      "step": 592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8055853920515576e-05,
+      "loss": 0.6121,
+      "step": 593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.805227354099535e-05,
+      "loss": 0.6477,
+      "step": 594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8048693161475116e-05,
+      "loss": 0.6887,
+      "step": 595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.804511278195489e-05,
+      "loss": 0.6153,
+      "step": 596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8041532402434655e-05,
+      "loss": 0.5794,
+      "step": 597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.803795202291443e-05,
+      "loss": 0.662,
+      "step": 598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.80343716433942e-05,
+      "loss": 0.6223,
+      "step": 599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8030791263873975e-05,
+      "loss": 0.6186,
+      "step": 600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.802721088435375e-05,
+      "loss": 0.6053,
+      "step": 601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.8023630504833514e-05,
+      "loss": 0.658,
+      "step": 602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.802005012531329e-05,
+      "loss": 0.6329,
+      "step": 603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.801646974579306e-05,
+      "loss": 0.5941,
+      "step": 604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.801288936627283e-05,
+      "loss": 0.6245,
+      "step": 605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.80093089867526e-05,
+      "loss": 0.6253,
+      "step": 606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.800572860723237e-05,
+      "loss": 0.7008,
+      "step": 607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.800214822771214e-05,
+      "loss": 0.629,
+      "step": 608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799856784819191e-05,
+      "loss": 0.6127,
+      "step": 609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799498746867168e-05,
+      "loss": 0.6018,
+      "step": 610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.799140708915145e-05,
+      "loss": 0.6164,
+      "step": 611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7987826709631226e-05,
+      "loss": 0.6795,
+      "step": 612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.798424633011099e-05,
+      "loss": 0.6771,
+      "step": 613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7980665950590766e-05,
+      "loss": 0.5993,
+      "step": 614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.797708557107053e-05,
+      "loss": 0.5903,
+      "step": 615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7973505191550305e-05,
+      "loss": 0.6103,
+      "step": 616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.796992481203008e-05,
+      "loss": 0.6272,
+      "step": 617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7966344432509845e-05,
+      "loss": 0.744,
+      "step": 618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.796276405298962e-05,
+      "loss": 0.7213,
+      "step": 619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795918367346939e-05,
+      "loss": 0.654,
+      "step": 620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795560329394916e-05,
+      "loss": 0.5945,
+      "step": 621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.795202291442893e-05,
+      "loss": 0.6277,
+      "step": 622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7948442534908704e-05,
+      "loss": 0.6289,
+      "step": 623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.794486215538848e-05,
+      "loss": 0.6487,
+      "step": 624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7941281775868244e-05,
+      "loss": 0.6325,
+      "step": 625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793770139634802e-05,
+      "loss": 0.7204,
+      "step": 626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.793412101682779e-05,
+      "loss": 0.6242,
+      "step": 627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.7930540637307556e-05,
+      "loss": 0.6503,
+      "step": 628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.792696025778733e-05,
+      "loss": 0.6597,
+      "step": 629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.79233798782671e-05,
+      "loss": 0.6289,
+      "step": 630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.791979949874687e-05,
+      "loss": 0.6229,
+      "step": 631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.791621911922664e-05,
+      "loss": 0.6277,
+      "step": 632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.791263873970641e-05,
+      "loss": 0.643,
+      "step": 633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.790905836018618e-05,
+      "loss": 0.5702,
+      "step": 634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7905477980665955e-05,
+      "loss": 0.6031,
+      "step": 635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.790189760114572e-05,
+      "loss": 0.6371,
+      "step": 636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7898317221625495e-05,
+      "loss": 0.7021,
+      "step": 637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.789473684210526e-05,
+      "loss": 0.6573,
+      "step": 638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7891156462585034e-05,
+      "loss": 0.6184,
+      "step": 639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.788757608306481e-05,
+      "loss": 0.6831,
+      "step": 640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7883995703544574e-05,
+      "loss": 0.5919,
+      "step": 641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.788041532402435e-05,
+      "loss": 0.6523,
+      "step": 642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.787683494450412e-05,
+      "loss": 0.6552,
+      "step": 643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.787325456498389e-05,
+      "loss": 0.6059,
+      "step": 644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.786967418546366e-05,
+      "loss": 0.6209,
+      "step": 645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.786609380594343e-05,
+      "loss": 0.6632,
+      "step": 646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7862513426423206e-05,
+      "loss": 0.6659,
+      "step": 647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.785893304690298e-05,
+      "loss": 0.6899,
+      "step": 648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7855352667382746e-05,
+      "loss": 0.5999,
+      "step": 649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.785177228786252e-05,
+      "loss": 0.5852,
+      "step": 650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7848191908342285e-05,
+      "loss": 0.6576,
+      "step": 651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.784461152882206e-05,
+      "loss": 0.7359,
+      "step": 652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.784103114930183e-05,
+      "loss": 0.5948,
+      "step": 653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.78374507697816e-05,
+      "loss": 0.6502,
+      "step": 654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.783387039026137e-05,
+      "loss": 0.6524,
+      "step": 655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.783029001074114e-05,
+      "loss": 0.5936,
+      "step": 656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.782670963122091e-05,
+      "loss": 0.6206,
+      "step": 657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.7823129251700684e-05,
+      "loss": 0.6518,
+      "step": 658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.781954887218045e-05,
+      "loss": 0.6619,
+      "step": 659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7815968492660224e-05,
+      "loss": 0.6513,
+      "step": 660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.781238811314e-05,
+      "loss": 0.6366,
+      "step": 661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7808807733619763e-05,
+      "loss": 0.5907,
+      "step": 662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7805227354099537e-05,
+      "loss": 0.6417,
+      "step": 663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.78016469745793e-05,
+      "loss": 0.6137,
+      "step": 664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7798066595059076e-05,
+      "loss": 0.5948,
+      "step": 665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.779448621553885e-05,
+      "loss": 0.6987,
+      "step": 666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7790905836018616e-05,
+      "loss": 0.6289,
+      "step": 667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.778732545649839e-05,
+      "loss": 0.6208,
+      "step": 668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.778374507697816e-05,
+      "loss": 0.6273,
+      "step": 669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7780164697457935e-05,
+      "loss": 0.5621,
+      "step": 670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.777658431793771e-05,
+      "loss": 0.6657,
+      "step": 671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7773003938417475e-05,
+      "loss": 0.579,
+      "step": 672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.776942355889725e-05,
+      "loss": 0.582,
+      "step": 673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7765843179377015e-05,
+      "loss": 0.6313,
+      "step": 674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.776226279985679e-05,
+      "loss": 0.674,
+      "step": 675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.775868242033656e-05,
+      "loss": 0.7313,
+      "step": 676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.775510204081633e-05,
+      "loss": 0.6226,
+      "step": 677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.77515216612961e-05,
+      "loss": 0.4999,
+      "step": 678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7747941281775874e-05,
+      "loss": 0.6622,
+      "step": 679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.774436090225564e-05,
+      "loss": 0.6635,
+      "step": 680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.774078052273541e-05,
+      "loss": 0.714,
+      "step": 681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.773720014321518e-05,
+      "loss": 0.68,
+      "step": 682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.773361976369495e-05,
+      "loss": 0.602,
+      "step": 683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7730039384174726e-05,
+      "loss": 0.5681,
+      "step": 684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.772645900465449e-05,
+      "loss": 0.684,
+      "step": 685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.7722878625134266e-05,
+      "loss": 0.589,
+      "step": 686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.771929824561404e-05,
+      "loss": 0.6831,
+      "step": 687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7715717866093805e-05,
+      "loss": 0.6739,
+      "step": 688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.771213748657358e-05,
+      "loss": 0.6169,
+      "step": 689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7708557107053345e-05,
+      "loss": 0.676,
+      "step": 690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.770497672753312e-05,
+      "loss": 0.6173,
+      "step": 691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.770139634801289e-05,
+      "loss": 0.6763,
+      "step": 692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7697815968492664e-05,
+      "loss": 0.6784,
+      "step": 693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.769423558897244e-05,
+      "loss": 0.6773,
+      "step": 694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7690655209452204e-05,
+      "loss": 0.5804,
+      "step": 695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.768707482993198e-05,
+      "loss": 0.6445,
+      "step": 696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.768349445041175e-05,
+      "loss": 0.6043,
+      "step": 697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.767991407089152e-05,
+      "loss": 0.7136,
+      "step": 698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.767633369137129e-05,
+      "loss": 0.6129,
+      "step": 699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7672753311851056e-05,
+      "loss": 0.6159,
+      "step": 700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.766917293233083e-05,
+      "loss": 0.6054,
+      "step": 701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.76655925528106e-05,
+      "loss": 0.6427,
+      "step": 702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.766201217329037e-05,
+      "loss": 0.641,
+      "step": 703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.765843179377014e-05,
+      "loss": 0.6576,
+      "step": 704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7654851414249916e-05,
+      "loss": 0.6144,
+      "step": 705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.765127103472968e-05,
+      "loss": 0.5989,
+      "step": 706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7647690655209455e-05,
+      "loss": 0.632,
+      "step": 707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.764411027568922e-05,
+      "loss": 0.6109,
+      "step": 708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7640529896168995e-05,
+      "loss": 0.5963,
+      "step": 709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.763694951664877e-05,
+      "loss": 0.5481,
+      "step": 710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.7633369137128534e-05,
+      "loss": 0.6597,
+      "step": 711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762978875760831e-05,
+      "loss": 0.6289,
+      "step": 712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762620837808808e-05,
+      "loss": 0.585,
+      "step": 713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.762262799856785e-05,
+      "loss": 0.6017,
+      "step": 714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.6541,
+      "step": 715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7615467239527394e-05,
+      "loss": 0.5871,
+      "step": 716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.761188686000717e-05,
+      "loss": 0.6162,
+      "step": 717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.760830648048693e-05,
+      "loss": 0.6705,
+      "step": 718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7604726100966706e-05,
+      "loss": 0.688,
+      "step": 719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.760114572144648e-05,
+      "loss": 0.6886,
+      "step": 720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7597565341926246e-05,
+      "loss": 0.6604,
+      "step": 721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.759398496240602e-05,
+      "loss": 0.6039,
+      "step": 722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.759040458288579e-05,
+      "loss": 0.6018,
+      "step": 723
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.758682420336556e-05,
+      "loss": 0.6566,
+      "step": 724
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.758324382384533e-05,
+      "loss": 0.6245,
+      "step": 725
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75796634443251e-05,
+      "loss": 0.5896,
+      "step": 726
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.757608306480487e-05,
+      "loss": 0.615,
+      "step": 727
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7572502685284645e-05,
+      "loss": 0.5735,
+      "step": 728
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.756892230576441e-05,
+      "loss": 0.6692,
+      "step": 729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7565341926244184e-05,
+      "loss": 0.5977,
+      "step": 730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.756176154672396e-05,
+      "loss": 0.6393,
+      "step": 731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7558181167203724e-05,
+      "loss": 0.624,
+      "step": 732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.75546007876835e-05,
+      "loss": 0.6192,
+      "step": 733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7551020408163263e-05,
+      "loss": 0.6107,
+      "step": 734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.754744002864304e-05,
+      "loss": 0.6727,
+      "step": 735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.754385964912281e-05,
+      "loss": 0.6203,
+      "step": 736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7540279269602576e-05,
+      "loss": 0.589,
+      "step": 737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.753669889008235e-05,
+      "loss": 0.5663,
+      "step": 738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.753311851056212e-05,
+      "loss": 0.6419,
+      "step": 739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7529538131041896e-05,
+      "loss": 0.7377,
+      "step": 740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.752595775152167e-05,
+      "loss": 0.6478,
+      "step": 741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.7522377372001435e-05,
+      "loss": 0.6434,
+      "step": 742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.751879699248121e-05,
+      "loss": 0.5502,
+      "step": 743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7515216612960975e-05,
+      "loss": 0.5807,
+      "step": 744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.751163623344075e-05,
+      "loss": 0.6415,
+      "step": 745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.750805585392052e-05,
+      "loss": 0.6201,
+      "step": 746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.750447547440029e-05,
+      "loss": 0.6885,
+      "step": 747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.750089509488006e-05,
+      "loss": 0.6618,
+      "step": 748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.749731471535983e-05,
+      "loss": 0.629,
+      "step": 749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.74937343358396e-05,
+      "loss": 0.6916,
+      "step": 750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7490153956319374e-05,
+      "loss": 0.6061,
+      "step": 751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.748657357679914e-05,
+      "loss": 0.6416,
+      "step": 752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7482993197278913e-05,
+      "loss": 0.639,
+      "step": 753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7479412817758687e-05,
+      "loss": 0.6176,
+      "step": 754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.747583243823845e-05,
+      "loss": 0.686,
+      "step": 755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7472252058718226e-05,
+      "loss": 0.5925,
+      "step": 756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.746867167919799e-05,
+      "loss": 0.6369,
+      "step": 757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7465091299677766e-05,
+      "loss": 0.6552,
+      "step": 758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.746151092015754e-05,
+      "loss": 0.5725,
+      "step": 759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7457930540637305e-05,
+      "loss": 0.5793,
+      "step": 760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.745435016111708e-05,
+      "loss": 0.651,
+      "step": 761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.745076978159685e-05,
+      "loss": 0.5542,
+      "step": 762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7447189402076625e-05,
+      "loss": 0.5755,
+      "step": 763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.74436090225564e-05,
+      "loss": 0.5698,
+      "step": 764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7440028643036165e-05,
+      "loss": 0.5871,
+      "step": 765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.743644826351594e-05,
+      "loss": 0.5948,
+      "step": 766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.7432867883995704e-05,
+      "loss": 0.5909,
+      "step": 767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742928750447548e-05,
+      "loss": 0.571,
+      "step": 768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742570712495525e-05,
+      "loss": 0.6713,
+      "step": 769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.742212674543502e-05,
+      "loss": 0.6979,
+      "step": 770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.741854636591479e-05,
+      "loss": 0.6522,
+      "step": 771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.741496598639456e-05,
+      "loss": 0.5555,
+      "step": 772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.741138560687433e-05,
+      "loss": 0.6757,
+      "step": 773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.74078052273541e-05,
+      "loss": 0.6042,
+      "step": 774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.740422484783387e-05,
+      "loss": 0.6209,
+      "step": 775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.740064446831364e-05,
+      "loss": 0.6783,
+      "step": 776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7397064088793416e-05,
+      "loss": 0.6464,
+      "step": 777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.739348370927318e-05,
+      "loss": 0.5771,
+      "step": 778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7389903329752955e-05,
+      "loss": 0.6112,
+      "step": 779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.738632295023273e-05,
+      "loss": 0.5757,
+      "step": 780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7382742570712495e-05,
+      "loss": 0.5478,
+      "step": 781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.737916219119227e-05,
+      "loss": 0.6216,
+      "step": 782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7375581811672034e-05,
+      "loss": 0.6566,
+      "step": 783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.737200143215181e-05,
+      "loss": 0.5559,
+      "step": 784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.5706,
+      "step": 785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7364840673111354e-05,
+      "loss": 0.7015,
+      "step": 786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.736126029359113e-05,
+      "loss": 0.5659,
+      "step": 787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7357679914070894e-05,
+      "loss": 0.6617,
+      "step": 788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.735409953455067e-05,
+      "loss": 0.5842,
+      "step": 789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.735051915503044e-05,
+      "loss": 0.5757,
+      "step": 790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7346938775510206e-05,
+      "loss": 0.5771,
+      "step": 791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.734335839598998e-05,
+      "loss": 0.6813,
+      "step": 792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7339778016469746e-05,
+      "loss": 0.5735,
+      "step": 793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.733619763694952e-05,
+      "loss": 0.5787,
+      "step": 794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.733261725742929e-05,
+      "loss": 0.5939,
+      "step": 795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732903687790906e-05,
+      "loss": 0.5919,
+      "step": 796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.732545649838883e-05,
+      "loss": 0.759,
+      "step": 797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.7321876118868605e-05,
+      "loss": 0.5852,
+      "step": 798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.731829573934837e-05,
+      "loss": 0.691,
+      "step": 799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7314715359828145e-05,
+      "loss": 0.7036,
+      "step": 800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.731113498030791e-05,
+      "loss": 0.6744,
+      "step": 801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7307554600787684e-05,
+      "loss": 0.5779,
+      "step": 802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.730397422126746e-05,
+      "loss": 0.616,
+      "step": 803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7300393841747224e-05,
+      "loss": 0.6257,
+      "step": 804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7296813462227e-05,
+      "loss": 0.5997,
+      "step": 805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.729323308270677e-05,
+      "loss": 0.6384,
+      "step": 806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.728965270318654e-05,
+      "loss": 0.5652,
+      "step": 807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.728607232366631e-05,
+      "loss": 0.6217,
+      "step": 808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.728249194414608e-05,
+      "loss": 0.6238,
+      "step": 809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7278911564625856e-05,
+      "loss": 0.6617,
+      "step": 810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.727533118510562e-05,
+      "loss": 0.5764,
+      "step": 811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7271750805585396e-05,
+      "loss": 0.5569,
+      "step": 812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.726817042606517e-05,
+      "loss": 0.6364,
+      "step": 813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7264590046544936e-05,
+      "loss": 0.6016,
+      "step": 814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.726100966702471e-05,
+      "loss": 0.684,
+      "step": 815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.725742928750448e-05,
+      "loss": 0.5959,
+      "step": 816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.725384890798425e-05,
+      "loss": 0.6257,
+      "step": 817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.725026852846402e-05,
+      "loss": 0.61,
+      "step": 818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.724668814894379e-05,
+      "loss": 0.6271,
+      "step": 819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.724310776942356e-05,
+      "loss": 0.5828,
+      "step": 820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7239527389903334e-05,
+      "loss": 0.6316,
+      "step": 821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.72359470103831e-05,
+      "loss": 0.6191,
+      "step": 822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7232366630862874e-05,
+      "loss": 0.584,
+      "step": 823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722878625134265e-05,
+      "loss": 0.6755,
+      "step": 824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.7225205871822413e-05,
+      "loss": 0.6218,
+      "step": 825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.722162549230219e-05,
+      "loss": 0.5374,
+      "step": 826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.721804511278195e-05,
+      "loss": 0.621,
+      "step": 827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7214464733261726e-05,
+      "loss": 0.6616,
+      "step": 828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.72108843537415e-05,
+      "loss": 0.5874,
+      "step": 829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7207303974221266e-05,
+      "loss": 0.6147,
+      "step": 830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.720372359470104e-05,
+      "loss": 0.5938,
+      "step": 831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7200143215180805e-05,
+      "loss": 0.6514,
+      "step": 832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7196562835660585e-05,
+      "loss": 0.6204,
+      "step": 833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.719298245614036e-05,
+      "loss": 0.5867,
+      "step": 834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7189402076620125e-05,
+      "loss": 0.642,
+      "step": 835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.71858216970999e-05,
+      "loss": 0.5799,
+      "step": 836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7182241317579665e-05,
+      "loss": 0.67,
+      "step": 837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.717866093805944e-05,
+      "loss": 0.5356,
+      "step": 838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.717508055853921e-05,
+      "loss": 0.566,
+      "step": 839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.717150017901898e-05,
+      "loss": 0.6405,
+      "step": 840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.716791979949875e-05,
+      "loss": 0.6777,
+      "step": 841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7164339419978524e-05,
+      "loss": 0.6366,
+      "step": 842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.716075904045829e-05,
+      "loss": 0.6348,
+      "step": 843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7157178660938063e-05,
+      "loss": 0.6344,
+      "step": 844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.715359828141783e-05,
+      "loss": 0.7312,
+      "step": 845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.71500179018976e-05,
+      "loss": 0.6219,
+      "step": 846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7146437522377376e-05,
+      "loss": 0.552,
+      "step": 847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.714285714285714e-05,
+      "loss": 0.6205,
+      "step": 848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7139276763336916e-05,
+      "loss": 0.6674,
+      "step": 849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.713569638381668e-05,
+      "loss": 0.5688,
+      "step": 850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7132116004296455e-05,
+      "loss": 0.6034,
+      "step": 851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.712853562477623e-05,
+      "loss": 0.5472,
+      "step": 852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.7124955245255995e-05,
+      "loss": 0.6654,
+      "step": 853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.712137486573577e-05,
+      "loss": 0.6103,
+      "step": 854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.711779448621554e-05,
+      "loss": 0.6437,
+      "step": 855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7114214106695315e-05,
+      "loss": 0.5895,
+      "step": 856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.711063372717509e-05,
+      "loss": 0.6347,
+      "step": 857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7107053347654854e-05,
+      "loss": 0.6135,
+      "step": 858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.710347296813463e-05,
+      "loss": 0.5635,
+      "step": 859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7099892588614394e-05,
+      "loss": 0.5908,
+      "step": 860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.709631220909417e-05,
+      "loss": 0.5871,
+      "step": 861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.709273182957394e-05,
+      "loss": 0.6856,
+      "step": 862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7089151450053707e-05,
+      "loss": 0.5981,
+      "step": 863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.708557107053348e-05,
+      "loss": 0.6354,
+      "step": 864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.708199069101325e-05,
+      "loss": 0.5895,
+      "step": 865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.707841031149302e-05,
+      "loss": 0.6859,
+      "step": 866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.707482993197279e-05,
+      "loss": 0.5745,
+      "step": 867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.707124955245256e-05,
+      "loss": 0.6336,
+      "step": 868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.706766917293233e-05,
+      "loss": 0.6705,
+      "step": 869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7064088793412105e-05,
+      "loss": 0.6259,
+      "step": 870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.706050841389187e-05,
+      "loss": 0.6278,
+      "step": 871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7056928034371645e-05,
+      "loss": 0.602,
+      "step": 872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.705334765485142e-05,
+      "loss": 0.6464,
+      "step": 873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7049767275331184e-05,
+      "loss": 0.7884,
+      "step": 874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.704618689581096e-05,
+      "loss": 0.5779,
+      "step": 875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7042606516290724e-05,
+      "loss": 0.6103,
+      "step": 876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.70390261367705e-05,
+      "loss": 0.6722,
+      "step": 877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.703544575725027e-05,
+      "loss": 0.6517,
+      "step": 878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.703186537773004e-05,
+      "loss": 0.5835,
+      "step": 879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702828499820982e-05,
+      "loss": 0.5853,
+      "step": 880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.702470461868958e-05,
+      "loss": 0.5948,
+      "step": 881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.7021124239169356e-05,
+      "loss": 0.5676,
+      "step": 882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.701754385964913e-05,
+      "loss": 0.6713,
+      "step": 883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7013963480128896e-05,
+      "loss": 0.5899,
+      "step": 884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.701038310060867e-05,
+      "loss": 0.5555,
+      "step": 885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.7006802721088436e-05,
+      "loss": 0.5958,
+      "step": 886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.700322234156821e-05,
+      "loss": 0.6801,
+      "step": 887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.699964196204798e-05,
+      "loss": 0.6851,
+      "step": 888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.699606158252775e-05,
+      "loss": 0.6117,
+      "step": 889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.699248120300752e-05,
+      "loss": 0.621,
+      "step": 890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6988900823487295e-05,
+      "loss": 0.5166,
+      "step": 891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.698532044396706e-05,
+      "loss": 0.6238,
+      "step": 892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6981740064446834e-05,
+      "loss": 0.5591,
+      "step": 893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.69781596849266e-05,
+      "loss": 0.6256,
+      "step": 894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6974579305406374e-05,
+      "loss": 0.6114,
+      "step": 895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.697099892588615e-05,
+      "loss": 0.6451,
+      "step": 896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6967418546365914e-05,
+      "loss": 0.6534,
+      "step": 897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.696383816684569e-05,
+      "loss": 0.7154,
+      "step": 898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.696025778732546e-05,
+      "loss": 0.6499,
+      "step": 899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6956677407805226e-05,
+      "loss": 0.6366,
+      "step": 900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6953097028285e-05,
+      "loss": 0.6129,
+      "step": 901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6949516648764766e-05,
+      "loss": 0.5906,
+      "step": 902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6945936269244546e-05,
+      "loss": 0.6005,
+      "step": 903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.694235588972431e-05,
+      "loss": 0.7187,
+      "step": 904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6938775510204086e-05,
+      "loss": 0.638,
+      "step": 905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.693519513068386e-05,
+      "loss": 0.5951,
+      "step": 906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.6931614751163625e-05,
+      "loss": 0.6234,
+      "step": 907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.69280343716434e-05,
+      "loss": 0.6947,
+      "step": 908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692445399212317e-05,
+      "loss": 0.5582,
+      "step": 909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.692087361260294e-05,
+      "loss": 0.5085,
+      "step": 910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.691729323308271e-05,
+      "loss": 0.6483,
+      "step": 911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.691371285356248e-05,
+      "loss": 0.5519,
+      "step": 912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.691013247404225e-05,
+      "loss": 0.7038,
+      "step": 913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6906552094522024e-05,
+      "loss": 0.5783,
+      "step": 914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.690297171500179e-05,
+      "loss": 0.5942,
+      "step": 915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6899391335481563e-05,
+      "loss": 0.6228,
+      "step": 916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.689581095596134e-05,
+      "loss": 0.6308,
+      "step": 917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.68922305764411e-05,
+      "loss": 0.6745,
+      "step": 918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6888650196920876e-05,
+      "loss": 0.6317,
+      "step": 919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.688506981740064e-05,
+      "loss": 0.6484,
+      "step": 920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6881489437880416e-05,
+      "loss": 0.676,
+      "step": 921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.687790905836019e-05,
+      "loss": 0.6246,
+      "step": 922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6874328678839955e-05,
+      "loss": 0.7367,
+      "step": 923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.687074829931973e-05,
+      "loss": 0.6958,
+      "step": 924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6867167919799495e-05,
+      "loss": 0.6646,
+      "step": 925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6863587540279275e-05,
+      "loss": 0.6246,
+      "step": 926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.686000716075905e-05,
+      "loss": 0.6119,
+      "step": 927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6856426781238815e-05,
+      "loss": 0.6369,
+      "step": 928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.685284640171859e-05,
+      "loss": 0.6242,
+      "step": 929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6849266022198354e-05,
+      "loss": 0.5706,
+      "step": 930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.684568564267813e-05,
+      "loss": 0.6378,
+      "step": 931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.68421052631579e-05,
+      "loss": 0.5909,
+      "step": 932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.683852488363767e-05,
+      "loss": 0.6158,
+      "step": 933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.683494450411744e-05,
+      "loss": 0.649,
+      "step": 934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.6831364124597213e-05,
+      "loss": 0.6368,
+      "step": 935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682778374507698e-05,
+      "loss": 0.589,
+      "step": 936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682420336555675e-05,
+      "loss": 0.6531,
+      "step": 937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.682062298603652e-05,
+      "loss": 0.6511,
+      "step": 938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.681704260651629e-05,
+      "loss": 0.5871,
+      "step": 939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6813462226996066e-05,
+      "loss": 0.5874,
+      "step": 940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.680988184747583e-05,
+      "loss": 0.5381,
+      "step": 941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6806301467955605e-05,
+      "loss": 0.6121,
+      "step": 942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.680272108843537e-05,
+      "loss": 0.662,
+      "step": 943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6799140708915145e-05,
+      "loss": 0.5993,
+      "step": 944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.679556032939492e-05,
+      "loss": 0.6021,
+      "step": 945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6791979949874685e-05,
+      "loss": 0.5491,
+      "step": 946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.678839957035446e-05,
+      "loss": 0.519,
+      "step": 947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.678481919083423e-05,
+      "loss": 0.6508,
+      "step": 948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6781238811314e-05,
+      "loss": 0.6183,
+      "step": 949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.677765843179378e-05,
+      "loss": 0.5984,
+      "step": 950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6774078052273544e-05,
+      "loss": 0.6919,
+      "step": 951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.677049767275332e-05,
+      "loss": 0.5597,
+      "step": 952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.676691729323309e-05,
+      "loss": 0.558,
+      "step": 953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6763336913712857e-05,
+      "loss": 0.6319,
+      "step": 954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.675975653419263e-05,
+      "loss": 0.6414,
+      "step": 955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6756176154672396e-05,
+      "loss": 0.7025,
+      "step": 956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.675259577515217e-05,
+      "loss": 0.599,
+      "step": 957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.674901539563194e-05,
+      "loss": 0.603,
+      "step": 958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.674543501611171e-05,
+      "loss": 0.59,
+      "step": 959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.674185463659148e-05,
+      "loss": 0.5935,
+      "step": 960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.673827425707125e-05,
+      "loss": 0.6459,
+      "step": 961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.673469387755102e-05,
+      "loss": 0.61,
+      "step": 962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6731113498030795e-05,
+      "loss": 0.6742,
+      "step": 963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.672753311851056e-05,
+      "loss": 0.4971,
+      "step": 964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6723952738990334e-05,
+      "loss": 0.677,
+      "step": 965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.672037235947011e-05,
+      "loss": 0.6509,
+      "step": 966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6716791979949874e-05,
+      "loss": 0.6588,
+      "step": 967
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.671321160042965e-05,
+      "loss": 0.6436,
+      "step": 968
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6709631220909414e-05,
+      "loss": 0.6011,
+      "step": 969
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.670605084138919e-05,
+      "loss": 0.5969,
+      "step": 970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.670247046186896e-05,
+      "loss": 0.5635,
+      "step": 971
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6698890082348726e-05,
+      "loss": 0.5857,
+      "step": 972
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6695309702828506e-05,
+      "loss": 0.5805,
+      "step": 973
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.669172932330827e-05,
+      "loss": 0.5955,
+      "step": 974
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6688148943788046e-05,
+      "loss": 0.6117,
+      "step": 975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.668456856426782e-05,
+      "loss": 0.5805,
+      "step": 976
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6680988184747586e-05,
+      "loss": 0.6374,
+      "step": 977
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.667740780522736e-05,
+      "loss": 0.5811,
+      "step": 978
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6673827425707125e-05,
+      "loss": 0.6176,
+      "step": 979
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.66702470461869e-05,
+      "loss": 0.592,
+      "step": 980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.5797,
+      "step": 981
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.666308628714644e-05,
+      "loss": 0.5918,
+      "step": 982
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.665950590762621e-05,
+      "loss": 0.5987,
+      "step": 983
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6655925528105984e-05,
+      "loss": 0.5861,
+      "step": 984
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.665234514858575e-05,
+      "loss": 0.6632,
+      "step": 985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6648764769065524e-05,
+      "loss": 0.6804,
+      "step": 986
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.664518438954529e-05,
+      "loss": 0.6425,
+      "step": 987
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6641604010025064e-05,
+      "loss": 0.6847,
+      "step": 988
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.663802363050484e-05,
+      "loss": 0.6121,
+      "step": 989
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.66344432509846e-05,
+      "loss": 0.6,
+      "step": 990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6630862871464376e-05,
+      "loss": 0.6774,
+      "step": 991
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662728249194415e-05,
+      "loss": 0.6269,
+      "step": 992
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6623702112423916e-05,
+      "loss": 0.5897,
+      "step": 993
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.662012173290369e-05,
+      "loss": 0.6187,
+      "step": 994
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6616541353383456e-05,
+      "loss": 0.5873,
+      "step": 995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6612960973863236e-05,
+      "loss": 0.638,
+      "step": 996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6609380594343e-05,
+      "loss": 0.5715,
+      "step": 997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6605800214822775e-05,
+      "loss": 0.6211,
+      "step": 998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.660221983530255e-05,
+      "loss": 0.577,
+      "step": 999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6598639455782315e-05,
+      "loss": 0.5802,
+      "step": 1000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.659505907626209e-05,
+      "loss": 0.596,
+      "step": 1001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.659147869674186e-05,
+      "loss": 0.5693,
+      "step": 1002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.658789831722163e-05,
+      "loss": 0.5805,
+      "step": 1003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.65843179377014e-05,
+      "loss": 0.5854,
+      "step": 1004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.658073755818117e-05,
+      "loss": 0.6109,
+      "step": 1005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.657715717866094e-05,
+      "loss": 0.5919,
+      "step": 1006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6573576799140714e-05,
+      "loss": 0.6611,
+      "step": 1007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.656999641962048e-05,
+      "loss": 0.6238,
+      "step": 1008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.656641604010025e-05,
+      "loss": 0.691,
+      "step": 1009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6562835660580026e-05,
+      "loss": 0.6763,
+      "step": 1010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.655925528105979e-05,
+      "loss": 0.6107,
+      "step": 1011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6555674901539566e-05,
+      "loss": 0.6566,
+      "step": 1012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.655209452201933e-05,
+      "loss": 0.5724,
+      "step": 1013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6548514142499105e-05,
+      "loss": 0.6358,
+      "step": 1014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.654493376297888e-05,
+      "loss": 0.6135,
+      "step": 1015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6541353383458645e-05,
+      "loss": 0.6206,
+      "step": 1016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.653777300393842e-05,
+      "loss": 0.6334,
+      "step": 1017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6534192624418185e-05,
+      "loss": 0.6253,
+      "step": 1018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.653061224489796e-05,
+      "loss": 0.6363,
+      "step": 1019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.652703186537774e-05,
+      "loss": 0.655,
+      "step": 1020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6523451485857504e-05,
+      "loss": 0.5597,
+      "step": 1021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.651987110633728e-05,
+      "loss": 0.613,
+      "step": 1022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.6516290726817044e-05,
+      "loss": 0.6009,
+      "step": 1023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.651271034729682e-05,
+      "loss": 0.6324,
+      "step": 1024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.650912996777659e-05,
+      "loss": 0.5835,
+      "step": 1025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.650554958825636e-05,
+      "loss": 0.5916,
+      "step": 1026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.650196920873613e-05,
+      "loss": 0.6254,
+      "step": 1027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.64983888292159e-05,
+      "loss": 0.6446,
+      "step": 1028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.649480844969567e-05,
+      "loss": 0.639,
+      "step": 1029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.649122807017544e-05,
+      "loss": 0.6257,
+      "step": 1030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.648764769065521e-05,
+      "loss": 0.7095,
+      "step": 1031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.648406731113498e-05,
+      "loss": 0.6922,
+      "step": 1032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6480486931614755e-05,
+      "loss": 0.6045,
+      "step": 1033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.647690655209452e-05,
+      "loss": 0.6134,
+      "step": 1034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6473326172574295e-05,
+      "loss": 0.6234,
+      "step": 1035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.646974579305406e-05,
+      "loss": 0.6726,
+      "step": 1036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6466165413533835e-05,
+      "loss": 0.6133,
+      "step": 1037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.646258503401361e-05,
+      "loss": 0.6138,
+      "step": 1038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6459004654493374e-05,
+      "loss": 0.6091,
+      "step": 1039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.645542427497315e-05,
+      "loss": 0.536,
+      "step": 1040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.645184389545292e-05,
+      "loss": 0.5595,
+      "step": 1041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.644826351593269e-05,
+      "loss": 0.633,
+      "step": 1042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.644468313641247e-05,
+      "loss": 0.5593,
+      "step": 1043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.644110275689223e-05,
+      "loss": 0.5435,
+      "step": 1044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6437522377372007e-05,
+      "loss": 0.6041,
+      "step": 1045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.643394199785178e-05,
+      "loss": 0.6958,
+      "step": 1046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6430361618331546e-05,
+      "loss": 0.5786,
+      "step": 1047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.642678123881132e-05,
+      "loss": 0.6889,
+      "step": 1048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.6423200859291086e-05,
+      "loss": 0.5802,
+      "step": 1049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.641962047977086e-05,
+      "loss": 0.5233,
+      "step": 1050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.641604010025063e-05,
+      "loss": 0.6665,
+      "step": 1051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.64124597207304e-05,
+      "loss": 0.6309,
+      "step": 1052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.640887934121017e-05,
+      "loss": 0.697,
+      "step": 1053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.640529896168994e-05,
+      "loss": 0.5385,
+      "step": 1054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.640171858216971e-05,
+      "loss": 0.6051,
+      "step": 1055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6398138202649484e-05,
+      "loss": 0.5907,
+      "step": 1056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.639455782312925e-05,
+      "loss": 0.5226,
+      "step": 1057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6390977443609024e-05,
+      "loss": 0.6127,
+      "step": 1058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.63873970640888e-05,
+      "loss": 0.7291,
+      "step": 1059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6383816684568564e-05,
+      "loss": 0.4886,
+      "step": 1060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.638023630504834e-05,
+      "loss": 0.6304,
+      "step": 1061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.63766559255281e-05,
+      "loss": 0.5866,
+      "step": 1062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6373075546007876e-05,
+      "loss": 0.6125,
+      "step": 1063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.636949516648765e-05,
+      "loss": 0.6053,
+      "step": 1064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6365914786967416e-05,
+      "loss": 0.6521,
+      "step": 1065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6362334407447196e-05,
+      "loss": 0.6062,
+      "step": 1066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.635875402792696e-05,
+      "loss": 0.5406,
+      "step": 1067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6355173648406736e-05,
+      "loss": 0.5386,
+      "step": 1068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.635159326888651e-05,
+      "loss": 0.6663,
+      "step": 1069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6348012889366275e-05,
+      "loss": 0.5704,
+      "step": 1070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.634443250984605e-05,
+      "loss": 0.612,
+      "step": 1071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6340852130325815e-05,
+      "loss": 0.6014,
+      "step": 1072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.633727175080559e-05,
+      "loss": 0.5785,
+      "step": 1073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.633369137128536e-05,
+      "loss": 0.6531,
+      "step": 1074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.633011099176513e-05,
+      "loss": 0.5799,
+      "step": 1075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.63265306122449e-05,
+      "loss": 0.6021,
+      "step": 1076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6322950232724674e-05,
+      "loss": 0.6322,
+      "step": 1077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.631936985320444e-05,
+      "loss": 0.5358,
+      "step": 1078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.6315789473684214e-05,
+      "loss": 0.5858,
+      "step": 1079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.631220909416398e-05,
+      "loss": 0.6852,
+      "step": 1080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.630862871464375e-05,
+      "loss": 0.5826,
+      "step": 1081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6305048335123526e-05,
+      "loss": 0.6315,
+      "step": 1082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.630146795560329e-05,
+      "loss": 0.5886,
+      "step": 1083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6297887576083066e-05,
+      "loss": 0.6191,
+      "step": 1084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.629430719656284e-05,
+      "loss": 0.5792,
+      "step": 1085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6290726817042606e-05,
+      "loss": 0.6413,
+      "step": 1086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.628714643752238e-05,
+      "loss": 0.5461,
+      "step": 1087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6283566058002145e-05,
+      "loss": 0.6258,
+      "step": 1088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.627998567848192e-05,
+      "loss": 0.6321,
+      "step": 1089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.627640529896169e-05,
+      "loss": 0.5986,
+      "step": 1090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6272824919441465e-05,
+      "loss": 0.6726,
+      "step": 1091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.626924453992124e-05,
+      "loss": 0.6255,
+      "step": 1092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6265664160401004e-05,
+      "loss": 0.6877,
+      "step": 1093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.626208378088078e-05,
+      "loss": 0.6733,
+      "step": 1094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.625850340136055e-05,
+      "loss": 0.5915,
+      "step": 1095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.625492302184032e-05,
+      "loss": 0.5628,
+      "step": 1096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.625134264232009e-05,
+      "loss": 0.6341,
+      "step": 1097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.624776226279986e-05,
+      "loss": 0.6154,
+      "step": 1098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.624418188327963e-05,
+      "loss": 0.6416,
+      "step": 1099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.62406015037594e-05,
+      "loss": 0.5294,
+      "step": 1100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.623702112423917e-05,
+      "loss": 0.6237,
+      "step": 1101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.623344074471894e-05,
+      "loss": 0.6366,
+      "step": 1102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6229860365198716e-05,
+      "loss": 0.5905,
+      "step": 1103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.622627998567848e-05,
+      "loss": 0.6122,
+      "step": 1104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6222699606158255e-05,
+      "loss": 0.561,
+      "step": 1105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.621911922663802e-05,
+      "loss": 0.5946,
+      "step": 1106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.6215538847117795e-05,
+      "loss": 0.6495,
+      "step": 1107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.621195846759757e-05,
+      "loss": 0.5825,
+      "step": 1108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6208378088077335e-05,
+      "loss": 0.6402,
+      "step": 1109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.620479770855711e-05,
+      "loss": 0.5731,
+      "step": 1110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.620121732903688e-05,
+      "loss": 0.6238,
+      "step": 1111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.619763694951665e-05,
+      "loss": 0.5702,
+      "step": 1112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.619405656999643e-05,
+      "loss": 0.6366,
+      "step": 1113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6190476190476194e-05,
+      "loss": 0.6204,
+      "step": 1114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.618689581095597e-05,
+      "loss": 0.5949,
+      "step": 1115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6183315431435733e-05,
+      "loss": 0.6061,
+      "step": 1116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.617973505191551e-05,
+      "loss": 0.5727,
+      "step": 1117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.617615467239528e-05,
+      "loss": 0.5692,
+      "step": 1118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6172574292875046e-05,
+      "loss": 0.6131,
+      "step": 1119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.616899391335482e-05,
+      "loss": 0.5821,
+      "step": 1120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.616541353383459e-05,
+      "loss": 0.6537,
+      "step": 1121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.616183315431436e-05,
+      "loss": 0.6501,
+      "step": 1122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.615825277479413e-05,
+      "loss": 0.533,
+      "step": 1123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.61546723952739e-05,
+      "loss": 0.5931,
+      "step": 1124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.615109201575367e-05,
+      "loss": 0.5532,
+      "step": 1125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6147511636233445e-05,
+      "loss": 0.5964,
+      "step": 1126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.614393125671321e-05,
+      "loss": 0.6437,
+      "step": 1127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6140350877192985e-05,
+      "loss": 0.6301,
+      "step": 1128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.613677049767275e-05,
+      "loss": 0.627,
+      "step": 1129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6133190118152524e-05,
+      "loss": 0.6636,
+      "step": 1130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.61296097386323e-05,
+      "loss": 0.6019,
+      "step": 1131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6126029359112064e-05,
+      "loss": 0.6136,
+      "step": 1132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.612244897959184e-05,
+      "loss": 0.6426,
+      "step": 1133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.611886860007161e-05,
+      "loss": 0.575,
+      "step": 1134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.6115288220551377e-05,
+      "loss": 0.6652,
+      "step": 1135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6111707841031157e-05,
+      "loss": 0.644,
+      "step": 1136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.610812746151092e-05,
+      "loss": 0.578,
+      "step": 1137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6104547081990696e-05,
+      "loss": 0.545,
+      "step": 1138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.610096670247047e-05,
+      "loss": 0.582,
+      "step": 1139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6097386322950236e-05,
+      "loss": 0.6065,
+      "step": 1140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.609380594343001e-05,
+      "loss": 0.5412,
+      "step": 1141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6090225563909775e-05,
+      "loss": 0.6027,
+      "step": 1142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.608664518438955e-05,
+      "loss": 0.6968,
+      "step": 1143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.608306480486932e-05,
+      "loss": 0.6591,
+      "step": 1144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.607948442534909e-05,
+      "loss": 0.5979,
+      "step": 1145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.607590404582886e-05,
+      "loss": 0.5719,
+      "step": 1146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.607232366630863e-05,
+      "loss": 0.5839,
+      "step": 1147
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.60687432867884e-05,
+      "loss": 0.5571,
+      "step": 1148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6065162907268174e-05,
+      "loss": 0.6108,
+      "step": 1149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.606158252774794e-05,
+      "loss": 0.6502,
+      "step": 1150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6058002148227714e-05,
+      "loss": 0.5885,
+      "step": 1151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.605442176870749e-05,
+      "loss": 0.5603,
+      "step": 1152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.605084138918725e-05,
+      "loss": 0.567,
+      "step": 1153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6047261009667026e-05,
+      "loss": 0.6602,
+      "step": 1154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.604368063014679e-05,
+      "loss": 0.6226,
+      "step": 1155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6040100250626566e-05,
+      "loss": 0.588,
+      "step": 1156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.603651987110634e-05,
+      "loss": 0.631,
+      "step": 1157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6032939491586106e-05,
+      "loss": 0.6639,
+      "step": 1158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.602935911206588e-05,
+      "loss": 0.617,
+      "step": 1159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.602577873254565e-05,
+      "loss": 0.566,
+      "step": 1160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6022198353025425e-05,
+      "loss": 0.6075,
+      "step": 1161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.60186179735052e-05,
+      "loss": 0.5346,
+      "step": 1162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6015037593984965e-05,
+      "loss": 0.756,
+      "step": 1163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.601145721446474e-05,
+      "loss": 0.5488,
+      "step": 1164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6007876834944504e-05,
+      "loss": 0.6731,
+      "step": 1165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.600429645542428e-05,
+      "loss": 0.5547,
+      "step": 1166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.600071607590405e-05,
+      "loss": 0.5873,
+      "step": 1167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.599713569638382e-05,
+      "loss": 0.544,
+      "step": 1168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.599355531686359e-05,
+      "loss": 0.6154,
+      "step": 1169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5989974937343364e-05,
+      "loss": 0.5926,
+      "step": 1170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.598639455782313e-05,
+      "loss": 0.5547,
+      "step": 1171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.59828141783029e-05,
+      "loss": 0.5935,
+      "step": 1172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.597923379878267e-05,
+      "loss": 0.7053,
+      "step": 1173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.597565341926244e-05,
+      "loss": 0.5968,
+      "step": 1174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5972073039742216e-05,
+      "loss": 0.5802,
+      "step": 1175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.596849266022198e-05,
+      "loss": 0.6311,
+      "step": 1176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5964912280701756e-05,
+      "loss": 0.6786,
+      "step": 1177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.596133190118153e-05,
+      "loss": 0.623,
+      "step": 1178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5957751521661295e-05,
+      "loss": 0.5511,
+      "step": 1179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.595417114214107e-05,
+      "loss": 0.5798,
+      "step": 1180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5950590762620835e-05,
+      "loss": 0.6675,
+      "step": 1181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.594701038310061e-05,
+      "loss": 0.7011,
+      "step": 1182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.594343000358038e-05,
+      "loss": 0.6311,
+      "step": 1183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5939849624060154e-05,
+      "loss": 0.6911,
+      "step": 1184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.593626924453993e-05,
+      "loss": 0.6358,
+      "step": 1185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5932688865019694e-05,
+      "loss": 0.6025,
+      "step": 1186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592910848549947e-05,
+      "loss": 0.5972,
+      "step": 1187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592552810597924e-05,
+      "loss": 0.608,
+      "step": 1188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.592194772645901e-05,
+      "loss": 0.575,
+      "step": 1189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.591836734693878e-05,
+      "loss": 0.7249,
+      "step": 1190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.5914786967418546e-05,
+      "loss": 0.5996,
+      "step": 1191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.591120658789832e-05,
+      "loss": 0.5665,
+      "step": 1192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.590762620837809e-05,
+      "loss": 0.583,
+      "step": 1193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.590404582885786e-05,
+      "loss": 0.6985,
+      "step": 1194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.590046544933763e-05,
+      "loss": 0.6615,
+      "step": 1195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5896885069817405e-05,
+      "loss": 0.6313,
+      "step": 1196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.589330469029717e-05,
+      "loss": 0.6556,
+      "step": 1197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5889724310776945e-05,
+      "loss": 0.5848,
+      "step": 1198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.588614393125671e-05,
+      "loss": 0.5918,
+      "step": 1199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5882563551736485e-05,
+      "loss": 0.6167,
+      "step": 1200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.587898317221626e-05,
+      "loss": 0.6604,
+      "step": 1201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5875402792696024e-05,
+      "loss": 0.6393,
+      "step": 1202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.58718224131758e-05,
+      "loss": 0.673,
+      "step": 1203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.586824203365557e-05,
+      "loss": 0.5803,
+      "step": 1204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.586466165413534e-05,
+      "loss": 0.6283,
+      "step": 1205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.586108127461512e-05,
+      "loss": 0.5731,
+      "step": 1206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5857500895094883e-05,
+      "loss": 0.5175,
+      "step": 1207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.585392051557466e-05,
+      "loss": 0.5852,
+      "step": 1208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.585034013605442e-05,
+      "loss": 0.6266,
+      "step": 1209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5846759756534196e-05,
+      "loss": 0.6389,
+      "step": 1210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.584317937701397e-05,
+      "loss": 0.5959,
+      "step": 1211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5839598997493736e-05,
+      "loss": 0.6002,
+      "step": 1212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.583601861797351e-05,
+      "loss": 0.6004,
+      "step": 1213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.583243823845328e-05,
+      "loss": 0.5752,
+      "step": 1214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582885785893305e-05,
+      "loss": 0.6275,
+      "step": 1215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582527747941282e-05,
+      "loss": 0.5978,
+      "step": 1216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.582169709989259e-05,
+      "loss": 0.6065,
+      "step": 1217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.581811672037236e-05,
+      "loss": 0.5409,
+      "step": 1218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.5814536340852135e-05,
+      "loss": 0.6012,
+      "step": 1219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.58109559613319e-05,
+      "loss": 0.7826,
+      "step": 1220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5807375581811674e-05,
+      "loss": 0.6438,
+      "step": 1221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.580379520229145e-05,
+      "loss": 0.6654,
+      "step": 1222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5800214822771214e-05,
+      "loss": 0.5663,
+      "step": 1223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.579663444325099e-05,
+      "loss": 0.6455,
+      "step": 1224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5793054063730753e-05,
+      "loss": 0.6404,
+      "step": 1225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5789473684210527e-05,
+      "loss": 0.53,
+      "step": 1226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.57858933046903e-05,
+      "loss": 0.5531,
+      "step": 1227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5782312925170066e-05,
+      "loss": 0.5585,
+      "step": 1228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.577873254564984e-05,
+      "loss": 0.6588,
+      "step": 1229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.577515216612961e-05,
+      "loss": 0.6389,
+      "step": 1230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5771571786609386e-05,
+      "loss": 0.5763,
+      "step": 1231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.576799140708916e-05,
+      "loss": 0.6936,
+      "step": 1232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5764411027568925e-05,
+      "loss": 0.4588,
+      "step": 1233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.57608306480487e-05,
+      "loss": 0.6626,
+      "step": 1234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5757250268528465e-05,
+      "loss": 0.5558,
+      "step": 1235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.575366988900824e-05,
+      "loss": 0.5934,
+      "step": 1236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.575008950948801e-05,
+      "loss": 0.6543,
+      "step": 1237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.574650912996778e-05,
+      "loss": 0.5989,
+      "step": 1238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.574292875044755e-05,
+      "loss": 0.5251,
+      "step": 1239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5739348370927324e-05,
+      "loss": 0.7058,
+      "step": 1240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.573576799140709e-05,
+      "loss": 0.643,
+      "step": 1241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5732187611886864e-05,
+      "loss": 0.5821,
+      "step": 1242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.572860723236663e-05,
+      "loss": 0.5795,
+      "step": 1243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.57250268528464e-05,
+      "loss": 0.6112,
+      "step": 1244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5721446473326176e-05,
+      "loss": 0.605,
+      "step": 1245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.571786609380594e-05,
+      "loss": 0.621,
+      "step": 1246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.5714285714285716e-05,
+      "loss": 0.4887,
+      "step": 1247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.571070533476548e-05,
+      "loss": 0.698,
+      "step": 1248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5707124955245256e-05,
+      "loss": 0.5479,
+      "step": 1249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.570354457572503e-05,
+      "loss": 0.6316,
+      "step": 1250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5699964196204795e-05,
+      "loss": 0.6238,
+      "step": 1251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.569638381668457e-05,
+      "loss": 0.5282,
+      "step": 1252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.569280343716434e-05,
+      "loss": 0.6507,
+      "step": 1253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5689223057644115e-05,
+      "loss": 0.6992,
+      "step": 1254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.568564267812389e-05,
+      "loss": 0.655,
+      "step": 1255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5682062298603654e-05,
+      "loss": 0.5074,
+      "step": 1256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.567848191908343e-05,
+      "loss": 0.5345,
+      "step": 1257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5674901539563194e-05,
+      "loss": 0.6028,
+      "step": 1258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.567132116004297e-05,
+      "loss": 0.6915,
+      "step": 1259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.566774078052274e-05,
+      "loss": 0.6796,
+      "step": 1260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.566416040100251e-05,
+      "loss": 0.5023,
+      "step": 1261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.566058002148228e-05,
+      "loss": 0.6313,
+      "step": 1262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.565699964196205e-05,
+      "loss": 0.6047,
+      "step": 1263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.565341926244182e-05,
+      "loss": 0.6533,
+      "step": 1264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.564983888292159e-05,
+      "loss": 0.6586,
+      "step": 1265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.564625850340136e-05,
+      "loss": 0.5987,
+      "step": 1266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.564267812388113e-05,
+      "loss": 0.5772,
+      "step": 1267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5639097744360906e-05,
+      "loss": 0.651,
+      "step": 1268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.563551736484067e-05,
+      "loss": 0.6039,
+      "step": 1269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5631936985320445e-05,
+      "loss": 0.6249,
+      "step": 1270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562835660580022e-05,
+      "loss": 0.5785,
+      "step": 1271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5624776226279985e-05,
+      "loss": 0.5673,
+      "step": 1272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.562119584675976e-05,
+      "loss": 0.6881,
+      "step": 1273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.5617615467239524e-05,
+      "loss": 0.6203,
+      "step": 1274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 0.5912,
+      "step": 1275
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.561045470819907e-05,
+      "loss": 0.6112,
+      "step": 1276
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5606874328678844e-05,
+      "loss": 0.6251,
+      "step": 1277
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.560329394915862e-05,
+      "loss": 0.7052,
+      "step": 1278
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5599713569638384e-05,
+      "loss": 0.6772,
+      "step": 1279
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.559613319011816e-05,
+      "loss": 0.6207,
+      "step": 1280
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.559255281059793e-05,
+      "loss": 0.5513,
+      "step": 1281
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5588972431077696e-05,
+      "loss": 0.5603,
+      "step": 1282
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.558539205155747e-05,
+      "loss": 0.623,
+      "step": 1283
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5581811672037236e-05,
+      "loss": 0.5386,
+      "step": 1284
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.557823129251701e-05,
+      "loss": 0.6579,
+      "step": 1285
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.557465091299678e-05,
+      "loss": 0.648,
+      "step": 1286
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.557107053347655e-05,
+      "loss": 0.604,
+      "step": 1287
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.556749015395632e-05,
+      "loss": 0.6753,
+      "step": 1288
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5563909774436095e-05,
+      "loss": 0.5936,
+      "step": 1289
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.556032939491586e-05,
+      "loss": 0.6524,
+      "step": 1290
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5556749015395635e-05,
+      "loss": 0.6395,
+      "step": 1291
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.55531686358754e-05,
+      "loss": 0.5559,
+      "step": 1292
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5549588256355174e-05,
+      "loss": 0.6252,
+      "step": 1293
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.554600787683495e-05,
+      "loss": 0.6472,
+      "step": 1294
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5542427497314714e-05,
+      "loss": 0.5291,
+      "step": 1295
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.553884711779449e-05,
+      "loss": 0.5875,
+      "step": 1296
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.553526673827426e-05,
+      "loss": 0.6038,
+      "step": 1297
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.553168635875403e-05,
+      "loss": 0.5771,
+      "step": 1298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.55281059792338e-05,
+      "loss": 0.6105,
+      "step": 1299
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.552452559971357e-05,
+      "loss": 0.6352,
+      "step": 1300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5520945220193346e-05,
+      "loss": 0.5363,
+      "step": 1301
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.551736484067311e-05,
+      "loss": 0.6191,
+      "step": 1302
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.5513784461152886e-05,
+      "loss": 0.4904,
+      "step": 1303
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.551020408163266e-05,
+      "loss": 0.5796,
+      "step": 1304
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5506623702112425e-05,
+      "loss": 0.6163,
+      "step": 1305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.55030433225922e-05,
+      "loss": 0.6488,
+      "step": 1306
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.549946294307197e-05,
+      "loss": 0.7144,
+      "step": 1307
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.549588256355174e-05,
+      "loss": 0.6226,
+      "step": 1308
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.549230218403151e-05,
+      "loss": 0.5813,
+      "step": 1309
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.548872180451128e-05,
+      "loss": 0.5756,
+      "step": 1310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.548514142499105e-05,
+      "loss": 0.581,
+      "step": 1311
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5481561045470824e-05,
+      "loss": 0.5377,
+      "step": 1312
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.547798066595059e-05,
+      "loss": 0.6405,
+      "step": 1313
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5474400286430364e-05,
+      "loss": 0.5658,
+      "step": 1314
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.547081990691014e-05,
+      "loss": 0.6155,
+      "step": 1315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5467239527389903e-05,
+      "loss": 0.6922,
+      "step": 1316
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5463659147869677e-05,
+      "loss": 0.5488,
+      "step": 1317
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.546007876834944e-05,
+      "loss": 0.5357,
+      "step": 1318
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5456498388829216e-05,
+      "loss": 0.561,
+      "step": 1319
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.545291800930899e-05,
+      "loss": 0.6186,
+      "step": 1320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5449337629788756e-05,
+      "loss": 0.5704,
+      "step": 1321
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.544575725026853e-05,
+      "loss": 0.5193,
+      "step": 1322
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.54421768707483e-05,
+      "loss": 0.622,
+      "step": 1323
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5438596491228075e-05,
+      "loss": 0.5216,
+      "step": 1324
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.543501611170785e-05,
+      "loss": 0.5669,
+      "step": 1325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5431435732187615e-05,
+      "loss": 0.6909,
+      "step": 1326
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542785535266739e-05,
+      "loss": 0.6022,
+      "step": 1327
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.5424274973147155e-05,
+      "loss": 0.5354,
+      "step": 1328
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.542069459362693e-05,
+      "loss": 0.588,
+      "step": 1329
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.54171142141067e-05,
+      "loss": 0.6173,
+      "step": 1330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.541353383458647e-05,
+      "loss": 0.6279,
+      "step": 1331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.540995345506624e-05,
+      "loss": 0.667,
+      "step": 1332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5406373075546014e-05,
+      "loss": 0.469,
+      "step": 1333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.540279269602578e-05,
+      "loss": 0.5679,
+      "step": 1334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.539921231650555e-05,
+      "loss": 0.5864,
+      "step": 1335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.539563193698532e-05,
+      "loss": 0.5953,
+      "step": 1336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.539205155746509e-05,
+      "loss": 0.6446,
+      "step": 1337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5388471177944866e-05,
+      "loss": 0.5452,
+      "step": 1338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.538489079842463e-05,
+      "loss": 0.4833,
+      "step": 1339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5381310418904406e-05,
+      "loss": 0.7119,
+      "step": 1340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.537773003938417e-05,
+      "loss": 0.6352,
+      "step": 1341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5374149659863945e-05,
+      "loss": 0.6251,
+      "step": 1342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.537056928034372e-05,
+      "loss": 0.4452,
+      "step": 1343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5366988900823485e-05,
+      "loss": 0.5991,
+      "step": 1344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.536340852130326e-05,
+      "loss": 0.6204,
+      "step": 1345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.535982814178303e-05,
+      "loss": 0.5764,
+      "step": 1346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5356247762262804e-05,
+      "loss": 0.685,
+      "step": 1347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.535266738274258e-05,
+      "loss": 0.605,
+      "step": 1348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5349087003222344e-05,
+      "loss": 0.5017,
+      "step": 1349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.534550662370212e-05,
+      "loss": 0.6132,
+      "step": 1350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.534192624418189e-05,
+      "loss": 0.6392,
+      "step": 1351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.533834586466166e-05,
+      "loss": 0.6204,
+      "step": 1352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.533476548514143e-05,
+      "loss": 0.6816,
+      "step": 1353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.5331185105621196e-05,
+      "loss": 0.5694,
+      "step": 1354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532760472610097e-05,
+      "loss": 0.6299,
+      "step": 1355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532402434658074e-05,
+      "loss": 0.6828,
+      "step": 1356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.532044396706051e-05,
+      "loss": 0.6593,
+      "step": 1357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.531686358754028e-05,
+      "loss": 0.6334,
+      "step": 1358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.531328320802005e-05,
+      "loss": 0.6353,
+      "step": 1359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.530970282849982e-05,
+      "loss": 0.5285,
+      "step": 1360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5306122448979595e-05,
+      "loss": 0.6161,
+      "step": 1361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.530254206945936e-05,
+      "loss": 0.5342,
+      "step": 1362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5298961689939135e-05,
+      "loss": 0.6492,
+      "step": 1363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.529538131041891e-05,
+      "loss": 0.5478,
+      "step": 1364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5291800930898674e-05,
+      "loss": 0.6477,
+      "step": 1365
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.528822055137845e-05,
+      "loss": 0.5803,
+      "step": 1366
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5284640171858214e-05,
+      "loss": 0.5768,
+      "step": 1367
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.528105979233799e-05,
+      "loss": 0.6276,
+      "step": 1368
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.527747941281776e-05,
+      "loss": 0.5352,
+      "step": 1369
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5273899033297534e-05,
+      "loss": 0.6251,
+      "step": 1370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.527031865377731e-05,
+      "loss": 0.6439,
+      "step": 1371
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.526673827425707e-05,
+      "loss": 0.592,
+      "step": 1372
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5263157894736846e-05,
+      "loss": 0.579,
+      "step": 1373
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.525957751521662e-05,
+      "loss": 0.6963,
+      "step": 1374
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5255997135696386e-05,
+      "loss": 0.6745,
+      "step": 1375
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.525241675617616e-05,
+      "loss": 0.7231,
+      "step": 1376
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5248836376655926e-05,
+      "loss": 0.5597,
+      "step": 1377
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.52452559971357e-05,
+      "loss": 0.6008,
+      "step": 1378
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.524167561761547e-05,
+      "loss": 0.6338,
+      "step": 1379
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.5436,
+      "step": 1380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.523451485857501e-05,
+      "loss": 0.5605,
+      "step": 1381
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5230934479054785e-05,
+      "loss": 0.594,
+      "step": 1382
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522735409953455e-05,
+      "loss": 0.5767,
+      "step": 1383
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5223773720014324e-05,
+      "loss": 0.5586,
+      "step": 1384
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.522019334049409e-05,
+      "loss": 0.5544,
+      "step": 1385
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.5216612960973864e-05,
+      "loss": 0.5581,
+      "step": 1386
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.521303258145364e-05,
+      "loss": 0.6487,
+      "step": 1387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5209452201933403e-05,
+      "loss": 0.5819,
+      "step": 1388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.520587182241318e-05,
+      "loss": 0.5502,
+      "step": 1389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.520229144289295e-05,
+      "loss": 0.5956,
+      "step": 1390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5198711063372716e-05,
+      "loss": 0.6002,
+      "step": 1391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.519513068385249e-05,
+      "loss": 0.6072,
+      "step": 1392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.519155030433226e-05,
+      "loss": 0.6331,
+      "step": 1393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5187969924812036e-05,
+      "loss": 0.5829,
+      "step": 1394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.51843895452918e-05,
+      "loss": 0.4877,
+      "step": 1395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5180809165771575e-05,
+      "loss": 0.5958,
+      "step": 1396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.517722878625135e-05,
+      "loss": 0.6928,
+      "step": 1397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5173648406731115e-05,
+      "loss": 0.4927,
+      "step": 1398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.517006802721089e-05,
+      "loss": 0.7108,
+      "step": 1399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.516648764769066e-05,
+      "loss": 0.5632,
+      "step": 1400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.516290726817043e-05,
+      "loss": 0.5975,
+      "step": 1401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.51593268886502e-05,
+      "loss": 0.6039,
+      "step": 1402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.515574650912997e-05,
+      "loss": 0.6618,
+      "step": 1403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.515216612960974e-05,
+      "loss": 0.4991,
+      "step": 1404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5148585750089514e-05,
+      "loss": 0.5985,
+      "step": 1405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.514500537056928e-05,
+      "loss": 0.594,
+      "step": 1406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5141424991049053e-05,
+      "loss": 0.7162,
+      "step": 1407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5137844611528827e-05,
+      "loss": 0.652,
+      "step": 1408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.513426423200859e-05,
+      "loss": 0.6115,
+      "step": 1409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5130683852488366e-05,
+      "loss": 0.5318,
+      "step": 1410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.512710347296813e-05,
+      "loss": 0.6125,
+      "step": 1411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5123523093447906e-05,
+      "loss": 0.5787,
+      "step": 1412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511994271392768e-05,
+      "loss": 0.6564,
+      "step": 1413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.5116362334407445e-05,
+      "loss": 0.6142,
+      "step": 1414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.511278195488722e-05,
+      "loss": 0.6241,
+      "step": 1415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5109201575366985e-05,
+      "loss": 0.6926,
+      "step": 1416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5105621195846765e-05,
+      "loss": 0.6123,
+      "step": 1417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.510204081632654e-05,
+      "loss": 0.5987,
+      "step": 1418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5098460436806305e-05,
+      "loss": 0.6282,
+      "step": 1419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.509488005728608e-05,
+      "loss": 0.56,
+      "step": 1420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5091299677765844e-05,
+      "loss": 0.5879,
+      "step": 1421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.508771929824562e-05,
+      "loss": 0.6116,
+      "step": 1422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.508413891872539e-05,
+      "loss": 0.5266,
+      "step": 1423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.508055853920516e-05,
+      "loss": 0.6385,
+      "step": 1424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.507697815968493e-05,
+      "loss": 0.5186,
+      "step": 1425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.50733977801647e-05,
+      "loss": 0.626,
+      "step": 1426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.506981740064447e-05,
+      "loss": 0.6803,
+      "step": 1427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.506623702112424e-05,
+      "loss": 0.6777,
+      "step": 1428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.506265664160401e-05,
+      "loss": 0.5614,
+      "step": 1429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.505907626208378e-05,
+      "loss": 0.611,
+      "step": 1430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5055495882563556e-05,
+      "loss": 0.618,
+      "step": 1431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.505191550304332e-05,
+      "loss": 0.5187,
+      "step": 1432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5048335123523095e-05,
+      "loss": 0.6462,
+      "step": 1433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.504475474400286e-05,
+      "loss": 0.5559,
+      "step": 1434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5041174364482635e-05,
+      "loss": 0.5559,
+      "step": 1435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.503759398496241e-05,
+      "loss": 0.5214,
+      "step": 1436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5034013605442174e-05,
+      "loss": 0.5505,
+      "step": 1437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.503043322592195e-05,
+      "loss": 0.6321,
+      "step": 1438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.502685284640172e-05,
+      "loss": 0.5473,
+      "step": 1439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5023272466881494e-05,
+      "loss": 0.6754,
+      "step": 1440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501969208736127e-05,
+      "loss": 0.6333,
+      "step": 1441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.5016111707841034e-05,
+      "loss": 0.6189,
+      "step": 1442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.501253132832081e-05,
+      "loss": 0.5874,
+      "step": 1443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.500895094880058e-05,
+      "loss": 0.5685,
+      "step": 1444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.5005370569280346e-05,
+      "loss": 0.6053,
+      "step": 1445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.500179018976012e-05,
+      "loss": 0.6454,
+      "step": 1446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4998209810239886e-05,
+      "loss": 0.5845,
+      "step": 1447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.499462943071966e-05,
+      "loss": 0.5664,
+      "step": 1448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.499104905119943e-05,
+      "loss": 0.5718,
+      "step": 1449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.49874686716792e-05,
+      "loss": 0.5774,
+      "step": 1450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.498388829215897e-05,
+      "loss": 0.6617,
+      "step": 1451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.498030791263874e-05,
+      "loss": 0.6046,
+      "step": 1452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.497672753311851e-05,
+      "loss": 0.5685,
+      "step": 1453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4973147153598285e-05,
+      "loss": 0.6086,
+      "step": 1454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.496956677407805e-05,
+      "loss": 0.5459,
+      "step": 1455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4965986394557824e-05,
+      "loss": 0.6609,
+      "step": 1456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.49624060150376e-05,
+      "loss": 0.6074,
+      "step": 1457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4958825635517364e-05,
+      "loss": 0.6126,
+      "step": 1458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.495524525599714e-05,
+      "loss": 0.5966,
+      "step": 1459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4951664876476904e-05,
+      "loss": 0.6241,
+      "step": 1460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.494808449695668e-05,
+      "loss": 0.5558,
+      "step": 1461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.494450411743645e-05,
+      "loss": 0.6559,
+      "step": 1462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.494092373791622e-05,
+      "loss": 0.6989,
+      "step": 1463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4937343358395996e-05,
+      "loss": 0.601,
+      "step": 1464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.493376297887576e-05,
+      "loss": 0.6591,
+      "step": 1465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4930182599355536e-05,
+      "loss": 0.6226,
+      "step": 1466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.492660221983531e-05,
+      "loss": 0.5921,
+      "step": 1467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4923021840315076e-05,
+      "loss": 0.6414,
+      "step": 1468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.491944146079485e-05,
+      "loss": 0.6509,
+      "step": 1469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.4915861081274615e-05,
+      "loss": 0.555,
+      "step": 1470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.491228070175439e-05,
+      "loss": 0.6069,
+      "step": 1471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.490870032223416e-05,
+      "loss": 0.6315,
+      "step": 1472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.490511994271393e-05,
+      "loss": 0.573,
+      "step": 1473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.49015395631937e-05,
+      "loss": 0.576,
+      "step": 1474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4897959183673474e-05,
+      "loss": 0.5823,
+      "step": 1475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.489437880415324e-05,
+      "loss": 0.6114,
+      "step": 1476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4890798424633014e-05,
+      "loss": 0.5705,
+      "step": 1477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.488721804511278e-05,
+      "loss": 0.6238,
+      "step": 1478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4883637665592554e-05,
+      "loss": 0.5922,
+      "step": 1479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.488005728607233e-05,
+      "loss": 0.5138,
+      "step": 1480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.487647690655209e-05,
+      "loss": 0.5931,
+      "step": 1481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4872896527031866e-05,
+      "loss": 0.6637,
+      "step": 1482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.486931614751164e-05,
+      "loss": 0.5682,
+      "step": 1483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4865735767991406e-05,
+      "loss": 0.5952,
+      "step": 1484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.486215538847118e-05,
+      "loss": 0.6338,
+      "step": 1485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4858575008950945e-05,
+      "loss": 0.5874,
+      "step": 1486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4854994629430725e-05,
+      "loss": 0.5771,
+      "step": 1487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.485141424991049e-05,
+      "loss": 0.5712,
+      "step": 1488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4847833870390265e-05,
+      "loss": 0.5979,
+      "step": 1489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.484425349087004e-05,
+      "loss": 0.5108,
+      "step": 1490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4840673111349805e-05,
+      "loss": 0.5922,
+      "step": 1491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.483709273182958e-05,
+      "loss": 0.5498,
+      "step": 1492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.483351235230935e-05,
+      "loss": 0.5532,
+      "step": 1493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482993197278912e-05,
+      "loss": 0.6701,
+      "step": 1494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482635159326889e-05,
+      "loss": 0.6121,
+      "step": 1495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.482277121374866e-05,
+      "loss": 0.5311,
+      "step": 1496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481919083422843e-05,
+      "loss": 0.417,
+      "step": 1497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.4815610454708203e-05,
+      "loss": 0.5659,
+      "step": 1498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.481203007518797e-05,
+      "loss": 0.6641,
+      "step": 1499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.480844969566774e-05,
+      "loss": 0.6595,
+      "step": 1500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4804869316147516e-05,
+      "loss": 0.6554,
+      "step": 1501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.480128893662728e-05,
+      "loss": 0.6241,
+      "step": 1502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4797708557107056e-05,
+      "loss": 0.678,
+      "step": 1503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.479412817758682e-05,
+      "loss": 0.5559,
+      "step": 1504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4790547798066595e-05,
+      "loss": 0.6223,
+      "step": 1505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.478696741854637e-05,
+      "loss": 0.4652,
+      "step": 1506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4783387039026135e-05,
+      "loss": 0.4802,
+      "step": 1507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.477980665950591e-05,
+      "loss": 0.6548,
+      "step": 1508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.477622627998568e-05,
+      "loss": 0.5873,
+      "step": 1509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4772645900465455e-05,
+      "loss": 0.5789,
+      "step": 1510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.476906552094523e-05,
+      "loss": 0.6162,
+      "step": 1511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4765485141424994e-05,
+      "loss": 0.6971,
+      "step": 1512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.476190476190477e-05,
+      "loss": 0.6091,
+      "step": 1513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4758324382384534e-05,
+      "loss": 0.6469,
+      "step": 1514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.475474400286431e-05,
+      "loss": 0.5736,
+      "step": 1515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.475116362334408e-05,
+      "loss": 0.5954,
+      "step": 1516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4747583243823847e-05,
+      "loss": 0.5116,
+      "step": 1517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.474400286430362e-05,
+      "loss": 0.5675,
+      "step": 1518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.474042248478339e-05,
+      "loss": 0.5664,
+      "step": 1519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.473684210526316e-05,
+      "loss": 0.6149,
+      "step": 1520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.473326172574293e-05,
+      "loss": 0.6712,
+      "step": 1521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.47296813462227e-05,
+      "loss": 0.5912,
+      "step": 1522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.472610096670247e-05,
+      "loss": 0.5311,
+      "step": 1523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4722520587182245e-05,
+      "loss": 0.5679,
+      "step": 1524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471894020766201e-05,
+      "loss": 0.5523,
+      "step": 1525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4715359828141785e-05,
+      "loss": 0.5278,
+      "step": 1526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.471177944862155e-05,
+      "loss": 0.5091,
+      "step": 1527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4708199069101324e-05,
+      "loss": 0.5762,
+      "step": 1528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47046186895811e-05,
+      "loss": 0.5392,
+      "step": 1529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4701038310060864e-05,
+      "loss": 0.6172,
+      "step": 1530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469745793054064e-05,
+      "loss": 0.6087,
+      "step": 1531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469387755102041e-05,
+      "loss": 0.5225,
+      "step": 1532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4690297171500184e-05,
+      "loss": 0.5681,
+      "step": 1533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468671679197996e-05,
+      "loss": 0.5435,
+      "step": 1534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468313641245972e-05,
+      "loss": 0.5714,
+      "step": 1535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4679556032939496e-05,
+      "loss": 0.6349,
+      "step": 1536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467597565341927e-05,
+      "loss": 0.5765,
+      "step": 1537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4672395273899036e-05,
+      "loss": 0.5316,
+      "step": 1538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.466881489437881e-05,
+      "loss": 0.5947,
+      "step": 1539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4665234514858576e-05,
+      "loss": 0.5568,
+      "step": 1540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.466165413533835e-05,
+      "loss": 0.58,
+      "step": 1541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465807375581812e-05,
+      "loss": 0.6005,
+      "step": 1542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465449337629789e-05,
+      "loss": 0.4952,
+      "step": 1543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465091299677766e-05,
+      "loss": 0.6461,
+      "step": 1544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464733261725743e-05,
+      "loss": 0.5788,
+      "step": 1545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46437522377372e-05,
+      "loss": 0.5522,
+      "step": 1546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4640171858216974e-05,
+      "loss": 0.5527,
+      "step": 1547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463659147869674e-05,
+      "loss": 0.6151,
+      "step": 1548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4633011099176514e-05,
+      "loss": 0.6223,
+      "step": 1549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462943071965629e-05,
+      "loss": 0.5724,
+      "step": 1550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4625850340136054e-05,
+      "loss": 0.6143,
+      "step": 1551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.462226996061583e-05,
+      "loss": 0.6349,
+      "step": 1552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461868958109559e-05,
+      "loss": 0.5159,
+      "step": 1553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4615109201575366e-05,
+      "loss": 0.6576,
+      "step": 1554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.461152882205514e-05,
+      "loss": 0.596,
+      "step": 1555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4607948442534906e-05,
+      "loss": 0.6156,
+      "step": 1556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4604368063014686e-05,
+      "loss": 0.5106,
+      "step": 1557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.460078768349445e-05,
+      "loss": 0.5206,
+      "step": 1558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4597207303974226e-05,
+      "loss": 0.6039,
+      "step": 1559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4593626924454e-05,
+      "loss": 0.4772,
+      "step": 1560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4590046544933765e-05,
+      "loss": 0.4786,
+      "step": 1561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.458646616541354e-05,
+      "loss": 0.548,
+      "step": 1562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4582885785893305e-05,
+      "loss": 0.5059,
+      "step": 1563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.457930540637308e-05,
+      "loss": 0.6421,
+      "step": 1564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.457572502685285e-05,
+      "loss": 0.6247,
+      "step": 1565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.457214464733262e-05,
+      "loss": 0.5947,
+      "step": 1566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456856426781239e-05,
+      "loss": 0.5823,
+      "step": 1567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4564983888292164e-05,
+      "loss": 0.6371,
+      "step": 1568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456140350877193e-05,
+      "loss": 0.6449,
+      "step": 1569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4557823129251704e-05,
+      "loss": 0.6754,
+      "step": 1570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.455424274973147e-05,
+      "loss": 0.4881,
+      "step": 1571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.455066237021124e-05,
+      "loss": 0.7085,
+      "step": 1572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4547081990691016e-05,
+      "loss": 0.5704,
+      "step": 1573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.454350161117078e-05,
+      "loss": 0.6045,
+      "step": 1574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4539921231650556e-05,
+      "loss": 0.5463,
+      "step": 1575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.453634085213033e-05,
+      "loss": 0.572,
+      "step": 1576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4532760472610095e-05,
+      "loss": 0.5698,
+      "step": 1577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452918009308987e-05,
+      "loss": 0.5057,
+      "step": 1578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4525599713569635e-05,
+      "loss": 0.6738,
+      "step": 1579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4522019334049415e-05,
+      "loss": 0.6001,
+      "step": 1580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451843895452918e-05,
+      "loss": 0.5715,
+      "step": 1581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4514858575008955e-05,
+      "loss": 0.6163,
+      "step": 1582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451127819548873e-05,
+      "loss": 0.6159,
+      "step": 1583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4507697815968494e-05,
+      "loss": 0.6334,
+      "step": 1584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.450411743644827e-05,
+      "loss": 0.5583,
+      "step": 1585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.450053705692804e-05,
+      "loss": 0.6087,
+      "step": 1586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.449695667740781e-05,
+      "loss": 0.5355,
+      "step": 1587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.449337629788758e-05,
+      "loss": 0.4843,
+      "step": 1588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.448979591836735e-05,
+      "loss": 0.5777,
+      "step": 1589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.448621553884712e-05,
+      "loss": 0.5346,
+      "step": 1590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.448263515932689e-05,
+      "loss": 0.6187,
+      "step": 1591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.447905477980666e-05,
+      "loss": 0.6896,
+      "step": 1592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.447547440028643e-05,
+      "loss": 0.6801,
+      "step": 1593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4471894020766206e-05,
+      "loss": 0.4759,
+      "step": 1594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.446831364124597e-05,
+      "loss": 0.551,
+      "step": 1595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4464733261725745e-05,
+      "loss": 0.5738,
+      "step": 1596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.446115288220551e-05,
+      "loss": 0.5983,
+      "step": 1597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4457572502685285e-05,
+      "loss": 0.5627,
+      "step": 1598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.445399212316506e-05,
+      "loss": 0.4733,
+      "step": 1599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4450411743644825e-05,
+      "loss": 0.5362,
+      "step": 1600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.44468313641246e-05,
+      "loss": 0.5614,
+      "step": 1601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.444325098460437e-05,
+      "loss": 0.5023,
+      "step": 1602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4439670605084144e-05,
+      "loss": 0.5647,
+      "step": 1603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.443609022556392e-05,
+      "loss": 0.5889,
+      "step": 1604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4432509846043684e-05,
+      "loss": 0.5965,
+      "step": 1605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442892946652346e-05,
+      "loss": 0.5547,
+      "step": 1606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.442534908700322e-05,
+      "loss": 0.6271,
+      "step": 1607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4421768707482997e-05,
+      "loss": 0.5556,
+      "step": 1608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.441818832796277e-05,
+      "loss": 0.5789,
+      "step": 1609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4414607948442536e-05,
+      "loss": 0.517,
+      "step": 1610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.441102756892231e-05,
+      "loss": 0.6455,
+      "step": 1611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.440744718940208e-05,
+      "loss": 0.589,
+      "step": 1612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.440386680988185e-05,
+      "loss": 0.6421,
+      "step": 1613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.440028643036162e-05,
+      "loss": 0.5981,
+      "step": 1614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.439670605084139e-05,
+      "loss": 0.6576,
+      "step": 1615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.439312567132116e-05,
+      "loss": 0.5703,
+      "step": 1616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4389545291800935e-05,
+      "loss": 0.4365,
+      "step": 1617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43859649122807e-05,
+      "loss": 0.5076,
+      "step": 1618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4382384532760475e-05,
+      "loss": 0.5462,
+      "step": 1619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.437880415324025e-05,
+      "loss": 0.6459,
+      "step": 1620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4375223773720014e-05,
+      "loss": 0.5757,
+      "step": 1621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.437164339419979e-05,
+      "loss": 0.6967,
+      "step": 1622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4368063014679554e-05,
+      "loss": 0.5278,
+      "step": 1623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.436448263515933e-05,
+      "loss": 0.5205,
+      "step": 1624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43609022556391e-05,
+      "loss": 0.6064,
+      "step": 1625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4357321876118866e-05,
+      "loss": 0.6615,
+      "step": 1626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4353741496598646e-05,
+      "loss": 0.6491,
+      "step": 1627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.435016111707841e-05,
+      "loss": 0.6626,
+      "step": 1628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4346580737558186e-05,
+      "loss": 0.593,
+      "step": 1629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.434300035803796e-05,
+      "loss": 0.6242,
+      "step": 1630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4339419978517726e-05,
+      "loss": 0.6035,
+      "step": 1631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.43358395989975e-05,
+      "loss": 0.533,
+      "step": 1632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4332259219477265e-05,
+      "loss": 0.7427,
+      "step": 1633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432867883995704e-05,
+      "loss": 0.6095,
+      "step": 1634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432509846043681e-05,
+      "loss": 0.5284,
+      "step": 1635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.432151808091658e-05,
+      "loss": 0.5954,
+      "step": 1636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431793770139635e-05,
+      "loss": 0.66,
+      "step": 1637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431435732187612e-05,
+      "loss": 0.513,
+      "step": 1638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.431077694235589e-05,
+      "loss": 0.5648,
+      "step": 1639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4307196562835664e-05,
+      "loss": 0.5534,
+      "step": 1640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.430361618331543e-05,
+      "loss": 0.5335,
+      "step": 1641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4300035803795204e-05,
+      "loss": 0.5943,
+      "step": 1642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.429645542427498e-05,
+      "loss": 0.5934,
+      "step": 1643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.429287504475474e-05,
+      "loss": 0.5636,
+      "step": 1644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4289294665234516e-05,
+      "loss": 0.5353,
+      "step": 1645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.428571428571428e-05,
+      "loss": 0.5416,
+      "step": 1646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4282133906194056e-05,
+      "loss": 0.6131,
+      "step": 1647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.427855352667383e-05,
+      "loss": 0.4745,
+      "step": 1648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4274973147153596e-05,
+      "loss": 0.5616,
+      "step": 1649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4271392767633376e-05,
+      "loss": 0.5155,
+      "step": 1650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.426781238811314e-05,
+      "loss": 0.5055,
+      "step": 1651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4264232008592915e-05,
+      "loss": 0.6725,
+      "step": 1652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.426065162907269e-05,
+      "loss": 0.586,
+      "step": 1653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4257071249552455e-05,
+      "loss": 0.5439,
+      "step": 1654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.425349087003223e-05,
+      "loss": 0.558,
+      "step": 1655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4249910490511994e-05,
+      "loss": 0.5842,
+      "step": 1656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.424633011099177e-05,
+      "loss": 0.568,
+      "step": 1657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.424274973147154e-05,
+      "loss": 0.7056,
+      "step": 1658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.423916935195131e-05,
+      "loss": 0.6327,
+      "step": 1659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.423558897243108e-05,
+      "loss": 0.5781,
+      "step": 1660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4232008592910854e-05,
+      "loss": 0.4768,
+      "step": 1661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422842821339062e-05,
+      "loss": 0.5787,
+      "step": 1662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422484783387039e-05,
+      "loss": 0.6237,
+      "step": 1663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.422126745435016e-05,
+      "loss": 0.6425,
+      "step": 1664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421768707482993e-05,
+      "loss": 0.5489,
+      "step": 1665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.4214106695309706e-05,
+      "loss": 0.624,
+      "step": 1666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.421052631578947e-05,
+      "loss": 0.5854,
+      "step": 1667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4206945936269245e-05,
+      "loss": 0.511,
+      "step": 1668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.420336555674902e-05,
+      "loss": 0.5266,
+      "step": 1669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4199785177228785e-05,
+      "loss": 0.4752,
+      "step": 1670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.419620479770856e-05,
+      "loss": 0.5736,
+      "step": 1671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4192624418188325e-05,
+      "loss": 0.5388,
+      "step": 1672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.41890440386681e-05,
+      "loss": 0.5742,
+      "step": 1673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.418546365914787e-05,
+      "loss": 0.5816,
+      "step": 1674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4181883279627644e-05,
+      "loss": 0.5309,
+      "step": 1675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.417830290010742e-05,
+      "loss": 0.7421,
+      "step": 1676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4174722520587184e-05,
+      "loss": 0.546,
+      "step": 1677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.417114214106696e-05,
+      "loss": 0.4903,
+      "step": 1678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.416756176154673e-05,
+      "loss": 0.5564,
+      "step": 1679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.41639813820265e-05,
+      "loss": 0.5716,
+      "step": 1680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.416040100250627e-05,
+      "loss": 0.577,
+      "step": 1681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4156820622986036e-05,
+      "loss": 0.6849,
+      "step": 1682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.415324024346581e-05,
+      "loss": 0.498,
+      "step": 1683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.414965986394558e-05,
+      "loss": 0.5434,
+      "step": 1684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.414607948442535e-05,
+      "loss": 0.4949,
+      "step": 1685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.414249910490512e-05,
+      "loss": 0.5895,
+      "step": 1686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4138918725384895e-05,
+      "loss": 0.5622,
+      "step": 1687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.413533834586466e-05,
+      "loss": 0.5697,
+      "step": 1688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4131757966344435e-05,
+      "loss": 0.5218,
+      "step": 1689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.41281775868242e-05,
+      "loss": 0.5226,
+      "step": 1690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4124597207303975e-05,
+      "loss": 0.475,
+      "step": 1691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.412101682778375e-05,
+      "loss": 0.7018,
+      "step": 1692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.4117436448263514e-05,
+      "loss": 0.6124,
+      "step": 1693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411385606874329e-05,
+      "loss": 0.6216,
+      "step": 1694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.411027568922306e-05,
+      "loss": 0.5741,
+      "step": 1695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.410669530970283e-05,
+      "loss": 0.6843,
+      "step": 1696
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.410311493018261e-05,
+      "loss": 0.5556,
+      "step": 1697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.409953455066237e-05,
+      "loss": 0.5489,
+      "step": 1698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4095954171142147e-05,
+      "loss": 0.5112,
+      "step": 1699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.409237379162191e-05,
+      "loss": 0.5657,
+      "step": 1700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4088793412101686e-05,
+      "loss": 0.5506,
+      "step": 1701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.408521303258146e-05,
+      "loss": 0.5791,
+      "step": 1702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4081632653061226e-05,
+      "loss": 0.6327,
+      "step": 1703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4078052273541e-05,
+      "loss": 0.5488,
+      "step": 1704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.407447189402077e-05,
+      "loss": 0.5774,
+      "step": 1705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.407089151450054e-05,
+      "loss": 0.6631,
+      "step": 1706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.406731113498031e-05,
+      "loss": 0.5484,
+      "step": 1707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.406373075546008e-05,
+      "loss": 0.5741,
+      "step": 1708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.406015037593985e-05,
+      "loss": 0.5808,
+      "step": 1709
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4056569996419625e-05,
+      "loss": 0.5774,
+      "step": 1710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.405298961689939e-05,
+      "loss": 0.5927,
+      "step": 1711
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4049409237379164e-05,
+      "loss": 0.5192,
+      "step": 1712
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.404582885785894e-05,
+      "loss": 0.5443,
+      "step": 1713
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4042248478338704e-05,
+      "loss": 0.5761,
+      "step": 1714
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.403866809881848e-05,
+      "loss": 0.4661,
+      "step": 1715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.403508771929824e-05,
+      "loss": 0.6437,
+      "step": 1716
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4031507339778016e-05,
+      "loss": 0.5932,
+      "step": 1717
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.402792696025779e-05,
+      "loss": 0.5955,
+      "step": 1718
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4024346580737556e-05,
+      "loss": 0.5573,
+      "step": 1719
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4020766201217336e-05,
+      "loss": 0.6023,
+      "step": 1720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.40171858216971e-05,
+      "loss": 0.6293,
+      "step": 1721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.4013605442176876e-05,
+      "loss": 0.6128,
+      "step": 1722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.401002506265665e-05,
+      "loss": 0.6772,
+      "step": 1723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4006444683136415e-05,
+      "loss": 0.5513,
+      "step": 1724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.400286430361619e-05,
+      "loss": 0.5122,
+      "step": 1725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3999283924095955e-05,
+      "loss": 0.6084,
+      "step": 1726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.399570354457573e-05,
+      "loss": 0.5187,
+      "step": 1727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.39921231650555e-05,
+      "loss": 0.5665,
+      "step": 1728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.398854278553527e-05,
+      "loss": 0.5808,
+      "step": 1729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.398496240601504e-05,
+      "loss": 0.52,
+      "step": 1730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3981382026494814e-05,
+      "loss": 0.5874,
+      "step": 1731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.397780164697458e-05,
+      "loss": 0.6608,
+      "step": 1732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3974221267454354e-05,
+      "loss": 0.6173,
+      "step": 1733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.397064088793412e-05,
+      "loss": 0.6382,
+      "step": 1734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.396706050841389e-05,
+      "loss": 0.58,
+      "step": 1735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3963480128893666e-05,
+      "loss": 0.5153,
+      "step": 1736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.395989974937343e-05,
+      "loss": 0.4597,
+      "step": 1737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3956319369853206e-05,
+      "loss": 0.5951,
+      "step": 1738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.395273899033297e-05,
+      "loss": 0.6288,
+      "step": 1739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3949158610812746e-05,
+      "loss": 0.5175,
+      "step": 1740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.394557823129252e-05,
+      "loss": 0.532,
+      "step": 1741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3941997851772285e-05,
+      "loss": 0.699,
+      "step": 1742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.393841747225206e-05,
+      "loss": 0.5902,
+      "step": 1743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.393483709273183e-05,
+      "loss": 0.532,
+      "step": 1744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3931256713211605e-05,
+      "loss": 0.4617,
+      "step": 1745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.392767633369138e-05,
+      "loss": 0.4764,
+      "step": 1746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3924095954171144e-05,
+      "loss": 0.6017,
+      "step": 1747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.392051557465092e-05,
+      "loss": 0.5482,
+      "step": 1748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.3916935195130684e-05,
+      "loss": 0.6826,
+      "step": 1749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.391335481561046e-05,
+      "loss": 0.5642,
+      "step": 1750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.390977443609023e-05,
+      "loss": 0.5885,
+      "step": 1751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.390619405657e-05,
+      "loss": 0.5092,
+      "step": 1752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.390261367704977e-05,
+      "loss": 0.504,
+      "step": 1753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.389903329752954e-05,
+      "loss": 0.5721,
+      "step": 1754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.389545291800931e-05,
+      "loss": 0.5995,
+      "step": 1755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.389187253848908e-05,
+      "loss": 0.5277,
+      "step": 1756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.388829215896885e-05,
+      "loss": 0.6987,
+      "step": 1757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.388471177944862e-05,
+      "loss": 0.6451,
+      "step": 1758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3881131399928395e-05,
+      "loss": 0.5194,
+      "step": 1759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.387755102040816e-05,
+      "loss": 0.6607,
+      "step": 1760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3873970640887935e-05,
+      "loss": 0.5656,
+      "step": 1761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.387039026136771e-05,
+      "loss": 0.5351,
+      "step": 1762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3866809881847475e-05,
+      "loss": 0.5519,
+      "step": 1763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.386322950232725e-05,
+      "loss": 0.5667,
+      "step": 1764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 0.7036,
+      "step": 1765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.385606874328679e-05,
+      "loss": 0.743,
+      "step": 1766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.385248836376656e-05,
+      "loss": 0.5666,
+      "step": 1767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3848907984246334e-05,
+      "loss": 0.5805,
+      "step": 1768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.384532760472611e-05,
+      "loss": 0.6224,
+      "step": 1769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3841747225205873e-05,
+      "loss": 0.6225,
+      "step": 1770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.383816684568565e-05,
+      "loss": 0.5741,
+      "step": 1771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.383458646616542e-05,
+      "loss": 0.6223,
+      "step": 1772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3831006086645186e-05,
+      "loss": 0.6328,
+      "step": 1773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.382742570712496e-05,
+      "loss": 0.5481,
+      "step": 1774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.3823845327604726e-05,
+      "loss": 0.7021,
+      "step": 1775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.38202649480845e-05,
+      "loss": 0.6657,
+      "step": 1776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381668456856427e-05,
+      "loss": 0.557,
+      "step": 1777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.381310418904404e-05,
+      "loss": 0.4988,
+      "step": 1778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.380952380952381e-05,
+      "loss": 0.623,
+      "step": 1779
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3805943430003585e-05,
+      "loss": 0.6144,
+      "step": 1780
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.380236305048335e-05,
+      "loss": 0.5874,
+      "step": 1781
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3798782670963125e-05,
+      "loss": 0.5982,
+      "step": 1782
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.379520229144289e-05,
+      "loss": 0.6197,
+      "step": 1783
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3791621911922664e-05,
+      "loss": 0.583,
+      "step": 1784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.378804153240244e-05,
+      "loss": 0.537,
+      "step": 1785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3784461152882204e-05,
+      "loss": 0.5788,
+      "step": 1786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.378088077336198e-05,
+      "loss": 0.5535,
+      "step": 1787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.377730039384175e-05,
+      "loss": 0.5758,
+      "step": 1788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3773720014321517e-05,
+      "loss": 0.4873,
+      "step": 1789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3770139634801297e-05,
+      "loss": 0.6618,
+      "step": 1790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.376655925528106e-05,
+      "loss": 0.6474,
+      "step": 1791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3762978875760836e-05,
+      "loss": 0.5596,
+      "step": 1792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.37593984962406e-05,
+      "loss": 0.5316,
+      "step": 1793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3755818116720376e-05,
+      "loss": 0.6508,
+      "step": 1794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.375223773720015e-05,
+      "loss": 0.5333,
+      "step": 1795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3748657357679915e-05,
+      "loss": 0.5417,
+      "step": 1796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.374507697815969e-05,
+      "loss": 0.595,
+      "step": 1797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.374149659863946e-05,
+      "loss": 0.5376,
+      "step": 1798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.373791621911923e-05,
+      "loss": 0.591,
+      "step": 1799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3734335839599e-05,
+      "loss": 0.6119,
+      "step": 1800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.373075546007877e-05,
+      "loss": 0.614,
+      "step": 1801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372717508055854e-05,
+      "loss": 0.6321,
+      "step": 1802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3723594701038314e-05,
+      "loss": 0.6899,
+      "step": 1803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.372001432151808e-05,
+      "loss": 0.5262,
+      "step": 1804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.3716433941997854e-05,
+      "loss": 0.5089,
+      "step": 1805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.371285356247763e-05,
+      "loss": 0.553,
+      "step": 1806
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.370927318295739e-05,
+      "loss": 0.454,
+      "step": 1807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3705692803437166e-05,
+      "loss": 0.5642,
+      "step": 1808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.370211242391693e-05,
+      "loss": 0.6454,
+      "step": 1809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3698532044396706e-05,
+      "loss": 0.6849,
+      "step": 1810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.369495166487648e-05,
+      "loss": 0.6618,
+      "step": 1811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3691371285356246e-05,
+      "loss": 0.5551,
+      "step": 1812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.368779090583602e-05,
+      "loss": 0.615,
+      "step": 1813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.368421052631579e-05,
+      "loss": 0.5259,
+      "step": 1814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3680630146795565e-05,
+      "loss": 0.5437,
+      "step": 1815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.367704976727534e-05,
+      "loss": 0.5036,
+      "step": 1816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3673469387755105e-05,
+      "loss": 0.6283,
+      "step": 1817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.366988900823488e-05,
+      "loss": 0.57,
+      "step": 1818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3666308628714644e-05,
+      "loss": 0.5112,
+      "step": 1819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.366272824919442e-05,
+      "loss": 0.5382,
+      "step": 1820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.365914786967419e-05,
+      "loss": 0.4672,
+      "step": 1821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.365556749015396e-05,
+      "loss": 0.633,
+      "step": 1822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.365198711063373e-05,
+      "loss": 0.5726,
+      "step": 1823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3648406731113504e-05,
+      "loss": 0.5635,
+      "step": 1824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.364482635159327e-05,
+      "loss": 0.5262,
+      "step": 1825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.364124597207304e-05,
+      "loss": 0.4827,
+      "step": 1826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.363766559255281e-05,
+      "loss": 0.6661,
+      "step": 1827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.363408521303258e-05,
+      "loss": 0.5094,
+      "step": 1828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3630504833512356e-05,
+      "loss": 0.6235,
+      "step": 1829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.362692445399212e-05,
+      "loss": 0.6308,
+      "step": 1830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3623344074471896e-05,
+      "loss": 0.6117,
+      "step": 1831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361976369495166e-05,
+      "loss": 0.5728,
+      "step": 1832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3616183315431435e-05,
+      "loss": 0.5946,
+      "step": 1833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.361260293591121e-05,
+      "loss": 0.5067,
+      "step": 1834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.3609022556390975e-05,
+      "loss": 0.6609,
+      "step": 1835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.360544217687075e-05,
+      "loss": 0.6028,
+      "step": 1836
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.360186179735052e-05,
+      "loss": 0.6133,
+      "step": 1837
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3598281417830294e-05,
+      "loss": 0.5761,
+      "step": 1838
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.359470103831007e-05,
+      "loss": 0.555,
+      "step": 1839
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3591120658789834e-05,
+      "loss": 0.5495,
+      "step": 1840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.358754027926961e-05,
+      "loss": 0.5613,
+      "step": 1841
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.358395989974938e-05,
+      "loss": 0.5829,
+      "step": 1842
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.358037952022915e-05,
+      "loss": 0.5161,
+      "step": 1843
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.357679914070892e-05,
+      "loss": 0.5699,
+      "step": 1844
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3573218761188686e-05,
+      "loss": 0.6122,
+      "step": 1845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.356963838166846e-05,
+      "loss": 0.5529,
+      "step": 1846
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.356605800214823e-05,
+      "loss": 0.6371,
+      "step": 1847
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3562477622628e-05,
+      "loss": 0.6103,
+      "step": 1848
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.355889724310777e-05,
+      "loss": 0.5255,
+      "step": 1849
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.355531686358754e-05,
+      "loss": 0.5652,
+      "step": 1850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.355173648406731e-05,
+      "loss": 0.5358,
+      "step": 1851
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3548156104547085e-05,
+      "loss": 0.5909,
+      "step": 1852
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.354457572502685e-05,
+      "loss": 0.512,
+      "step": 1853
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3540995345506625e-05,
+      "loss": 0.4926,
+      "step": 1854
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.35374149659864e-05,
+      "loss": 0.5834,
+      "step": 1855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3533834586466164e-05,
+      "loss": 0.5648,
+      "step": 1856
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.353025420694594e-05,
+      "loss": 0.5687,
+      "step": 1857
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3526673827425704e-05,
+      "loss": 0.5412,
+      "step": 1858
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.352309344790548e-05,
+      "loss": 0.5704,
+      "step": 1859
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.351951306838525e-05,
+      "loss": 0.7048,
+      "step": 1860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.3515932688865023e-05,
+      "loss": 0.6689,
+      "step": 1861
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.35123523093448e-05,
+      "loss": 0.4872,
+      "step": 1862
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.350877192982456e-05,
+      "loss": 0.5462,
+      "step": 1863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3505191550304336e-05,
+      "loss": 0.6642,
+      "step": 1864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.350161117078411e-05,
+      "loss": 0.5403,
+      "step": 1865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3498030791263876e-05,
+      "loss": 0.5752,
+      "step": 1866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.349445041174365e-05,
+      "loss": 0.6575,
+      "step": 1867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3490870032223415e-05,
+      "loss": 0.5337,
+      "step": 1868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.348728965270319e-05,
+      "loss": 0.5871,
+      "step": 1869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.348370927318296e-05,
+      "loss": 0.5808,
+      "step": 1870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.348012889366273e-05,
+      "loss": 0.5579,
+      "step": 1871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.34765485141425e-05,
+      "loss": 0.5518,
+      "step": 1872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3472968134622275e-05,
+      "loss": 0.5148,
+      "step": 1873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.346938775510204e-05,
+      "loss": 0.6324,
+      "step": 1874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3465807375581814e-05,
+      "loss": 0.566,
+      "step": 1875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.346222699606158e-05,
+      "loss": 0.6034,
+      "step": 1876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3458646616541354e-05,
+      "loss": 0.5314,
+      "step": 1877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.345506623702113e-05,
+      "loss": 0.5813,
+      "step": 1878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3451485857500893e-05,
+      "loss": 0.5657,
+      "step": 1879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3447905477980667e-05,
+      "loss": 0.6374,
+      "step": 1880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.344432509846044e-05,
+      "loss": 0.485,
+      "step": 1881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3440744718940206e-05,
+      "loss": 0.7071,
+      "step": 1882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.343716433941998e-05,
+      "loss": 0.5834,
+      "step": 1883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.343358395989975e-05,
+      "loss": 0.5357,
+      "step": 1884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3430003580379526e-05,
+      "loss": 0.5815,
+      "step": 1885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.342642320085929e-05,
+      "loss": 0.4756,
+      "step": 1886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3422842821339065e-05,
+      "loss": 0.6267,
+      "step": 1887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341926244181884e-05,
+      "loss": 0.5375,
+      "step": 1888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.3415682062298605e-05,
+      "loss": 0.5487,
+      "step": 1889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.341210168277838e-05,
+      "loss": 0.6143,
+      "step": 1890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340852130325815e-05,
+      "loss": 0.4972,
+      "step": 1891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.340494092373792e-05,
+      "loss": 0.6352,
+      "step": 1892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.340136054421769e-05,
+      "loss": 0.4939,
+      "step": 1893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.339778016469746e-05,
+      "loss": 0.5399,
+      "step": 1894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.339419978517723e-05,
+      "loss": 0.5985,
+      "step": 1895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3390619405657004e-05,
+      "loss": 0.4921,
+      "step": 1896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.338703902613677e-05,
+      "loss": 0.5539,
+      "step": 1897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.338345864661654e-05,
+      "loss": 0.5238,
+      "step": 1898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3379878267096316e-05,
+      "loss": 0.5503,
+      "step": 1899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.337629788757608e-05,
+      "loss": 0.5695,
+      "step": 1900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3372717508055856e-05,
+      "loss": 0.5713,
+      "step": 1901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.336913712853562e-05,
+      "loss": 0.4825,
+      "step": 1902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3365556749015396e-05,
+      "loss": 0.5135,
+      "step": 1903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.336197636949517e-05,
+      "loss": 0.5813,
+      "step": 1904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3358395989974935e-05,
+      "loss": 0.5237,
+      "step": 1905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.335481561045471e-05,
+      "loss": 0.5004,
+      "step": 1906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.335123523093448e-05,
+      "loss": 0.5308,
+      "step": 1907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3347654851414255e-05,
+      "loss": 0.5714,
+      "step": 1908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.334407447189403e-05,
+      "loss": 0.5937,
+      "step": 1909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3340494092373794e-05,
+      "loss": 0.5475,
+      "step": 1910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.333691371285357e-05,
+      "loss": 0.5961,
+      "step": 1911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.3333333333333334e-05,
+      "loss": 0.5257,
+      "step": 1912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332975295381311e-05,
+      "loss": 0.6015,
+      "step": 1913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332617257429288e-05,
+      "loss": 0.5174,
+      "step": 1914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.332259219477265e-05,
+      "loss": 0.5269,
+      "step": 1915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331901181525242e-05,
+      "loss": 0.5751,
+      "step": 1916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331543143573219e-05,
+      "loss": 0.5195,
+      "step": 1917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.331185105621196e-05,
+      "loss": 0.6208,
+      "step": 1918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.330827067669173e-05,
+      "loss": 0.6556,
+      "step": 1919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.33046902971715e-05,
+      "loss": 0.5423,
+      "step": 1920
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.330110991765127e-05,
+      "loss": 0.6356,
+      "step": 1921
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3297529538131046e-05,
+      "loss": 0.6353,
+      "step": 1922
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.329394915861081e-05,
+      "loss": 0.6217,
+      "step": 1923
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3290368779090585e-05,
+      "loss": 0.5675,
+      "step": 1924
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.328678839957035e-05,
+      "loss": 0.5722,
+      "step": 1925
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3283208020050125e-05,
+      "loss": 0.5821,
+      "step": 1926
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.32796276405299e-05,
+      "loss": 0.5778,
+      "step": 1927
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3276047261009664e-05,
+      "loss": 0.5082,
+      "step": 1928
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.327246688148944e-05,
+      "loss": 0.5407,
+      "step": 1929
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.326888650196921e-05,
+      "loss": 0.5227,
+      "step": 1930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3265306122448984e-05,
+      "loss": 0.5279,
+      "step": 1931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.326172574292876e-05,
+      "loss": 0.5863,
+      "step": 1932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3258145363408524e-05,
+      "loss": 0.5001,
+      "step": 1933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.32545649838883e-05,
+      "loss": 0.5613,
+      "step": 1934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.325098460436807e-05,
+      "loss": 0.6068,
+      "step": 1935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3247404224847836e-05,
+      "loss": 0.5651,
+      "step": 1936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.324382384532761e-05,
+      "loss": 0.5495,
+      "step": 1937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3240243465807376e-05,
+      "loss": 0.6224,
+      "step": 1938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.323666308628715e-05,
+      "loss": 0.4967,
+      "step": 1939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.323308270676692e-05,
+      "loss": 0.5605,
+      "step": 1940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322950232724669e-05,
+      "loss": 0.5213,
+      "step": 1941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322592194772646e-05,
+      "loss": 0.6543,
+      "step": 1942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.322234156820623e-05,
+      "loss": 0.5497,
+      "step": 1943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3218761188686e-05,
+      "loss": 0.5512,
+      "step": 1944
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3215180809165775e-05,
+      "loss": 0.5526,
+      "step": 1945
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.321160042964554e-05,
+      "loss": 0.5893,
+      "step": 1946
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.3208020050125314e-05,
+      "loss": 0.5441,
+      "step": 1947
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.320443967060509e-05,
+      "loss": 0.564,
+      "step": 1948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3200859291084854e-05,
+      "loss": 0.6024,
+      "step": 1949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319727891156463e-05,
+      "loss": 0.5891,
+      "step": 1950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3193698532044394e-05,
+      "loss": 0.5871,
+      "step": 1951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319011815252417e-05,
+      "loss": 0.5741,
+      "step": 1952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318653777300394e-05,
+      "loss": 0.595,
+      "step": 1953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318295739348371e-05,
+      "loss": 0.6189,
+      "step": 1954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3179377013963486e-05,
+      "loss": 0.4817,
+      "step": 1955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317579663444325e-05,
+      "loss": 0.5876,
+      "step": 1956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3172216254923026e-05,
+      "loss": 0.4917,
+      "step": 1957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31686358754028e-05,
+      "loss": 0.6094,
+      "step": 1958
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3165055495882565e-05,
+      "loss": 0.6331,
+      "step": 1959
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316147511636234e-05,
+      "loss": 0.6518,
+      "step": 1960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3157894736842105e-05,
+      "loss": 0.5646,
+      "step": 1961
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315431435732188e-05,
+      "loss": 0.6589,
+      "step": 1962
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.315073397780165e-05,
+      "loss": 0.6563,
+      "step": 1963
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314715359828142e-05,
+      "loss": 0.6513,
+      "step": 1964
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.314357321876119e-05,
+      "loss": 0.5403,
+      "step": 1965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3139992839240964e-05,
+      "loss": 0.5432,
+      "step": 1966
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.313641245972073e-05,
+      "loss": 0.5361,
+      "step": 1967
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3132832080200504e-05,
+      "loss": 0.612,
+      "step": 1968
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312925170068027e-05,
+      "loss": 0.6483,
+      "step": 1969
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3125671321160043e-05,
+      "loss": 0.5553,
+      "step": 1970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3122090941639817e-05,
+      "loss": 0.614,
+      "step": 1971
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311851056211958e-05,
+      "loss": 0.703,
+      "step": 1972
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3114930182599356e-05,
+      "loss": 0.6046,
+      "step": 1973
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.311134980307913e-05,
+      "loss": 0.4771,
+      "step": 1974
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3107769423558896e-05,
+      "loss": 0.5291,
+      "step": 1975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310418904403867e-05,
+      "loss": 0.6021,
+      "step": 1976
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.310060866451844e-05,
+      "loss": 0.5522,
+      "step": 1977
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3097028284998215e-05,
+      "loss": 0.5809,
+      "step": 1978
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.309344790547798e-05,
+      "loss": 0.5429,
+      "step": 1979
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3089867525957755e-05,
+      "loss": 0.5375,
+      "step": 1980
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.308628714643753e-05,
+      "loss": 0.6042,
+      "step": 1981
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3082706766917295e-05,
+      "loss": 0.4932,
+      "step": 1982
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.307912638739707e-05,
+      "loss": 0.4387,
+      "step": 1983
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.307554600787684e-05,
+      "loss": 0.5163,
+      "step": 1984
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.307196562835661e-05,
+      "loss": 0.6011,
+      "step": 1985
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.306838524883638e-05,
+      "loss": 0.5874,
+      "step": 1986
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.306480486931615e-05,
+      "loss": 0.5966,
+      "step": 1987
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.306122448979592e-05,
+      "loss": 0.6274,
+      "step": 1988
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.305764411027569e-05,
+      "loss": 0.5004,
+      "step": 1989
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.305406373075546e-05,
+      "loss": 0.6205,
+      "step": 1990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.305048335123523e-05,
+      "loss": 0.5466,
+      "step": 1991
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3046902971715006e-05,
+      "loss": 0.5868,
+      "step": 1992
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.304332259219477e-05,
+      "loss": 0.4791,
+      "step": 1993
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3039742212674546e-05,
+      "loss": 0.5104,
+      "step": 1994
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.303616183315431e-05,
+      "loss": 0.5207,
+      "step": 1995
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3032581453634085e-05,
+      "loss": 0.6129,
+      "step": 1996
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.302900107411386e-05,
+      "loss": 0.6093,
+      "step": 1997
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3025420694593625e-05,
+      "loss": 0.5994,
+      "step": 1998
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.30218403150734e-05,
+      "loss": 0.5276,
+      "step": 1999
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301825993555317e-05,
+      "loss": 0.5606,
+      "step": 2000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3014679556032944e-05,
+      "loss": 0.6799,
+      "step": 2001
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301109917651272e-05,
+      "loss": 0.5721,
+      "step": 2002
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.3007518796992484e-05,
+      "loss": 0.5478,
+      "step": 2003
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.300393841747226e-05,
+      "loss": 0.5859,
+      "step": 2004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.3000358037952024e-05,
+      "loss": 0.5824,
+      "step": 2005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29967776584318e-05,
+      "loss": 0.6479,
+      "step": 2006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.299319727891157e-05,
+      "loss": 0.617,
+      "step": 2007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2989616899391336e-05,
+      "loss": 0.4881,
+      "step": 2008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.298603651987111e-05,
+      "loss": 0.5721,
+      "step": 2009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.298245614035088e-05,
+      "loss": 0.586,
+      "step": 2010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.297887576083065e-05,
+      "loss": 0.6066,
+      "step": 2011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.297529538131042e-05,
+      "loss": 0.5018,
+      "step": 2012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.297171500179019e-05,
+      "loss": 0.5615,
+      "step": 2013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.296813462226996e-05,
+      "loss": 0.4561,
+      "step": 2014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2964554242749735e-05,
+      "loss": 0.4996,
+      "step": 2015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29609738632295e-05,
+      "loss": 0.6318,
+      "step": 2016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2957393483709275e-05,
+      "loss": 0.5684,
+      "step": 2017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.295381310418904e-05,
+      "loss": 0.5273,
+      "step": 2018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2950232724668814e-05,
+      "loss": 0.544,
+      "step": 2019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.294665234514859e-05,
+      "loss": 0.5642,
+      "step": 2020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2943071965628354e-05,
+      "loss": 0.5859,
+      "step": 2021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.293949158610813e-05,
+      "loss": 0.5463,
+      "step": 2022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29359112065879e-05,
+      "loss": 0.6638,
+      "step": 2023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2932330827067674e-05,
+      "loss": 0.5449,
+      "step": 2024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292875044754745e-05,
+      "loss": 0.6859,
+      "step": 2025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.292517006802721e-05,
+      "loss": 0.6469,
+      "step": 2026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2921589688506986e-05,
+      "loss": 0.6123,
+      "step": 2027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.291800930898676e-05,
+      "loss": 0.5294,
+      "step": 2028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2914428929466526e-05,
+      "loss": 0.5566,
+      "step": 2029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.29108485499463e-05,
+      "loss": 0.5503,
+      "step": 2030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.2907268170426066e-05,
+      "loss": 0.57,
+      "step": 2031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.290368779090584e-05,
+      "loss": 0.5174,
+      "step": 2032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.290010741138561e-05,
+      "loss": 0.6053,
+      "step": 2033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.289652703186538e-05,
+      "loss": 0.591,
+      "step": 2034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.289294665234515e-05,
+      "loss": 0.5899,
+      "step": 2035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.288936627282492e-05,
+      "loss": 0.5635,
+      "step": 2036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.288578589330469e-05,
+      "loss": 0.5854,
+      "step": 2037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2882205513784464e-05,
+      "loss": 0.5211,
+      "step": 2038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.287862513426423e-05,
+      "loss": 0.5171,
+      "step": 2039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2875044754744004e-05,
+      "loss": 0.5874,
+      "step": 2040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.287146437522378e-05,
+      "loss": 0.5348,
+      "step": 2041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2867883995703544e-05,
+      "loss": 0.5766,
+      "step": 2042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.286430361618332e-05,
+      "loss": 0.4977,
+      "step": 2043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.286072323666308e-05,
+      "loss": 0.5898,
+      "step": 2044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.5536,
+      "step": 2045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.285356247762263e-05,
+      "loss": 0.4984,
+      "step": 2046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.28499820981024e-05,
+      "loss": 0.5804,
+      "step": 2047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2846401718582176e-05,
+      "loss": 0.5773,
+      "step": 2048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.284282133906194e-05,
+      "loss": 0.5862,
+      "step": 2049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2839240959541715e-05,
+      "loss": 0.4808,
+      "step": 2050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.283566058002149e-05,
+      "loss": 0.6103,
+      "step": 2051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2832080200501255e-05,
+      "loss": 0.5211,
+      "step": 2052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.282849982098103e-05,
+      "loss": 0.4812,
+      "step": 2053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2824919441460795e-05,
+      "loss": 0.4791,
+      "step": 2054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.282133906194057e-05,
+      "loss": 0.5875,
+      "step": 2055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281775868242034e-05,
+      "loss": 0.5793,
+      "step": 2056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281417830290011e-05,
+      "loss": 0.5838,
+      "step": 2057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.281059792337988e-05,
+      "loss": 0.5305,
+      "step": 2058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.2807017543859654e-05,
+      "loss": 0.5481,
+      "step": 2059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.280343716433942e-05,
+      "loss": 0.6856,
+      "step": 2060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2799856784819193e-05,
+      "loss": 0.6025,
+      "step": 2061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.279627640529896e-05,
+      "loss": 0.4721,
+      "step": 2062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.279269602577873e-05,
+      "loss": 0.5398,
+      "step": 2063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2789115646258506e-05,
+      "loss": 0.5973,
+      "step": 2064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.278553526673827e-05,
+      "loss": 0.465,
+      "step": 2065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2781954887218046e-05,
+      "loss": 0.5377,
+      "step": 2066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.277837450769782e-05,
+      "loss": 0.4835,
+      "step": 2067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2774794128177585e-05,
+      "loss": 0.5671,
+      "step": 2068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.277121374865736e-05,
+      "loss": 0.6365,
+      "step": 2069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2767633369137125e-05,
+      "loss": 0.4857,
+      "step": 2070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2764052989616905e-05,
+      "loss": 0.5312,
+      "step": 2071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.276047261009667e-05,
+      "loss": 0.5724,
+      "step": 2072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2756892230576445e-05,
+      "loss": 0.5799,
+      "step": 2073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.275331185105622e-05,
+      "loss": 0.4915,
+      "step": 2074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2749731471535984e-05,
+      "loss": 0.5586,
+      "step": 2075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.274615109201576e-05,
+      "loss": 0.5392,
+      "step": 2076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.274257071249553e-05,
+      "loss": 0.6364,
+      "step": 2077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.27389903329753e-05,
+      "loss": 0.5663,
+      "step": 2078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.273540995345507e-05,
+      "loss": 0.509,
+      "step": 2079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2731829573934837e-05,
+      "loss": 0.6537,
+      "step": 2080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.272824919441461e-05,
+      "loss": 0.5423,
+      "step": 2081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.272466881489438e-05,
+      "loss": 0.6125,
+      "step": 2082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.272108843537415e-05,
+      "loss": 0.5696,
+      "step": 2083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.271750805585392e-05,
+      "loss": 0.4752,
+      "step": 2084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2713927676333696e-05,
+      "loss": 0.4748,
+      "step": 2085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.271034729681346e-05,
+      "loss": 0.4299,
+      "step": 2086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2706766917293235e-05,
+      "loss": 0.5412,
+      "step": 2087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.2703186537773e-05,
+      "loss": 0.5774,
+      "step": 2088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2699606158252775e-05,
+      "loss": 0.6302,
+      "step": 2089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.269602577873255e-05,
+      "loss": 0.533,
+      "step": 2090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2692445399212315e-05,
+      "loss": 0.5917,
+      "step": 2091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.268886501969209e-05,
+      "loss": 0.5974,
+      "step": 2092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.268528464017186e-05,
+      "loss": 0.5,
+      "step": 2093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2681704260651634e-05,
+      "loss": 0.5873,
+      "step": 2094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.267812388113141e-05,
+      "loss": 0.5619,
+      "step": 2095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2674543501611174e-05,
+      "loss": 0.5386,
+      "step": 2096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.267096312209095e-05,
+      "loss": 0.5437,
+      "step": 2097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.266738274257071e-05,
+      "loss": 0.4374,
+      "step": 2098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2663802363050486e-05,
+      "loss": 0.5232,
+      "step": 2099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.266022198353026e-05,
+      "loss": 0.5258,
+      "step": 2100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2656641604010026e-05,
+      "loss": 0.57,
+      "step": 2101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.26530612244898e-05,
+      "loss": 0.6091,
+      "step": 2102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264948084496957e-05,
+      "loss": 0.5511,
+      "step": 2103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264590046544934e-05,
+      "loss": 0.5616,
+      "step": 2104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.264232008592911e-05,
+      "loss": 0.5765,
+      "step": 2105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.263873970640888e-05,
+      "loss": 0.5844,
+      "step": 2106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.263515932688865e-05,
+      "loss": 0.5729,
+      "step": 2107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2631578947368425e-05,
+      "loss": 0.5336,
+      "step": 2108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262799856784819e-05,
+      "loss": 0.5787,
+      "step": 2109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2624418188327964e-05,
+      "loss": 0.5108,
+      "step": 2110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.262083780880774e-05,
+      "loss": 0.5289,
+      "step": 2111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2617257429287504e-05,
+      "loss": 0.5858,
+      "step": 2112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.261367704976728e-05,
+      "loss": 0.5633,
+      "step": 2113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.2610096670247044e-05,
+      "loss": 0.4087,
+      "step": 2114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260651629072682e-05,
+      "loss": 0.5981,
+      "step": 2115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.260293591120659e-05,
+      "loss": 0.4648,
+      "step": 2116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.259935553168636e-05,
+      "loss": 0.587,
+      "step": 2117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2595775152166136e-05,
+      "loss": 0.5355,
+      "step": 2118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25921947726459e-05,
+      "loss": 0.6453,
+      "step": 2119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2588614393125676e-05,
+      "loss": 0.5995,
+      "step": 2120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.258503401360545e-05,
+      "loss": 0.6684,
+      "step": 2121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2581453634085216e-05,
+      "loss": 0.4764,
+      "step": 2122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.257787325456499e-05,
+      "loss": 0.5348,
+      "step": 2123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2574292875044755e-05,
+      "loss": 0.5158,
+      "step": 2124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.257071249552453e-05,
+      "loss": 0.4682,
+      "step": 2125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.25671321160043e-05,
+      "loss": 0.6355,
+      "step": 2126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.256355173648407e-05,
+      "loss": 0.4701,
+      "step": 2127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.255997135696384e-05,
+      "loss": 0.6371,
+      "step": 2128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.255639097744361e-05,
+      "loss": 0.5596,
+      "step": 2129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.255281059792338e-05,
+      "loss": 0.6239,
+      "step": 2130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2549230218403154e-05,
+      "loss": 0.559,
+      "step": 2131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.254564983888292e-05,
+      "loss": 0.5023,
+      "step": 2132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2542069459362694e-05,
+      "loss": 0.5952,
+      "step": 2133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.253848907984247e-05,
+      "loss": 0.5605,
+      "step": 2134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.253490870032223e-05,
+      "loss": 0.6099,
+      "step": 2135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2531328320802006e-05,
+      "loss": 0.5917,
+      "step": 2136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252774794128177e-05,
+      "loss": 0.6167,
+      "step": 2137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2524167561761546e-05,
+      "loss": 0.623,
+      "step": 2138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.252058718224132e-05,
+      "loss": 0.5797,
+      "step": 2139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2517006802721085e-05,
+      "loss": 0.5233,
+      "step": 2140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2513426423200865e-05,
+      "loss": 0.5531,
+      "step": 2141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250984604368063e-05,
+      "loss": 0.5289,
+      "step": 2142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.2506265664160405e-05,
+      "loss": 0.5163,
+      "step": 2143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.250268528464018e-05,
+      "loss": 0.5263,
+      "step": 2144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2499104905119945e-05,
+      "loss": 0.5504,
+      "step": 2145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.249552452559972e-05,
+      "loss": 0.4836,
+      "step": 2146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2491944146079484e-05,
+      "loss": 0.4972,
+      "step": 2147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.248836376655926e-05,
+      "loss": 0.4438,
+      "step": 2148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.248478338703903e-05,
+      "loss": 0.5397,
+      "step": 2149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.24812030075188e-05,
+      "loss": 0.5417,
+      "step": 2150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.247762262799857e-05,
+      "loss": 0.4724,
+      "step": 2151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2474042248478343e-05,
+      "loss": 0.5661,
+      "step": 2152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.247046186895811e-05,
+      "loss": 0.5167,
+      "step": 2153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.246688148943788e-05,
+      "loss": 0.5365,
+      "step": 2154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.246330110991765e-05,
+      "loss": 0.6677,
+      "step": 2155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.245972073039742e-05,
+      "loss": 0.5533,
+      "step": 2156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2456140350877196e-05,
+      "loss": 0.5394,
+      "step": 2157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.245255997135696e-05,
+      "loss": 0.5488,
+      "step": 2158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2448979591836735e-05,
+      "loss": 0.4618,
+      "step": 2159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.244539921231651e-05,
+      "loss": 0.5091,
+      "step": 2160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2441818832796275e-05,
+      "loss": 0.5101,
+      "step": 2161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.243823845327605e-05,
+      "loss": 0.6284,
+      "step": 2162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2434658073755815e-05,
+      "loss": 0.5554,
+      "step": 2163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2431077694235595e-05,
+      "loss": 0.7178,
+      "step": 2164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.242749731471536e-05,
+      "loss": 0.5856,
+      "step": 2165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2423916935195134e-05,
+      "loss": 0.5083,
+      "step": 2166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.242033655567491e-05,
+      "loss": 0.5212,
+      "step": 2167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2416756176154674e-05,
+      "loss": 0.4753,
+      "step": 2168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.241317579663445e-05,
+      "loss": 0.6104,
+      "step": 2169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240959541711422e-05,
+      "loss": 0.6465,
+      "step": 2170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.2406015037593987e-05,
+      "loss": 0.5176,
+      "step": 2171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.240243465807376e-05,
+      "loss": 0.5056,
+      "step": 2172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2398854278553526e-05,
+      "loss": 0.4473,
+      "step": 2173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.23952738990333e-05,
+      "loss": 0.5411,
+      "step": 2174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.239169351951307e-05,
+      "loss": 0.4795,
+      "step": 2175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.238811313999284e-05,
+      "loss": 0.4843,
+      "step": 2176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.238453276047261e-05,
+      "loss": 0.5097,
+      "step": 2177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2380952380952385e-05,
+      "loss": 0.559,
+      "step": 2178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.237737200143215e-05,
+      "loss": 0.5143,
+      "step": 2179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2373791621911925e-05,
+      "loss": 0.6142,
+      "step": 2180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.237021124239169e-05,
+      "loss": 0.4832,
+      "step": 2181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2366630862871465e-05,
+      "loss": 0.5247,
+      "step": 2182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.236305048335124e-05,
+      "loss": 0.7269,
+      "step": 2183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2359470103831004e-05,
+      "loss": 0.4907,
+      "step": 2184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.235588972431078e-05,
+      "loss": 0.4738,
+      "step": 2185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.235230934479055e-05,
+      "loss": 0.5321,
+      "step": 2186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2348728965270324e-05,
+      "loss": 0.5884,
+      "step": 2187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.23451485857501e-05,
+      "loss": 0.5451,
+      "step": 2188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.234156820622986e-05,
+      "loss": 0.6522,
+      "step": 2189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2337987826709636e-05,
+      "loss": 0.5042,
+      "step": 2190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.23344074471894e-05,
+      "loss": 0.5528,
+      "step": 2191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2330827067669176e-05,
+      "loss": 0.5181,
+      "step": 2192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.232724668814895e-05,
+      "loss": 0.5564,
+      "step": 2193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.2323666308628716e-05,
+      "loss": 0.491,
+      "step": 2194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.232008592910849e-05,
+      "loss": 0.5331,
+      "step": 2195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231650554958826e-05,
+      "loss": 0.5436,
+      "step": 2196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.231292517006803e-05,
+      "loss": 0.6572,
+      "step": 2197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.23093447905478e-05,
+      "loss": 0.4474,
+      "step": 2198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.230576441102757e-05,
+      "loss": 0.655,
+      "step": 2199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.230218403150734e-05,
+      "loss": 0.6144,
+      "step": 2200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2298603651987114e-05,
+      "loss": 0.6665,
+      "step": 2201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.229502327246688e-05,
+      "loss": 0.5865,
+      "step": 2202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2291442892946654e-05,
+      "loss": 0.5429,
+      "step": 2203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228786251342643e-05,
+      "loss": 0.5418,
+      "step": 2204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2284282133906194e-05,
+      "loss": 0.6338,
+      "step": 2205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228070175438597e-05,
+      "loss": 0.6404,
+      "step": 2206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.227712137486573e-05,
+      "loss": 0.6367,
+      "step": 2207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2273540995345506e-05,
+      "loss": 0.496,
+      "step": 2208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.226996061582528e-05,
+      "loss": 0.5408,
+      "step": 2209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2266380236305046e-05,
+      "loss": 0.4896,
+      "step": 2210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2262799856784826e-05,
+      "loss": 0.6205,
+      "step": 2211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.225921947726459e-05,
+      "loss": 0.4627,
+      "step": 2212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2255639097744366e-05,
+      "loss": 0.5881,
+      "step": 2213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.225205871822414e-05,
+      "loss": 0.584,
+      "step": 2214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2248478338703905e-05,
+      "loss": 0.5998,
+      "step": 2215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.224489795918368e-05,
+      "loss": 0.515,
+      "step": 2216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2241317579663445e-05,
+      "loss": 0.5199,
+      "step": 2217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223773720014322e-05,
+      "loss": 0.5396,
+      "step": 2218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223415682062299e-05,
+      "loss": 0.4506,
+      "step": 2219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223057644110276e-05,
+      "loss": 0.5077,
+      "step": 2220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.222699606158253e-05,
+      "loss": 0.558,
+      "step": 2221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2223415682062304e-05,
+      "loss": 0.5803,
+      "step": 2222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221983530254207e-05,
+      "loss": 0.5788,
+      "step": 2223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2216254923021844e-05,
+      "loss": 0.5725,
+      "step": 2224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.221267454350161e-05,
+      "loss": 0.6469,
+      "step": 2225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220909416398138e-05,
+      "loss": 0.47,
+      "step": 2226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2205513784461156e-05,
+      "loss": 0.5209,
+      "step": 2227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220193340494092e-05,
+      "loss": 0.5511,
+      "step": 2228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2198353025420696e-05,
+      "loss": 0.6221,
+      "step": 2229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.219477264590046e-05,
+      "loss": 0.5558,
+      "step": 2230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2191192266380235e-05,
+      "loss": 0.518,
+      "step": 2231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.218761188686001e-05,
+      "loss": 0.5445,
+      "step": 2232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2184031507339775e-05,
+      "loss": 0.4157,
+      "step": 2233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2180451127819555e-05,
+      "loss": 0.546,
+      "step": 2234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.217687074829932e-05,
+      "loss": 0.4725,
+      "step": 2235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2173290368779095e-05,
+      "loss": 0.5547,
+      "step": 2236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.216970998925887e-05,
+      "loss": 0.591,
+      "step": 2237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2166129609738634e-05,
+      "loss": 0.5509,
+      "step": 2238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.216254923021841e-05,
+      "loss": 0.5257,
+      "step": 2239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.215896885069818e-05,
+      "loss": 0.5227,
+      "step": 2240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.215538847117795e-05,
+      "loss": 0.4425,
+      "step": 2241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.215180809165772e-05,
+      "loss": 0.544,
+      "step": 2242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.214822771213749e-05,
+      "loss": 0.4618,
+      "step": 2243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.214464733261726e-05,
+      "loss": 0.6331,
+      "step": 2244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.214106695309703e-05,
+      "loss": 0.5991,
+      "step": 2245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.21374865735768e-05,
+      "loss": 0.5319,
+      "step": 2246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.213390619405657e-05,
+      "loss": 0.4892,
+      "step": 2247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.213032581453634e-05,
+      "loss": 0.5594,
+      "step": 2248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.212674543501611e-05,
+      "loss": 0.5337,
+      "step": 2249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2123165055495885e-05,
+      "loss": 0.5779,
+      "step": 2250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.211958467597565e-05,
+      "loss": 0.5536,
+      "step": 2251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2116004296455425e-05,
+      "loss": 0.5581,
+      "step": 2252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.21124239169352e-05,
+      "loss": 0.5361,
+      "step": 2253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2108843537414965e-05,
+      "loss": 0.4488,
+      "step": 2254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.5706,
+      "step": 2255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.2101682778374504e-05,
+      "loss": 0.5284,
+      "step": 2256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2098102398854284e-05,
+      "loss": 0.5194,
+      "step": 2257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.209452201933405e-05,
+      "loss": 0.7088,
+      "step": 2258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2090941639813824e-05,
+      "loss": 0.5639,
+      "step": 2259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.20873612602936e-05,
+      "loss": 0.544,
+      "step": 2260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.208378088077336e-05,
+      "loss": 0.5429,
+      "step": 2261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2080200501253137e-05,
+      "loss": 0.5084,
+      "step": 2262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.207662012173291e-05,
+      "loss": 0.6323,
+      "step": 2263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2073039742212676e-05,
+      "loss": 0.5383,
+      "step": 2264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.206945936269245e-05,
+      "loss": 0.647,
+      "step": 2265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2065878983172216e-05,
+      "loss": 0.5436,
+      "step": 2266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.206229860365199e-05,
+      "loss": 0.533,
+      "step": 2267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.205871822413176e-05,
+      "loss": 0.5984,
+      "step": 2268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.205513784461153e-05,
+      "loss": 0.6679,
+      "step": 2269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.20515574650913e-05,
+      "loss": 0.5189,
+      "step": 2270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2047977085571075e-05,
+      "loss": 0.5295,
+      "step": 2271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.204439670605084e-05,
+      "loss": 0.4657,
+      "step": 2272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2040816326530615e-05,
+      "loss": 0.4519,
+      "step": 2273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.203723594701038e-05,
+      "loss": 0.5492,
+      "step": 2274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2033655567490154e-05,
+      "loss": 0.6169,
+      "step": 2275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.203007518796993e-05,
+      "loss": 0.5324,
+      "step": 2276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2026494808449694e-05,
+      "loss": 0.6099,
+      "step": 2277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.202291442892947e-05,
+      "loss": 0.591,
+      "step": 2278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.201933404940924e-05,
+      "loss": 0.5055,
+      "step": 2279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2015753669889006e-05,
+      "loss": 0.5024,
+      "step": 2280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2012173290368786e-05,
+      "loss": 0.5452,
+      "step": 2281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200859291084855e-05,
+      "loss": 0.6194,
+      "step": 2282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2005012531328326e-05,
+      "loss": 0.5482,
+      "step": 2283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.200143215180809e-05,
+      "loss": 0.5596,
+      "step": 2284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1997851772287866e-05,
+      "loss": 0.5726,
+      "step": 2285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.199427139276764e-05,
+      "loss": 0.4555,
+      "step": 2286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1990691013247405e-05,
+      "loss": 0.5135,
+      "step": 2287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.198711063372718e-05,
+      "loss": 0.4957,
+      "step": 2288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.198353025420695e-05,
+      "loss": 0.5383,
+      "step": 2289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.197994987468672e-05,
+      "loss": 0.5974,
+      "step": 2290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.197636949516649e-05,
+      "loss": 0.583,
+      "step": 2291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.197278911564626e-05,
+      "loss": 0.4856,
+      "step": 2292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.196920873612603e-05,
+      "loss": 0.485,
+      "step": 2293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1965628356605804e-05,
+      "loss": 0.4997,
+      "step": 2294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.196204797708557e-05,
+      "loss": 0.5476,
+      "step": 2295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1958467597565344e-05,
+      "loss": 0.5287,
+      "step": 2296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.195488721804512e-05,
+      "loss": 0.4709,
+      "step": 2297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.195130683852488e-05,
+      "loss": 0.5736,
+      "step": 2298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1947726459004656e-05,
+      "loss": 0.4615,
+      "step": 2299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.194414607948442e-05,
+      "loss": 0.5306,
+      "step": 2300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1940565699964196e-05,
+      "loss": 0.515,
+      "step": 2301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.193698532044397e-05,
+      "loss": 0.4308,
+      "step": 2302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1933404940923736e-05,
+      "loss": 0.554,
+      "step": 2303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1929824561403516e-05,
+      "loss": 0.571,
+      "step": 2304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.192624418188328e-05,
+      "loss": 0.4434,
+      "step": 2305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1922663802363055e-05,
+      "loss": 0.5615,
+      "step": 2306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.191908342284283e-05,
+      "loss": 0.6195,
+      "step": 2307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1915503043322595e-05,
+      "loss": 0.5385,
+      "step": 2308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.191192266380237e-05,
+      "loss": 0.5275,
+      "step": 2309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.1908342284282134e-05,
+      "loss": 0.4919,
+      "step": 2310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190476190476191e-05,
+      "loss": 0.5725,
+      "step": 2311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.190118152524168e-05,
+      "loss": 0.6384,
+      "step": 2312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.189760114572145e-05,
+      "loss": 0.4696,
+      "step": 2313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.189402076620122e-05,
+      "loss": 0.6236,
+      "step": 2314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1890440386680994e-05,
+      "loss": 0.5283,
+      "step": 2315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.188686000716076e-05,
+      "loss": 0.4128,
+      "step": 2316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.188327962764053e-05,
+      "loss": 0.5784,
+      "step": 2317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.18796992481203e-05,
+      "loss": 0.4659,
+      "step": 2318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.187611886860007e-05,
+      "loss": 0.4935,
+      "step": 2319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1872538489079846e-05,
+      "loss": 0.6415,
+      "step": 2320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.186895810955961e-05,
+      "loss": 0.521,
+      "step": 2321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1865377730039386e-05,
+      "loss": 0.4499,
+      "step": 2322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.186179735051915e-05,
+      "loss": 0.5992,
+      "step": 2323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1858216970998925e-05,
+      "loss": 0.65,
+      "step": 2324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.18546365914787e-05,
+      "loss": 0.504,
+      "step": 2325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1851056211958465e-05,
+      "loss": 0.5237,
+      "step": 2326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1847475832438245e-05,
+      "loss": 0.5104,
+      "step": 2327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.184389545291801e-05,
+      "loss": 0.5229,
+      "step": 2328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1840315073397784e-05,
+      "loss": 0.5868,
+      "step": 2329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.183673469387756e-05,
+      "loss": 0.5524,
+      "step": 2330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1833154314357324e-05,
+      "loss": 0.5459,
+      "step": 2331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.18295739348371e-05,
+      "loss": 0.505,
+      "step": 2332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.182599355531687e-05,
+      "loss": 0.608,
+      "step": 2333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.182241317579664e-05,
+      "loss": 0.4771,
+      "step": 2334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181883279627641e-05,
+      "loss": 0.468,
+      "step": 2335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.1815252416756176e-05,
+      "loss": 0.606,
+      "step": 2336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.181167203723595e-05,
+      "loss": 0.537,
+      "step": 2337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180809165771572e-05,
+      "loss": 0.5457,
+      "step": 2338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180451127819549e-05,
+      "loss": 0.564,
+      "step": 2339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.180093089867526e-05,
+      "loss": 0.5815,
+      "step": 2340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.179735051915503e-05,
+      "loss": 0.5523,
+      "step": 2341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.17937701396348e-05,
+      "loss": 0.5493,
+      "step": 2342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1790189760114575e-05,
+      "loss": 0.4832,
+      "step": 2343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.178660938059434e-05,
+      "loss": 0.5607,
+      "step": 2344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1783029001074115e-05,
+      "loss": 0.5883,
+      "step": 2345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.177944862155389e-05,
+      "loss": 0.7145,
+      "step": 2346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1775868242033654e-05,
+      "loss": 0.5397,
+      "step": 2347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.177228786251343e-05,
+      "loss": 0.5006,
+      "step": 2348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1768707482993194e-05,
+      "loss": 0.5218,
+      "step": 2349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.176512710347297e-05,
+      "loss": 0.6663,
+      "step": 2350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.176154672395275e-05,
+      "loss": 0.5234,
+      "step": 2351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.175796634443251e-05,
+      "loss": 0.6004,
+      "step": 2352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1754385964912287e-05,
+      "loss": 0.5742,
+      "step": 2353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.175080558539205e-05,
+      "loss": 0.6768,
+      "step": 2354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1747225205871826e-05,
+      "loss": 0.5252,
+      "step": 2355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.17436448263516e-05,
+      "loss": 0.5892,
+      "step": 2356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1740064446831366e-05,
+      "loss": 0.5906,
+      "step": 2357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.173648406731114e-05,
+      "loss": 0.5081,
+      "step": 2358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1732903687790905e-05,
+      "loss": 0.5539,
+      "step": 2359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.172932330827068e-05,
+      "loss": 0.5165,
+      "step": 2360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.172574292875045e-05,
+      "loss": 0.512,
+      "step": 2361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.172216254923022e-05,
+      "loss": 0.5394,
+      "step": 2362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171858216970999e-05,
+      "loss": 0.6091,
+      "step": 2363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1715001790189765e-05,
+      "loss": 0.6027,
+      "step": 2364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.171142141066953e-05,
+      "loss": 0.5369,
+      "step": 2365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1707841031149304e-05,
+      "loss": 0.5631,
+      "step": 2366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.170426065162907e-05,
+      "loss": 0.5243,
+      "step": 2367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.1700680272108844e-05,
+      "loss": 0.5887,
+      "step": 2368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.169709989258862e-05,
+      "loss": 0.5556,
+      "step": 2369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.169351951306838e-05,
+      "loss": 0.5364,
+      "step": 2370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1689939133548156e-05,
+      "loss": 0.6018,
+      "step": 2371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.168635875402793e-05,
+      "loss": 0.5348,
+      "step": 2372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1682778374507696e-05,
+      "loss": 0.5186,
+      "step": 2373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1679197994987476e-05,
+      "loss": 0.4728,
+      "step": 2374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.167561761546724e-05,
+      "loss": 0.4528,
+      "step": 2375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1672037235947016e-05,
+      "loss": 0.5571,
+      "step": 2376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.166845685642678e-05,
+      "loss": 0.4043,
+      "step": 2377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1664876476906555e-05,
+      "loss": 0.4134,
+      "step": 2378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.166129609738633e-05,
+      "loss": 0.4629,
+      "step": 2379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1657715717866095e-05,
+      "loss": 0.51,
+      "step": 2380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.165413533834587e-05,
+      "loss": 0.5561,
+      "step": 2381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.165055495882564e-05,
+      "loss": 0.588,
+      "step": 2382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.164697457930541e-05,
+      "loss": 0.4873,
+      "step": 2383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.164339419978518e-05,
+      "loss": 0.6017,
+      "step": 2384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.163981382026495e-05,
+      "loss": 0.594,
+      "step": 2385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.163623344074472e-05,
+      "loss": 0.5883,
+      "step": 2386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1632653061224494e-05,
+      "loss": 0.4341,
+      "step": 2387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.162907268170426e-05,
+      "loss": 0.5117,
+      "step": 2388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.162549230218403e-05,
+      "loss": 0.5621,
+      "step": 2389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1621911922663806e-05,
+      "loss": 0.5342,
+      "step": 2390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161833154314357e-05,
+      "loss": 0.5463,
+      "step": 2391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1614751163623346e-05,
+      "loss": 0.6766,
+      "step": 2392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.161117078410311e-05,
+      "loss": 0.7212,
+      "step": 2393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1607590404582886e-05,
+      "loss": 0.5314,
+      "step": 2394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.160401002506266e-05,
+      "loss": 0.5283,
+      "step": 2395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.1600429645542425e-05,
+      "loss": 0.4255,
+      "step": 2396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1596849266022205e-05,
+      "loss": 0.6521,
+      "step": 2397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.159326888650197e-05,
+      "loss": 0.5211,
+      "step": 2398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1589688506981745e-05,
+      "loss": 0.4782,
+      "step": 2399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.158610812746152e-05,
+      "loss": 0.5121,
+      "step": 2400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1582527747941284e-05,
+      "loss": 0.5334,
+      "step": 2401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.157894736842106e-05,
+      "loss": 0.5247,
+      "step": 2402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1575366988900824e-05,
+      "loss": 0.5578,
+      "step": 2403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.15717866093806e-05,
+      "loss": 0.6368,
+      "step": 2404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.156820622986037e-05,
+      "loss": 0.4942,
+      "step": 2405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.156462585034014e-05,
+      "loss": 0.4956,
+      "step": 2406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.156104547081991e-05,
+      "loss": 0.6905,
+      "step": 2407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.155746509129968e-05,
+      "loss": 0.5103,
+      "step": 2408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.155388471177945e-05,
+      "loss": 0.5185,
+      "step": 2409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.155030433225922e-05,
+      "loss": 0.5853,
+      "step": 2410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.154672395273899e-05,
+      "loss": 0.5051,
+      "step": 2411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.154314357321876e-05,
+      "loss": 0.5335,
+      "step": 2412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1539563193698536e-05,
+      "loss": 0.6019,
+      "step": 2413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.15359828141783e-05,
+      "loss": 0.6518,
+      "step": 2414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1532402434658075e-05,
+      "loss": 0.4623,
+      "step": 2415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.152882205513784e-05,
+      "loss": 0.6096,
+      "step": 2416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1525241675617615e-05,
+      "loss": 0.5918,
+      "step": 2417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.152166129609739e-05,
+      "loss": 0.5264,
+      "step": 2418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1518080916577154e-05,
+      "loss": 0.5305,
+      "step": 2419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.151450053705693e-05,
+      "loss": 0.6035,
+      "step": 2420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.15109201575367e-05,
+      "loss": 0.4898,
+      "step": 2421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1507339778016474e-05,
+      "loss": 0.6575,
+      "step": 2422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150375939849625e-05,
+      "loss": 0.4837,
+      "step": 2423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1500179018976013e-05,
+      "loss": 0.6948,
+      "step": 2424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.149659863945579e-05,
+      "loss": 0.4972,
+      "step": 2425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.149301825993556e-05,
+      "loss": 0.6387,
+      "step": 2426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1489437880415326e-05,
+      "loss": 0.5223,
+      "step": 2427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14858575008951e-05,
+      "loss": 0.5148,
+      "step": 2428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1482277121374866e-05,
+      "loss": 0.6471,
+      "step": 2429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.147869674185464e-05,
+      "loss": 0.5352,
+      "step": 2430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.147511636233441e-05,
+      "loss": 0.5651,
+      "step": 2431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.147153598281418e-05,
+      "loss": 0.52,
+      "step": 2432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.146795560329395e-05,
+      "loss": 0.4967,
+      "step": 2433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.146437522377372e-05,
+      "loss": 0.5273,
+      "step": 2434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.146079484425349e-05,
+      "loss": 0.4397,
+      "step": 2435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1457214464733265e-05,
+      "loss": 0.5105,
+      "step": 2436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.145363408521303e-05,
+      "loss": 0.5218,
+      "step": 2437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1450053705692804e-05,
+      "loss": 0.4524,
+      "step": 2438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.144647332617258e-05,
+      "loss": 0.4178,
+      "step": 2439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1442892946652344e-05,
+      "loss": 0.5054,
+      "step": 2440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.143931256713212e-05,
+      "loss": 0.5648,
+      "step": 2441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1435732187611883e-05,
+      "loss": 0.4437,
+      "step": 2442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1432151808091657e-05,
+      "loss": 0.5713,
+      "step": 2443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 0.474,
+      "step": 2444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14249910490512e-05,
+      "loss": 0.5267,
+      "step": 2445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1421410669530976e-05,
+      "loss": 0.5331,
+      "step": 2446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.141783029001074e-05,
+      "loss": 0.5616,
+      "step": 2447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1414249910490516e-05,
+      "loss": 0.493,
+      "step": 2448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.141066953097029e-05,
+      "loss": 0.8337,
+      "step": 2449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1407089151450055e-05,
+      "loss": 0.4678,
+      "step": 2450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140350877192983e-05,
+      "loss": 0.5144,
+      "step": 2451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1399928392409595e-05,
+      "loss": 0.5041,
+      "step": 2452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.139634801288937e-05,
+      "loss": 0.5506,
+      "step": 2453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.139276763336914e-05,
+      "loss": 0.6357,
+      "step": 2454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.138918725384891e-05,
+      "loss": 0.6384,
+      "step": 2455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.138560687432868e-05,
+      "loss": 0.5246,
+      "step": 2456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1382026494808454e-05,
+      "loss": 0.5569,
+      "step": 2457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.137844611528822e-05,
+      "loss": 0.4647,
+      "step": 2458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1374865735767994e-05,
+      "loss": 0.5038,
+      "step": 2459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.137128535624776e-05,
+      "loss": 0.6412,
+      "step": 2460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.136770497672753e-05,
+      "loss": 0.5624,
+      "step": 2461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1364124597207307e-05,
+      "loss": 0.5715,
+      "step": 2462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.136054421768707e-05,
+      "loss": 0.4044,
+      "step": 2463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1356963838166846e-05,
+      "loss": 0.5386,
+      "step": 2464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.135338345864662e-05,
+      "loss": 0.4954,
+      "step": 2465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1349803079126386e-05,
+      "loss": 0.5062,
+      "step": 2466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.134622269960616e-05,
+      "loss": 0.4768,
+      "step": 2467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.134264232008593e-05,
+      "loss": 0.519,
+      "step": 2468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1339061940565705e-05,
+      "loss": 0.5731,
+      "step": 2469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.133548156104547e-05,
+      "loss": 0.5175,
+      "step": 2470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1331901181525245e-05,
+      "loss": 0.5512,
+      "step": 2471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.132832080200502e-05,
+      "loss": 0.465,
+      "step": 2472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1324740422484784e-05,
+      "loss": 0.6735,
+      "step": 2473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.132116004296456e-05,
+      "loss": 0.5998,
+      "step": 2474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.131757966344433e-05,
+      "loss": 0.5154,
+      "step": 2475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.13139992839241e-05,
+      "loss": 0.5417,
+      "step": 2476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.131041890440387e-05,
+      "loss": 0.5574,
+      "step": 2477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130683852488364e-05,
+      "loss": 0.5108,
+      "step": 2478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130325814536341e-05,
+      "loss": 0.6596,
+      "step": 2479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.129967776584318e-05,
+      "loss": 0.5379,
+      "step": 2480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.129609738632295e-05,
+      "loss": 0.5656,
+      "step": 2481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.129251700680272e-05,
+      "loss": 0.4785,
+      "step": 2482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1288936627282496e-05,
+      "loss": 0.5957,
+      "step": 2483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.128535624776226e-05,
+      "loss": 0.6194,
+      "step": 2484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1281775868242036e-05,
+      "loss": 0.5529,
+      "step": 2485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.12781954887218e-05,
+      "loss": 0.4594,
+      "step": 2486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1274615109201575e-05,
+      "loss": 0.4822,
+      "step": 2487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.127103472968135e-05,
+      "loss": 0.605,
+      "step": 2488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1267454350161115e-05,
+      "loss": 0.5225,
+      "step": 2489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.126387397064089e-05,
+      "loss": 0.5966,
+      "step": 2490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.126029359112066e-05,
+      "loss": 0.5817,
+      "step": 2491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1256713211600434e-05,
+      "loss": 0.5759,
+      "step": 2492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.125313283208021e-05,
+      "loss": 0.5097,
+      "step": 2493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1249552452559974e-05,
+      "loss": 0.5724,
+      "step": 2494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.124597207303975e-05,
+      "loss": 0.4335,
+      "step": 2495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1242391693519514e-05,
+      "loss": 0.543,
+      "step": 2496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.123881131399929e-05,
+      "loss": 0.5588,
+      "step": 2497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.123523093447906e-05,
+      "loss": 0.6751,
+      "step": 2498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1231650554958826e-05,
+      "loss": 0.5191,
+      "step": 2499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.12280701754386e-05,
+      "loss": 0.5407,
+      "step": 2500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.122448979591837e-05,
+      "loss": 0.6184,
+      "step": 2501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.122090941639814e-05,
+      "loss": 0.5132,
+      "step": 2502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121732903687791e-05,
+      "loss": 0.5673,
+      "step": 2503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121374865735768e-05,
+      "loss": 0.4531,
+      "step": 2504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.121016827783745e-05,
+      "loss": 0.5522,
+      "step": 2505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1206587898317225e-05,
+      "loss": 0.5427,
+      "step": 2506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.120300751879699e-05,
+      "loss": 0.4971,
+      "step": 2507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.1199427139276765e-05,
+      "loss": 0.555,
+      "step": 2508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.119584675975654e-05,
+      "loss": 0.5698,
+      "step": 2509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1192266380236304e-05,
+      "loss": 0.4637,
+      "step": 2510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.118868600071608e-05,
+      "loss": 0.5228,
+      "step": 2511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1185105621195844e-05,
+      "loss": 0.6098,
+      "step": 2512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.118152524167562e-05,
+      "loss": 0.6346,
+      "step": 2513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.117794486215539e-05,
+      "loss": 0.4238,
+      "step": 2514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1174364482635163e-05,
+      "loss": 0.5675,
+      "step": 2515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.117078410311494e-05,
+      "loss": 0.649,
+      "step": 2516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.11672037235947e-05,
+      "loss": 0.4846,
+      "step": 2517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1163623344074476e-05,
+      "loss": 0.5403,
+      "step": 2518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.116004296455425e-05,
+      "loss": 0.5296,
+      "step": 2519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1156462585034016e-05,
+      "loss": 0.5902,
+      "step": 2520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.115288220551379e-05,
+      "loss": 0.4368,
+      "step": 2521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1149301825993555e-05,
+      "loss": 0.5584,
+      "step": 2522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.114572144647333e-05,
+      "loss": 0.4757,
+      "step": 2523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.11421410669531e-05,
+      "loss": 0.5877,
+      "step": 2524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.113856068743287e-05,
+      "loss": 0.4914,
+      "step": 2525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.113498030791264e-05,
+      "loss": 0.5404,
+      "step": 2526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.113139992839241e-05,
+      "loss": 0.5013,
+      "step": 2527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.112781954887218e-05,
+      "loss": 0.439,
+      "step": 2528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1124239169351954e-05,
+      "loss": 0.6282,
+      "step": 2529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.112065878983172e-05,
+      "loss": 0.4511,
+      "step": 2530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1117078410311494e-05,
+      "loss": 0.485,
+      "step": 2531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.111349803079127e-05,
+      "loss": 0.5204,
+      "step": 2532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1109917651271033e-05,
+      "loss": 0.3987,
+      "step": 2533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1106337271750807e-05,
+      "loss": 0.4939,
+      "step": 2534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.110275689223057e-05,
+      "loss": 0.5289,
+      "step": 2535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.1099176512710346e-05,
+      "loss": 0.5365,
+      "step": 2536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109559613319012e-05,
+      "loss": 0.5058,
+      "step": 2537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109201575366989e-05,
+      "loss": 0.6194,
+      "step": 2538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1088435374149666e-05,
+      "loss": 0.6293,
+      "step": 2539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.108485499462943e-05,
+      "loss": 0.5681,
+      "step": 2540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1081274615109205e-05,
+      "loss": 0.5682,
+      "step": 2541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107769423558898e-05,
+      "loss": 0.4808,
+      "step": 2542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1074113856068745e-05,
+      "loss": 0.6183,
+      "step": 2543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.107053347654852e-05,
+      "loss": 0.5851,
+      "step": 2544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1066953097028285e-05,
+      "loss": 0.4849,
+      "step": 2545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.106337271750806e-05,
+      "loss": 0.5878,
+      "step": 2546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.105979233798783e-05,
+      "loss": 0.5088,
+      "step": 2547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.10562119584676e-05,
+      "loss": 0.5229,
+      "step": 2548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.105263157894737e-05,
+      "loss": 0.5058,
+      "step": 2549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1049051199427144e-05,
+      "loss": 0.5277,
+      "step": 2550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.104547081990691e-05,
+      "loss": 0.625,
+      "step": 2551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.104189044038668e-05,
+      "loss": 0.6354,
+      "step": 2552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103831006086645e-05,
+      "loss": 0.475,
+      "step": 2553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103472968134622e-05,
+      "loss": 0.6019,
+      "step": 2554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1031149301825996e-05,
+      "loss": 0.4884,
+      "step": 2555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.102756892230576e-05,
+      "loss": 0.5939,
+      "step": 2556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1023988542785536e-05,
+      "loss": 0.5665,
+      "step": 2557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.102040816326531e-05,
+      "loss": 0.5886,
+      "step": 2558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1016827783745075e-05,
+      "loss": 0.4418,
+      "step": 2559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.101324740422485e-05,
+      "loss": 0.5431,
+      "step": 2560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100966702470462e-05,
+      "loss": 0.5075,
+      "step": 2561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1006086645184395e-05,
+      "loss": 0.4522,
+      "step": 2562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.100250626566416e-05,
+      "loss": 0.3914,
+      "step": 2563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0998925886143934e-05,
+      "loss": 0.5087,
+      "step": 2564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.099534550662371e-05,
+      "loss": 0.4733,
+      "step": 2565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0991765127103474e-05,
+      "loss": 0.6528,
+      "step": 2566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.098818474758325e-05,
+      "loss": 0.4891,
+      "step": 2567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.098460436806302e-05,
+      "loss": 0.5211,
+      "step": 2568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.098102398854279e-05,
+      "loss": 0.5317,
+      "step": 2569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.097744360902256e-05,
+      "loss": 0.5221,
+      "step": 2570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0973863229502326e-05,
+      "loss": 0.6378,
+      "step": 2571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.09702828499821e-05,
+      "loss": 0.5128,
+      "step": 2572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.096670247046187e-05,
+      "loss": 0.5679,
+      "step": 2573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.096312209094164e-05,
+      "loss": 0.5375,
+      "step": 2574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.095954171142141e-05,
+      "loss": 0.6255,
+      "step": 2575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0955961331901186e-05,
+      "loss": 0.5159,
+      "step": 2576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.095238095238095e-05,
+      "loss": 0.4957,
+      "step": 2577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0948800572860725e-05,
+      "loss": 0.5877,
+      "step": 2578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.094522019334049e-05,
+      "loss": 0.5147,
+      "step": 2579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0941639813820265e-05,
+      "loss": 0.5506,
+      "step": 2580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.093805943430004e-05,
+      "loss": 0.5598,
+      "step": 2581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0934479054779804e-05,
+      "loss": 0.4638,
+      "step": 2582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.093089867525958e-05,
+      "loss": 0.4809,
+      "step": 2583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.092731829573935e-05,
+      "loss": 0.589,
+      "step": 2584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0923737916219124e-05,
+      "loss": 0.4937,
+      "step": 2585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.09201575366989e-05,
+      "loss": 0.4035,
+      "step": 2586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0916577157178664e-05,
+      "loss": 0.5323,
+      "step": 2587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.091299677765844e-05,
+      "loss": 0.5327,
+      "step": 2588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.09094163981382e-05,
+      "loss": 0.5307,
+      "step": 2589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0905836018617976e-05,
+      "loss": 0.4069,
+      "step": 2590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.090225563909775e-05,
+      "loss": 0.573,
+      "step": 2591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.0898675259577516e-05,
+      "loss": 0.5012,
+      "step": 2592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.089509488005729e-05,
+      "loss": 0.5479,
+      "step": 2593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.089151450053706e-05,
+      "loss": 0.4372,
+      "step": 2594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.088793412101683e-05,
+      "loss": 0.6049,
+      "step": 2595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.08843537414966e-05,
+      "loss": 0.4684,
+      "step": 2596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.088077336197637e-05,
+      "loss": 0.5218,
+      "step": 2597
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.087719298245614e-05,
+      "loss": 0.6338,
+      "step": 2598
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0873612602935915e-05,
+      "loss": 0.4834,
+      "step": 2599
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.087003222341568e-05,
+      "loss": 0.7416,
+      "step": 2600
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0866451843895454e-05,
+      "loss": 0.6249,
+      "step": 2601
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.086287146437523e-05,
+      "loss": 0.5054,
+      "step": 2602
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0859291084854994e-05,
+      "loss": 0.4821,
+      "step": 2603
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.085571070533477e-05,
+      "loss": 0.4804,
+      "step": 2604
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0852130325814534e-05,
+      "loss": 0.5939,
+      "step": 2605
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.084854994629431e-05,
+      "loss": 0.514,
+      "step": 2606
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.084496956677408e-05,
+      "loss": 0.5561,
+      "step": 2607
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.084138918725385e-05,
+      "loss": 0.4916,
+      "step": 2608
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0837808807733626e-05,
+      "loss": 0.4342,
+      "step": 2609
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.083422842821339e-05,
+      "loss": 0.4551,
+      "step": 2610
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0830648048693166e-05,
+      "loss": 0.5356,
+      "step": 2611
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.082706766917294e-05,
+      "loss": 0.4579,
+      "step": 2612
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0823487289652705e-05,
+      "loss": 0.515,
+      "step": 2613
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081990691013248e-05,
+      "loss": 0.5724,
+      "step": 2614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0816326530612245e-05,
+      "loss": 0.5187,
+      "step": 2615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.081274615109202e-05,
+      "loss": 0.5579,
+      "step": 2616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.080916577157179e-05,
+      "loss": 0.5531,
+      "step": 2617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.080558539205156e-05,
+      "loss": 0.5583,
+      "step": 2618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.080200501253133e-05,
+      "loss": 0.5475,
+      "step": 2619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.0798424633011104e-05,
+      "loss": 0.5744,
+      "step": 2620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.079484425349087e-05,
+      "loss": 0.4234,
+      "step": 2621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0791263873970644e-05,
+      "loss": 0.4558,
+      "step": 2622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.078768349445041e-05,
+      "loss": 0.603,
+      "step": 2623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0784103114930183e-05,
+      "loss": 0.5247,
+      "step": 2624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0780522735409957e-05,
+      "loss": 0.5264,
+      "step": 2625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.077694235588972e-05,
+      "loss": 0.6069,
+      "step": 2626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0773361976369496e-05,
+      "loss": 0.5056,
+      "step": 2627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.076978159684926e-05,
+      "loss": 0.5756,
+      "step": 2628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0766201217329036e-05,
+      "loss": 0.4936,
+      "step": 2629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.076262083780881e-05,
+      "loss": 0.6088,
+      "step": 2630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.075904045828858e-05,
+      "loss": 0.4868,
+      "step": 2631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0755460078768355e-05,
+      "loss": 0.5276,
+      "step": 2632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.075187969924812e-05,
+      "loss": 0.5229,
+      "step": 2633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0748299319727895e-05,
+      "loss": 0.5948,
+      "step": 2634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.074471894020767e-05,
+      "loss": 0.5507,
+      "step": 2635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0741138560687435e-05,
+      "loss": 0.5184,
+      "step": 2636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.073755818116721e-05,
+      "loss": 0.5962,
+      "step": 2637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0733977801646974e-05,
+      "loss": 0.592,
+      "step": 2638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.073039742212675e-05,
+      "loss": 0.504,
+      "step": 2639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.072681704260652e-05,
+      "loss": 0.5334,
+      "step": 2640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.072323666308629e-05,
+      "loss": 0.5653,
+      "step": 2641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071965628356606e-05,
+      "loss": 0.5319,
+      "step": 2642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.071607590404583e-05,
+      "loss": 0.6091,
+      "step": 2643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.07124955245256e-05,
+      "loss": 0.554,
+      "step": 2644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.070891514500537e-05,
+      "loss": 0.545,
+      "step": 2645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.070533476548514e-05,
+      "loss": 0.5713,
+      "step": 2646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.070175438596491e-05,
+      "loss": 0.6839,
+      "step": 2647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0698174006444686e-05,
+      "loss": 0.5031,
+      "step": 2648
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.069459362692445e-05,
+      "loss": 0.4836,
+      "step": 2649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0691013247404225e-05,
+      "loss": 0.5058,
+      "step": 2650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0687432867884e-05,
+      "loss": 0.56,
+      "step": 2651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0683852488363765e-05,
+      "loss": 0.4621,
+      "step": 2652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.068027210884354e-05,
+      "loss": 0.5036,
+      "step": 2653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.067669172932331e-05,
+      "loss": 0.4893,
+      "step": 2654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0673111349803084e-05,
+      "loss": 0.5516,
+      "step": 2655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.066953097028285e-05,
+      "loss": 0.575,
+      "step": 2656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0665950590762624e-05,
+      "loss": 0.417,
+      "step": 2657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.06623702112424e-05,
+      "loss": 0.3888,
+      "step": 2658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0658789831722164e-05,
+      "loss": 0.5372,
+      "step": 2659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.065520945220194e-05,
+      "loss": 0.5202,
+      "step": 2660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.065162907268171e-05,
+      "loss": 0.5037,
+      "step": 2661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0648048693161476e-05,
+      "loss": 0.5272,
+      "step": 2662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.064446831364125e-05,
+      "loss": 0.5046,
+      "step": 2663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0640887934121016e-05,
+      "loss": 0.4577,
+      "step": 2664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.063730755460079e-05,
+      "loss": 0.5984,
+      "step": 2665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.063372717508056e-05,
+      "loss": 0.4086,
+      "step": 2666
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.063014679556033e-05,
+      "loss": 0.4369,
+      "step": 2667
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.06265664160401e-05,
+      "loss": 0.5723,
+      "step": 2668
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0622986036519875e-05,
+      "loss": 0.5225,
+      "step": 2669
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061940565699964e-05,
+      "loss": 0.5087,
+      "step": 2670
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0615825277479415e-05,
+      "loss": 0.4986,
+      "step": 2671
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.061224489795918e-05,
+      "loss": 0.5246,
+      "step": 2672
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0608664518438954e-05,
+      "loss": 0.7454,
+      "step": 2673
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.060508413891873e-05,
+      "loss": 0.5162,
+      "step": 2674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.0601503759398494e-05,
+      "loss": 0.4386,
+      "step": 2675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.059792337987827e-05,
+      "loss": 0.5048,
+      "step": 2676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.059434300035804e-05,
+      "loss": 0.5507,
+      "step": 2677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0590762620837814e-05,
+      "loss": 0.517,
+      "step": 2678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.058718224131759e-05,
+      "loss": 0.6341,
+      "step": 2679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.058360186179735e-05,
+      "loss": 0.4504,
+      "step": 2680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0580021482277126e-05,
+      "loss": 0.6321,
+      "step": 2681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.057644110275689e-05,
+      "loss": 0.5665,
+      "step": 2682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0572860723236666e-05,
+      "loss": 0.5031,
+      "step": 2683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.056928034371644e-05,
+      "loss": 0.6435,
+      "step": 2684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0565699964196206e-05,
+      "loss": 0.451,
+      "step": 2685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.056211958467598e-05,
+      "loss": 0.5544,
+      "step": 2686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.055853920515575e-05,
+      "loss": 0.5584,
+      "step": 2687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.055495882563552e-05,
+      "loss": 0.5872,
+      "step": 2688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.055137844611529e-05,
+      "loss": 0.5198,
+      "step": 2689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.054779806659506e-05,
+      "loss": 0.5689,
+      "step": 2690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.054421768707483e-05,
+      "loss": 0.5051,
+      "step": 2691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0540637307554604e-05,
+      "loss": 0.6671,
+      "step": 2692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.053705692803437e-05,
+      "loss": 0.5824,
+      "step": 2693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0533476548514144e-05,
+      "loss": 0.5052,
+      "step": 2694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.052989616899392e-05,
+      "loss": 0.4918,
+      "step": 2695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0526315789473684e-05,
+      "loss": 0.5857,
+      "step": 2696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.052273540995346e-05,
+      "loss": 0.5447,
+      "step": 2697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.051915503043322e-05,
+      "loss": 0.6089,
+      "step": 2698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0515574650912996e-05,
+      "loss": 0.5031,
+      "step": 2699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.051199427139277e-05,
+      "loss": 0.5294,
+      "step": 2700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050841389187254e-05,
+      "loss": 0.5118,
+      "step": 2701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0504833512352316e-05,
+      "loss": 0.4631,
+      "step": 2702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.050125313283208e-05,
+      "loss": 0.5845,
+      "step": 2703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0497672753311855e-05,
+      "loss": 0.6666,
+      "step": 2704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.049409237379163e-05,
+      "loss": 0.5143,
+      "step": 2705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0490511994271395e-05,
+      "loss": 0.5966,
+      "step": 2706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.048693161475117e-05,
+      "loss": 0.5018,
+      "step": 2707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0483351235230935e-05,
+      "loss": 0.4892,
+      "step": 2708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.047977085571071e-05,
+      "loss": 0.5877,
+      "step": 2709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.5866,
+      "step": 2710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.047261009667025e-05,
+      "loss": 0.5306,
+      "step": 2711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.046902971715002e-05,
+      "loss": 0.4555,
+      "step": 2712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0465449337629794e-05,
+      "loss": 0.482,
+      "step": 2713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.046186895810956e-05,
+      "loss": 0.5948,
+      "step": 2714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0458288578589333e-05,
+      "loss": 0.5513,
+      "step": 2715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.04547081990691e-05,
+      "loss": 0.5044,
+      "step": 2716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.045112781954887e-05,
+      "loss": 0.4561,
+      "step": 2717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0447547440028646e-05,
+      "loss": 0.5431,
+      "step": 2718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.044396706050841e-05,
+      "loss": 0.5417,
+      "step": 2719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0440386680988186e-05,
+      "loss": 0.4798,
+      "step": 2720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.043680630146795e-05,
+      "loss": 0.4288,
+      "step": 2721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0433225921947725e-05,
+      "loss": 0.4856,
+      "step": 2722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.04296455424275e-05,
+      "loss": 0.546,
+      "step": 2723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.042606516290727e-05,
+      "loss": 0.6363,
+      "step": 2724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0422484783387045e-05,
+      "loss": 0.5515,
+      "step": 2725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.041890440386681e-05,
+      "loss": 0.4827,
+      "step": 2726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0415324024346585e-05,
+      "loss": 0.6097,
+      "step": 2727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.041174364482636e-05,
+      "loss": 0.566,
+      "step": 2728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0408163265306124e-05,
+      "loss": 0.5457,
+      "step": 2729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.04045828857859e-05,
+      "loss": 0.5566,
+      "step": 2730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040100250626567e-05,
+      "loss": 0.5033,
+      "step": 2731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.039742212674544e-05,
+      "loss": 0.6272,
+      "step": 2732
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.039384174722521e-05,
+      "loss": 0.4215,
+      "step": 2733
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0390261367704977e-05,
+      "loss": 0.4577,
+      "step": 2734
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.038668098818475e-05,
+      "loss": 0.5127,
+      "step": 2735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.038310060866452e-05,
+      "loss": 0.6176,
+      "step": 2736
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.037952022914429e-05,
+      "loss": 0.488,
+      "step": 2737
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.037593984962406e-05,
+      "loss": 0.5688,
+      "step": 2738
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.037235947010383e-05,
+      "loss": 0.4817,
+      "step": 2739
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.03687790905836e-05,
+      "loss": 0.4249,
+      "step": 2740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0365198711063375e-05,
+      "loss": 0.5686,
+      "step": 2741
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.036161833154314e-05,
+      "loss": 0.5857,
+      "step": 2742
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0358037952022915e-05,
+      "loss": 0.5107,
+      "step": 2743
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.035445757250269e-05,
+      "loss": 0.6573,
+      "step": 2744
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0350877192982455e-05,
+      "loss": 0.5547,
+      "step": 2745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.034729681346223e-05,
+      "loss": 0.5256,
+      "step": 2746
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0343716433941994e-05,
+      "loss": 0.4825,
+      "step": 2747
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0340136054421774e-05,
+      "loss": 0.6603,
+      "step": 2748
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.033655567490154e-05,
+      "loss": 0.5491,
+      "step": 2749
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0332975295381314e-05,
+      "loss": 0.5546,
+      "step": 2750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.032939491586109e-05,
+      "loss": 0.4856,
+      "step": 2751
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.032581453634085e-05,
+      "loss": 0.5333,
+      "step": 2752
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0322234156820626e-05,
+      "loss": 0.5669,
+      "step": 2753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.03186537773004e-05,
+      "loss": 0.5927,
+      "step": 2754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0315073397780166e-05,
+      "loss": 0.6163,
+      "step": 2755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.031149301825994e-05,
+      "loss": 0.451,
+      "step": 2756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.0307912638739706e-05,
+      "loss": 0.5079,
+      "step": 2757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030433225921948e-05,
+      "loss": 0.4808,
+      "step": 2758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.030075187969925e-05,
+      "loss": 0.5342,
+      "step": 2759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.029717150017902e-05,
+      "loss": 0.4905,
+      "step": 2760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.029359112065879e-05,
+      "loss": 0.4659,
+      "step": 2761
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0290010741138565e-05,
+      "loss": 0.4424,
+      "step": 2762
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.028643036161833e-05,
+      "loss": 0.5626,
+      "step": 2763
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0282849982098104e-05,
+      "loss": 0.562,
+      "step": 2764
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.027926960257787e-05,
+      "loss": 0.4962,
+      "step": 2765
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0275689223057644e-05,
+      "loss": 0.5267,
+      "step": 2766
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.027210884353742e-05,
+      "loss": 0.4529,
+      "step": 2767
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0268528464017184e-05,
+      "loss": 0.5205,
+      "step": 2768
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.026494808449696e-05,
+      "loss": 0.3947,
+      "step": 2769
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.026136770497673e-05,
+      "loss": 0.5006,
+      "step": 2770
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.02577873254565e-05,
+      "loss": 0.4882,
+      "step": 2771
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0254206945936276e-05,
+      "loss": 0.3546,
+      "step": 2772
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.025062656641604e-05,
+      "loss": 0.5532,
+      "step": 2773
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0247046186895816e-05,
+      "loss": 0.4751,
+      "step": 2774
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.024346580737558e-05,
+      "loss": 0.3688,
+      "step": 2775
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0239885427855356e-05,
+      "loss": 0.432,
+      "step": 2776
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.023630504833513e-05,
+      "loss": 0.5745,
+      "step": 2777
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0232724668814895e-05,
+      "loss": 0.4726,
+      "step": 2778
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.022914428929467e-05,
+      "loss": 0.5436,
+      "step": 2779
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.022556390977444e-05,
+      "loss": 0.5466,
+      "step": 2780
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.022198353025421e-05,
+      "loss": 0.4763,
+      "step": 2781
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021840315073398e-05,
+      "loss": 0.5238,
+      "step": 2782
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021482277121375e-05,
+      "loss": 0.5015,
+      "step": 2783
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.021124239169352e-05,
+      "loss": 0.6661,
+      "step": 2784
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0207662012173294e-05,
+      "loss": 0.5508,
+      "step": 2785
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.020408163265306e-05,
+      "loss": 0.4218,
+      "step": 2786
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0200501253132834e-05,
+      "loss": 0.6453,
+      "step": 2787
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.019692087361261e-05,
+      "loss": 0.5053,
+      "step": 2788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.019334049409237e-05,
+      "loss": 0.6231,
+      "step": 2789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0189760114572146e-05,
+      "loss": 0.4891,
+      "step": 2790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.018617973505191e-05,
+      "loss": 0.4705,
+      "step": 2791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0182599355531686e-05,
+      "loss": 0.4452,
+      "step": 2792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.017901897601146e-05,
+      "loss": 0.5777,
+      "step": 2793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.017543859649123e-05,
+      "loss": 0.6075,
+      "step": 2794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0171858216971005e-05,
+      "loss": 0.5102,
+      "step": 2795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.016827783745077e-05,
+      "loss": 0.5495,
+      "step": 2796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0164697457930545e-05,
+      "loss": 0.5328,
+      "step": 2797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.016111707841032e-05,
+      "loss": 0.5199,
+      "step": 2798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0157536698890085e-05,
+      "loss": 0.4757,
+      "step": 2799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.015395631936986e-05,
+      "loss": 0.6732,
+      "step": 2800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0150375939849624e-05,
+      "loss": 0.6048,
+      "step": 2801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.01467955603294e-05,
+      "loss": 0.4561,
+      "step": 2802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.014321518080917e-05,
+      "loss": 0.6172,
+      "step": 2803
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7440963318095662,
+      "eval_loss": 0.5148146748542786,
+      "eval_runtime": 25.8168,
+      "eval_samples_per_second": 1737.045,
+      "eval_steps_per_second": 27.153,
+      "step": 2803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.013963480128894e-05,
+      "loss": 0.5686,
+      "step": 2804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.013605442176871e-05,
+      "loss": 0.4988,
+      "step": 2805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0132474042248483e-05,
+      "loss": 0.4112,
+      "step": 2806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.012889366272825e-05,
+      "loss": 0.5719,
+      "step": 2807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.012531328320802e-05,
+      "loss": 0.4022,
+      "step": 2808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.012173290368779e-05,
+      "loss": 0.3688,
+      "step": 2809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.011815252416756e-05,
+      "loss": 0.3802,
+      "step": 2810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0114572144647336e-05,
+      "loss": 0.4172,
+      "step": 2811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.01109917651271e-05,
+      "loss": 0.3883,
+      "step": 2812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0107411385606875e-05,
+      "loss": 0.4553,
+      "step": 2813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.010383100608664e-05,
+      "loss": 0.3907,
+      "step": 2814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0100250626566415e-05,
+      "loss": 0.4234,
+      "step": 2815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.009667024704619e-05,
+      "loss": 0.4394,
+      "step": 2816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0093089867525955e-05,
+      "loss": 0.3725,
+      "step": 2817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0089509488005735e-05,
+      "loss": 0.4525,
+      "step": 2818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.00859291084855e-05,
+      "loss": 0.5327,
+      "step": 2819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0082348728965274e-05,
+      "loss": 0.2809,
+      "step": 2820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.007876834944505e-05,
+      "loss": 0.3852,
+      "step": 2821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0075187969924814e-05,
+      "loss": 0.3061,
+      "step": 2822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.007160759040459e-05,
+      "loss": 0.4238,
+      "step": 2823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.006802721088436e-05,
+      "loss": 0.4511,
+      "step": 2824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0064446831364127e-05,
+      "loss": 0.3755,
+      "step": 2825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.00608664518439e-05,
+      "loss": 0.392,
+      "step": 2826
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0057286072323666e-05,
+      "loss": 0.4122,
+      "step": 2827
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.005370569280344e-05,
+      "loss": 0.3632,
+      "step": 2828
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.005012531328321e-05,
+      "loss": 0.3853,
+      "step": 2829
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.004654493376298e-05,
+      "loss": 0.435,
+      "step": 2830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.004296455424275e-05,
+      "loss": 0.4573,
+      "step": 2831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.003938417472252e-05,
+      "loss": 0.4782,
+      "step": 2832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.003580379520229e-05,
+      "loss": 0.4889,
+      "step": 2833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0032223415682065e-05,
+      "loss": 0.3583,
+      "step": 2834
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.002864303616183e-05,
+      "loss": 0.4254,
+      "step": 2835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0025062656641605e-05,
+      "loss": 0.5451,
+      "step": 2836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.002148227712138e-05,
+      "loss": 0.5426,
+      "step": 2837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0017901897601144e-05,
+      "loss": 0.5596,
+      "step": 2838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.001432151808092e-05,
+      "loss": 0.42,
+      "step": 2839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0010741138560684e-05,
+      "loss": 0.4457,
+      "step": 2840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0007160759040464e-05,
+      "loss": 0.4428,
+      "step": 2841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.000358037952024e-05,
+      "loss": 0.3812,
+      "step": 2842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4e-05,
+      "loss": 0.4849,
+      "step": 2843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.9996419620479776e-05,
+      "loss": 0.6266,
+      "step": 2844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.999283924095954e-05,
+      "loss": 0.3724,
+      "step": 2845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9989258861439316e-05,
+      "loss": 0.4688,
+      "step": 2846
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.998567848191909e-05,
+      "loss": 0.3871,
+      "step": 2847
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9982098102398856e-05,
+      "loss": 0.4298,
+      "step": 2848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.997851772287863e-05,
+      "loss": 0.4565,
+      "step": 2849
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9974937343358395e-05,
+      "loss": 0.4822,
+      "step": 2850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.997135696383817e-05,
+      "loss": 0.3903,
+      "step": 2851
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.996777658431794e-05,
+      "loss": 0.5263,
+      "step": 2852
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.996419620479771e-05,
+      "loss": 0.4243,
+      "step": 2853
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.996061582527748e-05,
+      "loss": 0.4144,
+      "step": 2854
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9957035445757254e-05,
+      "loss": 0.4077,
+      "step": 2855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.995345506623702e-05,
+      "loss": 0.4726,
+      "step": 2856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9949874686716794e-05,
+      "loss": 0.4155,
+      "step": 2857
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.994629430719656e-05,
+      "loss": 0.4958,
+      "step": 2858
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9942713927676334e-05,
+      "loss": 0.4706,
+      "step": 2859
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.993913354815611e-05,
+      "loss": 0.3751,
+      "step": 2860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.993555316863587e-05,
+      "loss": 0.3989,
+      "step": 2861
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9931972789115646e-05,
+      "loss": 0.4376,
+      "step": 2862
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.992839240959542e-05,
+      "loss": 0.4184,
+      "step": 2863
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9924812030075186e-05,
+      "loss": 0.3941,
+      "step": 2864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9921231650554966e-05,
+      "loss": 0.3355,
+      "step": 2865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.991765127103473e-05,
+      "loss": 0.4062,
+      "step": 2866
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9914070891514506e-05,
+      "loss": 0.4768,
+      "step": 2867
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.991049051199427e-05,
+      "loss": 0.4008,
+      "step": 2868
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9906910132474045e-05,
+      "loss": 0.4674,
+      "step": 2869
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.990332975295382e-05,
+      "loss": 0.4264,
+      "step": 2870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.9899749373433585e-05,
+      "loss": 0.3566,
+      "step": 2871
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989616899391336e-05,
+      "loss": 0.494,
+      "step": 2872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.989258861439313e-05,
+      "loss": 0.5344,
+      "step": 2873
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.98890082348729e-05,
+      "loss": 0.4867,
+      "step": 2874
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.988542785535267e-05,
+      "loss": 0.2628,
+      "step": 2875
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.988184747583244e-05,
+      "loss": 0.3515,
+      "step": 2876
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.987826709631221e-05,
+      "loss": 0.4316,
+      "step": 2877
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9874686716791984e-05,
+      "loss": 0.3518,
+      "step": 2878
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.987110633727175e-05,
+      "loss": 0.521,
+      "step": 2879
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.986752595775152e-05,
+      "loss": 0.547,
+      "step": 2880
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9863945578231296e-05,
+      "loss": 0.4946,
+      "step": 2881
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.986036519871106e-05,
+      "loss": 0.4224,
+      "step": 2882
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9856784819190836e-05,
+      "loss": 0.3359,
+      "step": 2883
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.98532044396706e-05,
+      "loss": 0.6293,
+      "step": 2884
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9849624060150376e-05,
+      "loss": 0.4912,
+      "step": 2885
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.984604368063015e-05,
+      "loss": 0.3694,
+      "step": 2886
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9842463301109915e-05,
+      "loss": 0.4132,
+      "step": 2887
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9838882921589695e-05,
+      "loss": 0.4966,
+      "step": 2888
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.983530254206946e-05,
+      "loss": 0.4504,
+      "step": 2889
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9831722162549235e-05,
+      "loss": 0.4614,
+      "step": 2890
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.982814178302901e-05,
+      "loss": 0.5544,
+      "step": 2891
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9824561403508774e-05,
+      "loss": 0.4262,
+      "step": 2892
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.982098102398855e-05,
+      "loss": 0.4017,
+      "step": 2893
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.9817400644468314e-05,
+      "loss": 0.3868,
+      "step": 2894
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981382026494809e-05,
+      "loss": 0.389,
+      "step": 2895
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.981023988542786e-05,
+      "loss": 0.395,
+      "step": 2896
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.980665950590763e-05,
+      "loss": 0.4909,
+      "step": 2897
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.98030791263874e-05,
+      "loss": 0.4479,
+      "step": 2898
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979949874686717e-05,
+      "loss": 0.5408,
+      "step": 2899
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979591836734694e-05,
+      "loss": 0.4198,
+      "step": 2900
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.979233798782671e-05,
+      "loss": 0.3484,
+      "step": 2901
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.978875760830648e-05,
+      "loss": 0.4362,
+      "step": 2902
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.978517722878625e-05,
+      "loss": 0.475,
+      "step": 2903
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9781596849266025e-05,
+      "loss": 0.4296,
+      "step": 2904
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.977801646974579e-05,
+      "loss": 0.4238,
+      "step": 2905
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9774436090225565e-05,
+      "loss": 0.4284,
+      "step": 2906
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.977085571070533e-05,
+      "loss": 0.3361,
+      "step": 2907
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9767275331185105e-05,
+      "loss": 0.447,
+      "step": 2908
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.976369495166488e-05,
+      "loss": 0.4453,
+      "step": 2909
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9760114572144644e-05,
+      "loss": 0.4961,
+      "step": 2910
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9756534192624424e-05,
+      "loss": 0.4606,
+      "step": 2911
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.975295381310419e-05,
+      "loss": 0.5482,
+      "step": 2912
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9749373433583964e-05,
+      "loss": 0.38,
+      "step": 2913
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.974579305406374e-05,
+      "loss": 0.3386,
+      "step": 2914
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.97422126745435e-05,
+      "loss": 0.3928,
+      "step": 2915
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9738632295023277e-05,
+      "loss": 0.4696,
+      "step": 2916
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.973505191550305e-05,
+      "loss": 0.345,
+      "step": 2917
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9731471535982816e-05,
+      "loss": 0.3804,
+      "step": 2918
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.972789115646259e-05,
+      "loss": 0.3772,
+      "step": 2919
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9724310776942356e-05,
+      "loss": 0.4399,
+      "step": 2920
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.972073039742213e-05,
+      "loss": 0.3797,
+      "step": 2921
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.97171500179019e-05,
+      "loss": 0.4201,
+      "step": 2922
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.971356963838167e-05,
+      "loss": 0.4553,
+      "step": 2923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970998925886144e-05,
+      "loss": 0.3922,
+      "step": 2924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970640887934121e-05,
+      "loss": 0.4259,
+      "step": 2925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.970282849982098e-05,
+      "loss": 0.3891,
+      "step": 2926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9699248120300755e-05,
+      "loss": 0.4531,
+      "step": 2927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.969566774078052e-05,
+      "loss": 0.3704,
+      "step": 2928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.9692087361260294e-05,
+      "loss": 0.3815,
+      "step": 2929
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.968850698174007e-05,
+      "loss": 0.38,
+      "step": 2930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9684926602219834e-05,
+      "loss": 0.4257,
+      "step": 2931
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.968134622269961e-05,
+      "loss": 0.3743,
+      "step": 2932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.967776584317937e-05,
+      "loss": 0.554,
+      "step": 2933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9674185463659146e-05,
+      "loss": 0.2964,
+      "step": 2934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9670605084138926e-05,
+      "loss": 0.4785,
+      "step": 2935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.966702470461869e-05,
+      "loss": 0.2569,
+      "step": 2936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9663444325098466e-05,
+      "loss": 0.4897,
+      "step": 2937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.965986394557823e-05,
+      "loss": 0.411,
+      "step": 2938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9656283566058006e-05,
+      "loss": 0.3609,
+      "step": 2939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.965270318653778e-05,
+      "loss": 0.5334,
+      "step": 2940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9649122807017545e-05,
+      "loss": 0.3892,
+      "step": 2941
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.964554242749732e-05,
+      "loss": 0.289,
+      "step": 2942
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9641962047977085e-05,
+      "loss": 0.4506,
+      "step": 2943
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.963838166845686e-05,
+      "loss": 0.4322,
+      "step": 2944
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.963480128893663e-05,
+      "loss": 0.4125,
+      "step": 2945
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.96312209094164e-05,
+      "loss": 0.3271,
+      "step": 2946
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.962764052989617e-05,
+      "loss": 0.3929,
+      "step": 2947
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9624060150375944e-05,
+      "loss": 0.3666,
+      "step": 2948
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.962047977085571e-05,
+      "loss": 0.3428,
+      "step": 2949
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9616899391335484e-05,
+      "loss": 0.4263,
+      "step": 2950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.961331901181525e-05,
+      "loss": 0.3981,
+      "step": 2951
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.960973863229502e-05,
+      "loss": 0.3866,
+      "step": 2952
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9606158252774796e-05,
+      "loss": 0.3535,
+      "step": 2953
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.960257787325456e-05,
+      "loss": 0.4332,
+      "step": 2954
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9598997493734336e-05,
+      "loss": 0.3517,
+      "step": 2955
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.959541711421411e-05,
+      "loss": 0.5091,
+      "step": 2956
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.9591836734693876e-05,
+      "loss": 0.445,
+      "step": 2957
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9588256355173656e-05,
+      "loss": 0.6204,
+      "step": 2958
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.958467597565342e-05,
+      "loss": 0.3353,
+      "step": 2959
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9581095596133195e-05,
+      "loss": 0.4284,
+      "step": 2960
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.957751521661296e-05,
+      "loss": 0.4291,
+      "step": 2961
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9573934837092735e-05,
+      "loss": 0.4546,
+      "step": 2962
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.957035445757251e-05,
+      "loss": 0.406,
+      "step": 2963
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9566774078052274e-05,
+      "loss": 0.3865,
+      "step": 2964
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.956319369853205e-05,
+      "loss": 0.3089,
+      "step": 2965
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.955961331901182e-05,
+      "loss": 0.4808,
+      "step": 2966
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.955603293949159e-05,
+      "loss": 0.5004,
+      "step": 2967
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.955245255997136e-05,
+      "loss": 0.448,
+      "step": 2968
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.954887218045113e-05,
+      "loss": 0.4416,
+      "step": 2969
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.95452918009309e-05,
+      "loss": 0.4764,
+      "step": 2970
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.954171142141067e-05,
+      "loss": 0.4883,
+      "step": 2971
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.953813104189044e-05,
+      "loss": 0.3057,
+      "step": 2972
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.953455066237021e-05,
+      "loss": 0.4343,
+      "step": 2973
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9530970282849986e-05,
+      "loss": 0.4173,
+      "step": 2974
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.952738990332975e-05,
+      "loss": 0.3265,
+      "step": 2975
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9523809523809526e-05,
+      "loss": 0.4307,
+      "step": 2976
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.952022914428929e-05,
+      "loss": 0.4153,
+      "step": 2977
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9516648764769065e-05,
+      "loss": 0.393,
+      "step": 2978
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.951306838524884e-05,
+      "loss": 0.3421,
+      "step": 2979
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9509488005728605e-05,
+      "loss": 0.4994,
+      "step": 2980
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9505907626208385e-05,
+      "loss": 0.3575,
+      "step": 2981
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.950232724668815e-05,
+      "loss": 0.3584,
+      "step": 2982
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9498746867167924e-05,
+      "loss": 0.3942,
+      "step": 2983
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.94951664876477e-05,
+      "loss": 0.4156,
+      "step": 2984
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.9491586108127464e-05,
+      "loss": 0.4289,
+      "step": 2985
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.948800572860724e-05,
+      "loss": 0.4448,
+      "step": 2986
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9484425349087003e-05,
+      "loss": 0.4498,
+      "step": 2987
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.948084496956678e-05,
+      "loss": 0.4722,
+      "step": 2988
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.947726459004655e-05,
+      "loss": 0.5337,
+      "step": 2989
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.6084,
+      "step": 2990
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.947010383100609e-05,
+      "loss": 0.3578,
+      "step": 2991
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.946652345148586e-05,
+      "loss": 0.414,
+      "step": 2992
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.946294307196563e-05,
+      "loss": 0.4819,
+      "step": 2993
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.94593626924454e-05,
+      "loss": 0.3882,
+      "step": 2994
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.945578231292517e-05,
+      "loss": 0.3866,
+      "step": 2995
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.945220193340494e-05,
+      "loss": 0.4578,
+      "step": 2996
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9448621553884715e-05,
+      "loss": 0.3266,
+      "step": 2997
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.944504117436448e-05,
+      "loss": 0.3085,
+      "step": 2998
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9441460794844255e-05,
+      "loss": 0.3302,
+      "step": 2999
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.943788041532403e-05,
+      "loss": 0.4691,
+      "step": 3000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9434300035803794e-05,
+      "loss": 0.3866,
+      "step": 3001
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.943071965628357e-05,
+      "loss": 0.5496,
+      "step": 3002
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9427139276763334e-05,
+      "loss": 0.515,
+      "step": 3003
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.942355889724311e-05,
+      "loss": 0.4467,
+      "step": 3004
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.941997851772288e-05,
+      "loss": 0.6531,
+      "step": 3005
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.941639813820265e-05,
+      "loss": 0.443,
+      "step": 3006
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9412817758682427e-05,
+      "loss": 0.4974,
+      "step": 3007
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.940923737916219e-05,
+      "loss": 0.3094,
+      "step": 3008
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9405656999641966e-05,
+      "loss": 0.5455,
+      "step": 3009
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.940207662012174e-05,
+      "loss": 0.3077,
+      "step": 3010
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9398496240601506e-05,
+      "loss": 0.3942,
+      "step": 3011
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.939491586108128e-05,
+      "loss": 0.3264,
+      "step": 3012
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.9391335481561045e-05,
+      "loss": 0.5687,
+      "step": 3013
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.938775510204082e-05,
+      "loss": 0.506,
+      "step": 3014
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.938417472252059e-05,
+      "loss": 0.3745,
+      "step": 3015
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.938059434300036e-05,
+      "loss": 0.4526,
+      "step": 3016
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.937701396348013e-05,
+      "loss": 0.3805,
+      "step": 3017
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.93734335839599e-05,
+      "loss": 0.3905,
+      "step": 3018
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936985320443967e-05,
+      "loss": 0.4848,
+      "step": 3019
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9366272824919444e-05,
+      "loss": 0.3944,
+      "step": 3020
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.936269244539921e-05,
+      "loss": 0.5307,
+      "step": 3021
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9359112065878984e-05,
+      "loss": 0.5042,
+      "step": 3022
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.935553168635876e-05,
+      "loss": 0.471,
+      "step": 3023
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.935195130683852e-05,
+      "loss": 0.5239,
+      "step": 3024
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9348370927318297e-05,
+      "loss": 0.4661,
+      "step": 3025
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.934479054779806e-05,
+      "loss": 0.4156,
+      "step": 3026
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9341210168277836e-05,
+      "loss": 0.3809,
+      "step": 3027
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9337629788757616e-05,
+      "loss": 0.4333,
+      "step": 3028
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.933404940923738e-05,
+      "loss": 0.3822,
+      "step": 3029
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9330469029717156e-05,
+      "loss": 0.4121,
+      "step": 3030
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.932688865019692e-05,
+      "loss": 0.3884,
+      "step": 3031
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9323308270676695e-05,
+      "loss": 0.4239,
+      "step": 3032
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931972789115647e-05,
+      "loss": 0.3942,
+      "step": 3033
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9316147511636235e-05,
+      "loss": 0.3742,
+      "step": 3034
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.931256713211601e-05,
+      "loss": 0.4985,
+      "step": 3035
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9308986752595774e-05,
+      "loss": 0.4653,
+      "step": 3036
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.930540637307555e-05,
+      "loss": 0.3345,
+      "step": 3037
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.930182599355532e-05,
+      "loss": 0.4653,
+      "step": 3038
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929824561403509e-05,
+      "loss": 0.5438,
+      "step": 3039
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.929466523451486e-05,
+      "loss": 0.3765,
+      "step": 3040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.9291084854994634e-05,
+      "loss": 0.3951,
+      "step": 3041
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.92875044754744e-05,
+      "loss": 0.3926,
+      "step": 3042
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.928392409595417e-05,
+      "loss": 0.3523,
+      "step": 3043
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.928034371643394e-05,
+      "loss": 0.4155,
+      "step": 3044
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.927676333691371e-05,
+      "loss": 0.39,
+      "step": 3045
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9273182957393486e-05,
+      "loss": 0.3374,
+      "step": 3046
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.926960257787325e-05,
+      "loss": 0.3034,
+      "step": 3047
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9266022198353026e-05,
+      "loss": 0.3872,
+      "step": 3048
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.92624418188328e-05,
+      "loss": 0.4143,
+      "step": 3049
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9258861439312565e-05,
+      "loss": 0.6241,
+      "step": 3050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9255281059792345e-05,
+      "loss": 0.44,
+      "step": 3051
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.925170068027211e-05,
+      "loss": 0.6088,
+      "step": 3052
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9248120300751885e-05,
+      "loss": 0.4237,
+      "step": 3053
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.924453992123165e-05,
+      "loss": 0.427,
+      "step": 3054
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9240959541711424e-05,
+      "loss": 0.5286,
+      "step": 3055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.92373791621912e-05,
+      "loss": 0.3715,
+      "step": 3056
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9233798782670964e-05,
+      "loss": 0.4204,
+      "step": 3057
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.923021840315074e-05,
+      "loss": 0.3723,
+      "step": 3058
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.922663802363051e-05,
+      "loss": 0.4216,
+      "step": 3059
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.922305764411028e-05,
+      "loss": 0.4352,
+      "step": 3060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.921947726459005e-05,
+      "loss": 0.4508,
+      "step": 3061
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9215896885069816e-05,
+      "loss": 0.4666,
+      "step": 3062
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.921231650554959e-05,
+      "loss": 0.3221,
+      "step": 3063
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920873612602936e-05,
+      "loss": 0.4238,
+      "step": 3064
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.920515574650913e-05,
+      "loss": 0.4535,
+      "step": 3065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.92015753669889e-05,
+      "loss": 0.3818,
+      "step": 3066
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9197994987468676e-05,
+      "loss": 0.5457,
+      "step": 3067
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.919441460794844e-05,
+      "loss": 0.4904,
+      "step": 3068
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.9190834228428215e-05,
+      "loss": 0.3519,
+      "step": 3069
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.918725384890798e-05,
+      "loss": 0.5294,
+      "step": 3070
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9183673469387755e-05,
+      "loss": 0.3621,
+      "step": 3071
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.918009308986753e-05,
+      "loss": 0.4134,
+      "step": 3072
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9176512710347294e-05,
+      "loss": 0.4336,
+      "step": 3073
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.917293233082707e-05,
+      "loss": 0.3231,
+      "step": 3074
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.916935195130684e-05,
+      "loss": 0.3386,
+      "step": 3075
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9165771571786614e-05,
+      "loss": 0.4978,
+      "step": 3076
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.916219119226639e-05,
+      "loss": 0.5019,
+      "step": 3077
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9158610812746153e-05,
+      "loss": 0.3825,
+      "step": 3078
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.915503043322593e-05,
+      "loss": 0.4164,
+      "step": 3079
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.915145005370569e-05,
+      "loss": 0.4286,
+      "step": 3080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9147869674185466e-05,
+      "loss": 0.3312,
+      "step": 3081
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.914428929466524e-05,
+      "loss": 0.3407,
+      "step": 3082
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9140708915145006e-05,
+      "loss": 0.2946,
+      "step": 3083
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.913712853562478e-05,
+      "loss": 0.4656,
+      "step": 3084
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.913354815610455e-05,
+      "loss": 0.3387,
+      "step": 3085
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.912996777658432e-05,
+      "loss": 0.4641,
+      "step": 3086
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.912638739706409e-05,
+      "loss": 0.4008,
+      "step": 3087
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.912280701754386e-05,
+      "loss": 0.4845,
+      "step": 3088
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.911922663802363e-05,
+      "loss": 0.5101,
+      "step": 3089
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9115646258503405e-05,
+      "loss": 0.4204,
+      "step": 3090
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.911206587898317e-05,
+      "loss": 0.4782,
+      "step": 3091
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9108485499462944e-05,
+      "loss": 0.6271,
+      "step": 3092
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.910490511994272e-05,
+      "loss": 0.4084,
+      "step": 3093
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9101324740422484e-05,
+      "loss": 0.4423,
+      "step": 3094
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.909774436090226e-05,
+      "loss": 0.3683,
+      "step": 3095
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9094163981382023e-05,
+      "loss": 0.5173,
+      "step": 3096
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.9090583601861797e-05,
+      "loss": 0.4448,
+      "step": 3097
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.908700322234157e-05,
+      "loss": 0.2896,
+      "step": 3098
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.908342284282134e-05,
+      "loss": 0.4696,
+      "step": 3099
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9079842463301116e-05,
+      "loss": 0.3557,
+      "step": 3100
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.907626208378088e-05,
+      "loss": 0.5205,
+      "step": 3101
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9072681704260656e-05,
+      "loss": 0.3718,
+      "step": 3102
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.906910132474043e-05,
+      "loss": 0.5444,
+      "step": 3103
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9065520945220195e-05,
+      "loss": 0.3566,
+      "step": 3104
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.906194056569997e-05,
+      "loss": 0.4336,
+      "step": 3105
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9058360186179735e-05,
+      "loss": 0.4268,
+      "step": 3106
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.905477980665951e-05,
+      "loss": 0.4778,
+      "step": 3107
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.905119942713928e-05,
+      "loss": 0.4711,
+      "step": 3108
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.904761904761905e-05,
+      "loss": 0.4145,
+      "step": 3109
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.904403866809882e-05,
+      "loss": 0.3242,
+      "step": 3110
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9040458288578594e-05,
+      "loss": 0.4029,
+      "step": 3111
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.903687790905836e-05,
+      "loss": 0.3238,
+      "step": 3112
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9033297529538134e-05,
+      "loss": 0.3961,
+      "step": 3113
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.90297171500179e-05,
+      "loss": 0.5941,
+      "step": 3114
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.902613677049767e-05,
+      "loss": 0.3977,
+      "step": 3115
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9022556390977447e-05,
+      "loss": 0.4322,
+      "step": 3116
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.901897601145721e-05,
+      "loss": 0.3783,
+      "step": 3117
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9015395631936986e-05,
+      "loss": 0.5055,
+      "step": 3118
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.901181525241675e-05,
+      "loss": 0.4073,
+      "step": 3119
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9008234872896526e-05,
+      "loss": 0.3897,
+      "step": 3120
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.9004654493376306e-05,
+      "loss": 0.4329,
+      "step": 3121
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.900107411385607e-05,
+      "loss": 0.5359,
+      "step": 3122
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8997493734335845e-05,
+      "loss": 0.369,
+      "step": 3123
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.899391335481561e-05,
+      "loss": 0.3719,
+      "step": 3124
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.8990332975295385e-05,
+      "loss": 0.4445,
+      "step": 3125
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.898675259577516e-05,
+      "loss": 0.3543,
+      "step": 3126
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8983172216254924e-05,
+      "loss": 0.3749,
+      "step": 3127
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.89795918367347e-05,
+      "loss": 0.3086,
+      "step": 3128
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8976011457214464e-05,
+      "loss": 0.4525,
+      "step": 3129
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.897243107769424e-05,
+      "loss": 0.5228,
+      "step": 3130
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.896885069817401e-05,
+      "loss": 0.2867,
+      "step": 3131
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.896527031865378e-05,
+      "loss": 0.6048,
+      "step": 3132
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.896168993913355e-05,
+      "loss": 0.3794,
+      "step": 3133
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.895810955961332e-05,
+      "loss": 0.4455,
+      "step": 3134
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.895452918009309e-05,
+      "loss": 0.3258,
+      "step": 3135
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.895094880057286e-05,
+      "loss": 0.3865,
+      "step": 3136
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.894736842105263e-05,
+      "loss": 0.4176,
+      "step": 3137
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.89437880415324e-05,
+      "loss": 0.5362,
+      "step": 3138
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8940207662012176e-05,
+      "loss": 0.3585,
+      "step": 3139
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.893662728249194e-05,
+      "loss": 0.4434,
+      "step": 3140
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8933046902971715e-05,
+      "loss": 0.3458,
+      "step": 3141
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.892946652345149e-05,
+      "loss": 0.5203,
+      "step": 3142
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8925886143931255e-05,
+      "loss": 0.3579,
+      "step": 3143
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.892230576441103e-05,
+      "loss": 0.4395,
+      "step": 3144
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.89187253848908e-05,
+      "loss": 0.4717,
+      "step": 3145
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8915145005370574e-05,
+      "loss": 0.3744,
+      "step": 3146
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.891156462585034e-05,
+      "loss": 0.3107,
+      "step": 3147
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8907984246330114e-05,
+      "loss": 0.4353,
+      "step": 3148
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.890440386680989e-05,
+      "loss": 0.4576,
+      "step": 3149
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8900823487289654e-05,
+      "loss": 0.4569,
+      "step": 3150
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.889724310776943e-05,
+      "loss": 0.4647,
+      "step": 3151
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.88936627282492e-05,
+      "loss": 0.4744,
+      "step": 3152
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.8890082348728966e-05,
+      "loss": 0.4556,
+      "step": 3153
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.888650196920874e-05,
+      "loss": 0.3896,
+      "step": 3154
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8882921589688506e-05,
+      "loss": 0.4563,
+      "step": 3155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.887934121016828e-05,
+      "loss": 0.4255,
+      "step": 3156
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.887576083064805e-05,
+      "loss": 0.5593,
+      "step": 3157
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.887218045112782e-05,
+      "loss": 0.4824,
+      "step": 3158
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.886860007160759e-05,
+      "loss": 0.5048,
+      "step": 3159
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8865019692087365e-05,
+      "loss": 0.4651,
+      "step": 3160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.886143931256713e-05,
+      "loss": 0.4694,
+      "step": 3161
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8857858933046905e-05,
+      "loss": 0.3475,
+      "step": 3162
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.885427855352667e-05,
+      "loss": 0.3827,
+      "step": 3163
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8850698174006444e-05,
+      "loss": 0.3438,
+      "step": 3164
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.884711779448622e-05,
+      "loss": 0.5392,
+      "step": 3165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8843537414965984e-05,
+      "loss": 0.3998,
+      "step": 3166
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.883995703544576e-05,
+      "loss": 0.3624,
+      "step": 3167
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.883637665592553e-05,
+      "loss": 0.4792,
+      "step": 3168
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8832796276405303e-05,
+      "loss": 0.438,
+      "step": 3169
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.882921589688508e-05,
+      "loss": 0.4338,
+      "step": 3170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.882563551736484e-05,
+      "loss": 0.3845,
+      "step": 3171
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8822055137844616e-05,
+      "loss": 0.3748,
+      "step": 3172
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881847475832438e-05,
+      "loss": 0.4438,
+      "step": 3173
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8814894378804156e-05,
+      "loss": 0.4445,
+      "step": 3174
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.881131399928393e-05,
+      "loss": 0.3569,
+      "step": 3175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.8807733619763695e-05,
+      "loss": 0.4231,
+      "step": 3176
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880415324024347e-05,
+      "loss": 0.412,
+      "step": 3177
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.880057286072324e-05,
+      "loss": 0.4981,
+      "step": 3178
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.879699248120301e-05,
+      "loss": 0.572,
+      "step": 3179
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.879341210168278e-05,
+      "loss": 0.3437,
+      "step": 3180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.878983172216255e-05,
+      "loss": 0.6056,
+      "step": 3181
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.878625134264232e-05,
+      "loss": 0.4567,
+      "step": 3182
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8782670963122094e-05,
+      "loss": 0.3125,
+      "step": 3183
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.877909058360186e-05,
+      "loss": 0.5031,
+      "step": 3184
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8775510204081634e-05,
+      "loss": 0.3438,
+      "step": 3185
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.877192982456141e-05,
+      "loss": 0.3281,
+      "step": 3186
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8768349445041173e-05,
+      "loss": 0.4085,
+      "step": 3187
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8764769065520947e-05,
+      "loss": 0.4002,
+      "step": 3188
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.876118868600071e-05,
+      "loss": 0.2696,
+      "step": 3189
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8757608306480486e-05,
+      "loss": 0.3961,
+      "step": 3190
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.875402792696026e-05,
+      "loss": 0.3838,
+      "step": 3191
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.875044754744003e-05,
+      "loss": 0.5462,
+      "step": 3192
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8746867167919806e-05,
+      "loss": 0.3589,
+      "step": 3193
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.874328678839957e-05,
+      "loss": 0.3827,
+      "step": 3194
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8739706408879345e-05,
+      "loss": 0.4316,
+      "step": 3195
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.873612602935912e-05,
+      "loss": 0.3469,
+      "step": 3196
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8732545649838885e-05,
+      "loss": 0.5556,
+      "step": 3197
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.872896527031866e-05,
+      "loss": 0.4227,
+      "step": 3198
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8725384890798425e-05,
+      "loss": 0.5299,
+      "step": 3199
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.87218045112782e-05,
+      "loss": 0.4113,
+      "step": 3200
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871822413175797e-05,
+      "loss": 0.5434,
+      "step": 3201
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871464375223774e-05,
+      "loss": 0.4742,
+      "step": 3202
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.871106337271751e-05,
+      "loss": 0.3065,
+      "step": 3203
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8707482993197284e-05,
+      "loss": 0.3423,
+      "step": 3204
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.870390261367705e-05,
+      "loss": 0.3476,
+      "step": 3205
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.870032223415682e-05,
+      "loss": 0.3447,
+      "step": 3206
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.869674185463659e-05,
+      "loss": 0.396,
+      "step": 3207
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.869316147511636e-05,
+      "loss": 0.3433,
+      "step": 3208
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.8689581095596136e-05,
+      "loss": 0.4453,
+      "step": 3209
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86860007160759e-05,
+      "loss": 0.3481,
+      "step": 3210
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8682420336555676e-05,
+      "loss": 0.4126,
+      "step": 3211
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.867883995703544e-05,
+      "loss": 0.435,
+      "step": 3212
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8675259577515215e-05,
+      "loss": 0.4852,
+      "step": 3213
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.867167919799499e-05,
+      "loss": 0.3973,
+      "step": 3214
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.866809881847476e-05,
+      "loss": 0.4107,
+      "step": 3215
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8664518438954535e-05,
+      "loss": 0.3268,
+      "step": 3216
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86609380594343e-05,
+      "loss": 0.3698,
+      "step": 3217
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8657357679914074e-05,
+      "loss": 0.3206,
+      "step": 3218
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.865377730039385e-05,
+      "loss": 0.3174,
+      "step": 3219
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8650196920873614e-05,
+      "loss": 0.3726,
+      "step": 3220
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.864661654135339e-05,
+      "loss": 0.3398,
+      "step": 3221
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.864303616183316e-05,
+      "loss": 0.3952,
+      "step": 3222
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.863945578231293e-05,
+      "loss": 0.3974,
+      "step": 3223
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.86358754027927e-05,
+      "loss": 0.3626,
+      "step": 3224
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8632295023272466e-05,
+      "loss": 0.3063,
+      "step": 3225
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.862871464375224e-05,
+      "loss": 0.3902,
+      "step": 3226
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.862513426423201e-05,
+      "loss": 0.4232,
+      "step": 3227
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.862155388471178e-05,
+      "loss": 0.4704,
+      "step": 3228
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.861797350519155e-05,
+      "loss": 0.5062,
+      "step": 3229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.861439312567132e-05,
+      "loss": 0.5101,
+      "step": 3230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.861081274615109e-05,
+      "loss": 0.4619,
+      "step": 3231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8607232366630865e-05,
+      "loss": 0.3369,
+      "step": 3232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.860365198711063e-05,
+      "loss": 0.4944,
+      "step": 3233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8600071607590405e-05,
+      "loss": 0.442,
+      "step": 3234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.4772,
+      "step": 3235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.8592910848549944e-05,
+      "loss": 0.3822,
+      "step": 3236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.858933046902972e-05,
+      "loss": 0.3861,
+      "step": 3237
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.858575008950949e-05,
+      "loss": 0.4226,
+      "step": 3238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8582169709989264e-05,
+      "loss": 0.4972,
+      "step": 3239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.857858933046904e-05,
+      "loss": 0.4008,
+      "step": 3240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8575008950948804e-05,
+      "loss": 0.325,
+      "step": 3241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.857142857142858e-05,
+      "loss": 0.3513,
+      "step": 3242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.856784819190834e-05,
+      "loss": 0.44,
+      "step": 3243
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8564267812388116e-05,
+      "loss": 0.3965,
+      "step": 3244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.856068743286789e-05,
+      "loss": 0.4325,
+      "step": 3245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8557107053347656e-05,
+      "loss": 0.3904,
+      "step": 3246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.855352667382743e-05,
+      "loss": 0.3913,
+      "step": 3247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8549946294307196e-05,
+      "loss": 0.4406,
+      "step": 3248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.854636591478697e-05,
+      "loss": 0.3969,
+      "step": 3249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.854278553526674e-05,
+      "loss": 0.4065,
+      "step": 3250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.853920515574651e-05,
+      "loss": 0.3389,
+      "step": 3251
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.853562477622628e-05,
+      "loss": 0.42,
+      "step": 3252
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8532044396706055e-05,
+      "loss": 0.5479,
+      "step": 3253
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.852846401718582e-05,
+      "loss": 0.4323,
+      "step": 3254
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8524883637665594e-05,
+      "loss": 0.3899,
+      "step": 3255
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.852130325814536e-05,
+      "loss": 0.4423,
+      "step": 3256
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8517722878625134e-05,
+      "loss": 0.2562,
+      "step": 3257
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.851414249910491e-05,
+      "loss": 0.2978,
+      "step": 3258
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8510562119584674e-05,
+      "loss": 0.3759,
+      "step": 3259
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850698174006445e-05,
+      "loss": 0.3908,
+      "step": 3260
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.850340136054422e-05,
+      "loss": 0.5065,
+      "step": 3261
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.849982098102399e-05,
+      "loss": 0.3927,
+      "step": 3262
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8496240601503766e-05,
+      "loss": 0.5582,
+      "step": 3263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.849266022198353e-05,
+      "loss": 0.3966,
+      "step": 3264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.8489079842463306e-05,
+      "loss": 0.4774,
+      "step": 3265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.848549946294307e-05,
+      "loss": 0.2459,
+      "step": 3266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8481919083422845e-05,
+      "loss": 0.5249,
+      "step": 3267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.847833870390262e-05,
+      "loss": 0.3488,
+      "step": 3268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8474758324382385e-05,
+      "loss": 0.4966,
+      "step": 3269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.847117794486216e-05,
+      "loss": 0.422,
+      "step": 3270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.846759756534193e-05,
+      "loss": 0.4201,
+      "step": 3271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.84640171858217e-05,
+      "loss": 0.364,
+      "step": 3272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.846043680630147e-05,
+      "loss": 0.461,
+      "step": 3273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.845685642678124e-05,
+      "loss": 0.3568,
+      "step": 3274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.845327604726101e-05,
+      "loss": 0.4067,
+      "step": 3275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8449695667740784e-05,
+      "loss": 0.419,
+      "step": 3276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.844611528822055e-05,
+      "loss": 0.4136,
+      "step": 3277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8442534908700323e-05,
+      "loss": 0.4813,
+      "step": 3278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8438954529180097e-05,
+      "loss": 0.4121,
+      "step": 3279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.843537414965986e-05,
+      "loss": 0.4894,
+      "step": 3280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8431793770139636e-05,
+      "loss": 0.3374,
+      "step": 3281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.84282133906194e-05,
+      "loss": 0.5839,
+      "step": 3282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8424633011099176e-05,
+      "loss": 0.4602,
+      "step": 3283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.842105263157895e-05,
+      "loss": 0.3554,
+      "step": 3284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841747225205872e-05,
+      "loss": 0.579,
+      "step": 3285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8413891872538495e-05,
+      "loss": 0.4097,
+      "step": 3286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.841031149301826e-05,
+      "loss": 0.3354,
+      "step": 3287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8406731113498035e-05,
+      "loss": 0.3823,
+      "step": 3288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.840315073397781e-05,
+      "loss": 0.4639,
+      "step": 3289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8399570354457575e-05,
+      "loss": 0.4718,
+      "step": 3290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.839598997493735e-05,
+      "loss": 0.4054,
+      "step": 3291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.8392409595417114e-05,
+      "loss": 0.4534,
+      "step": 3292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.838882921589689e-05,
+      "loss": 0.4495,
+      "step": 3293
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.838524883637666e-05,
+      "loss": 0.5076,
+      "step": 3294
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.838166845685643e-05,
+      "loss": 0.4129,
+      "step": 3295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.83780880773362e-05,
+      "loss": 0.4754,
+      "step": 3296
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.837450769781597e-05,
+      "loss": 0.4415,
+      "step": 3297
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.837092731829574e-05,
+      "loss": 0.433,
+      "step": 3298
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.836734693877551e-05,
+      "loss": 0.3958,
+      "step": 3299
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.836376655925528e-05,
+      "loss": 0.3989,
+      "step": 3300
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.836018617973505e-05,
+      "loss": 0.3021,
+      "step": 3301
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8356605800214826e-05,
+      "loss": 0.3164,
+      "step": 3302
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.835302542069459e-05,
+      "loss": 0.4279,
+      "step": 3303
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8349445041174365e-05,
+      "loss": 0.4972,
+      "step": 3304
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.834586466165413e-05,
+      "loss": 0.4424,
+      "step": 3305
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8342284282133905e-05,
+      "loss": 0.4124,
+      "step": 3306
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.833870390261368e-05,
+      "loss": 0.3529,
+      "step": 3307
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.833512352309345e-05,
+      "loss": 0.4596,
+      "step": 3308
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8331543143573224e-05,
+      "loss": 0.3146,
+      "step": 3309
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.832796276405299e-05,
+      "loss": 0.3578,
+      "step": 3310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8324382384532764e-05,
+      "loss": 0.3434,
+      "step": 3311
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.832080200501254e-05,
+      "loss": 0.5249,
+      "step": 3312
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8317221625492304e-05,
+      "loss": 0.4385,
+      "step": 3313
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.831364124597208e-05,
+      "loss": 0.4629,
+      "step": 3314
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.831006086645185e-05,
+      "loss": 0.3235,
+      "step": 3315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8306480486931616e-05,
+      "loss": 0.4696,
+      "step": 3316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.830290010741139e-05,
+      "loss": 0.4621,
+      "step": 3317
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.8299319727891156e-05,
+      "loss": 0.3326,
+      "step": 3318
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.829573934837093e-05,
+      "loss": 0.3116,
+      "step": 3319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.82921589688507e-05,
+      "loss": 0.4901,
+      "step": 3320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.828857858933047e-05,
+      "loss": 0.3755,
+      "step": 3321
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.828499820981024e-05,
+      "loss": 0.3566,
+      "step": 3322
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.828141783029001e-05,
+      "loss": 0.2747,
+      "step": 3323
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.827783745076978e-05,
+      "loss": 0.4301,
+      "step": 3324
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8274257071249555e-05,
+      "loss": 0.3058,
+      "step": 3325
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.827067669172932e-05,
+      "loss": 0.3791,
+      "step": 3326
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8267096312209094e-05,
+      "loss": 0.4096,
+      "step": 3327
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.826351593268887e-05,
+      "loss": 0.4196,
+      "step": 3328
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8259935553168634e-05,
+      "loss": 0.4521,
+      "step": 3329
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.825635517364841e-05,
+      "loss": 0.3263,
+      "step": 3330
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8252774794128174e-05,
+      "loss": 0.4274,
+      "step": 3331
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8249194414607954e-05,
+      "loss": 0.5008,
+      "step": 3332
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.824561403508773e-05,
+      "loss": 0.3901,
+      "step": 3333
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.824203365556749e-05,
+      "loss": 0.3705,
+      "step": 3334
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8238453276047266e-05,
+      "loss": 0.3683,
+      "step": 3335
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.823487289652703e-05,
+      "loss": 0.4732,
+      "step": 3336
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8231292517006806e-05,
+      "loss": 0.4548,
+      "step": 3337
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.822771213748658e-05,
+      "loss": 0.3717,
+      "step": 3338
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8224131757966346e-05,
+      "loss": 0.4885,
+      "step": 3339
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.822055137844612e-05,
+      "loss": 0.3431,
+      "step": 3340
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8216970998925885e-05,
+      "loss": 0.3396,
+      "step": 3341
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.821339061940566e-05,
+      "loss": 0.3483,
+      "step": 3342
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820981023988543e-05,
+      "loss": 0.4446,
+      "step": 3343
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.82062298603652e-05,
+      "loss": 0.4669,
+      "step": 3344
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.820264948084497e-05,
+      "loss": 0.5113,
+      "step": 3345
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8199069101324744e-05,
+      "loss": 0.4238,
+      "step": 3346
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.819548872180451e-05,
+      "loss": 0.3529,
+      "step": 3347
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.8191908342284284e-05,
+      "loss": 0.5202,
+      "step": 3348
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.818832796276405e-05,
+      "loss": 0.3336,
+      "step": 3349
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8184747583243824e-05,
+      "loss": 0.3214,
+      "step": 3350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.81811672037236e-05,
+      "loss": 0.4369,
+      "step": 3351
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.817758682420336e-05,
+      "loss": 0.5251,
+      "step": 3352
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8174006444683136e-05,
+      "loss": 0.3724,
+      "step": 3353
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.817042606516291e-05,
+      "loss": 0.3131,
+      "step": 3354
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.816684568564268e-05,
+      "loss": 0.4065,
+      "step": 3355
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8163265306122456e-05,
+      "loss": 0.4084,
+      "step": 3356
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.815968492660222e-05,
+      "loss": 0.4007,
+      "step": 3357
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8156104547081995e-05,
+      "loss": 0.4182,
+      "step": 3358
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.815252416756176e-05,
+      "loss": 0.4095,
+      "step": 3359
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8148943788041535e-05,
+      "loss": 0.3554,
+      "step": 3360
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.814536340852131e-05,
+      "loss": 0.4752,
+      "step": 3361
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8141783029001075e-05,
+      "loss": 0.3775,
+      "step": 3362
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.813820264948085e-05,
+      "loss": 0.464,
+      "step": 3363
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.813462226996062e-05,
+      "loss": 0.4678,
+      "step": 3364
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.813104189044039e-05,
+      "loss": 0.3389,
+      "step": 3365
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.812746151092016e-05,
+      "loss": 0.3473,
+      "step": 3366
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.812388113139993e-05,
+      "loss": 0.3582,
+      "step": 3367
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.81203007518797e-05,
+      "loss": 0.3182,
+      "step": 3368
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8116720372359473e-05,
+      "loss": 0.4884,
+      "step": 3369
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.811313999283924e-05,
+      "loss": 0.318,
+      "step": 3370
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.810955961331901e-05,
+      "loss": 0.5169,
+      "step": 3371
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8105979233798786e-05,
+      "loss": 0.4078,
+      "step": 3372
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.810239885427855e-05,
+      "loss": 0.3817,
+      "step": 3373
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8098818474758326e-05,
+      "loss": 0.5431,
+      "step": 3374
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.4777,
+      "step": 3375
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.8091657715717865e-05,
+      "loss": 0.3095,
+      "step": 3376
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.808807733619764e-05,
+      "loss": 0.4427,
+      "step": 3377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.808449695667741e-05,
+      "loss": 0.4577,
+      "step": 3378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8080916577157185e-05,
+      "loss": 0.366,
+      "step": 3379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.807733619763695e-05,
+      "loss": 0.5404,
+      "step": 3380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8073755818116725e-05,
+      "loss": 0.4392,
+      "step": 3381
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.80701754385965e-05,
+      "loss": 0.5187,
+      "step": 3382
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8066595059076264e-05,
+      "loss": 0.5312,
+      "step": 3383
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.806301467955604e-05,
+      "loss": 0.3005,
+      "step": 3384
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8059434300035804e-05,
+      "loss": 0.3031,
+      "step": 3385
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.805585392051558e-05,
+      "loss": 0.3393,
+      "step": 3386
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.805227354099535e-05,
+      "loss": 0.298,
+      "step": 3387
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8048693161475117e-05,
+      "loss": 0.4831,
+      "step": 3388
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.804511278195489e-05,
+      "loss": 0.3633,
+      "step": 3389
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.804153240243466e-05,
+      "loss": 0.5336,
+      "step": 3390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.803795202291443e-05,
+      "loss": 0.4444,
+      "step": 3391
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.80343716433942e-05,
+      "loss": 0.4793,
+      "step": 3392
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.803079126387397e-05,
+      "loss": 0.3473,
+      "step": 3393
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.802721088435374e-05,
+      "loss": 0.3945,
+      "step": 3394
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8023630504833515e-05,
+      "loss": 0.4703,
+      "step": 3395
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.802005012531328e-05,
+      "loss": 0.4953,
+      "step": 3396
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8016469745793055e-05,
+      "loss": 0.3697,
+      "step": 3397
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.801288936627282e-05,
+      "loss": 0.4831,
+      "step": 3398
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8009308986752595e-05,
+      "loss": 0.4412,
+      "step": 3399
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.800572860723237e-05,
+      "loss": 0.4384,
+      "step": 3400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.8002148227712134e-05,
+      "loss": 0.4128,
+      "step": 3401
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7998567848191914e-05,
+      "loss": 0.4312,
+      "step": 3402
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.799498746867168e-05,
+      "loss": 0.3627,
+      "step": 3403
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.7991407089151454e-05,
+      "loss": 0.3812,
+      "step": 3404
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 3.798782670963123e-05,
+      "loss": 0.5031,
+      "step": 3405
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.798424633011099e-05,
+      "loss": 0.5182,
+      "step": 3406
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7980665950590766e-05,
+      "loss": 0.3482,
+      "step": 3407
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.797708557107054e-05,
+      "loss": 0.4314,
+      "step": 3408
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7973505191550306e-05,
+      "loss": 0.4382,
+      "step": 3409
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.796992481203008e-05,
+      "loss": 0.446,
+      "step": 3410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7966344432509846e-05,
+      "loss": 0.4468,
+      "step": 3411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.796276405298962e-05,
+      "loss": 0.4628,
+      "step": 3412
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.795918367346939e-05,
+      "loss": 0.3871,
+      "step": 3413
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.795560329394916e-05,
+      "loss": 0.3307,
+      "step": 3414
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.795202291442893e-05,
+      "loss": 0.4665,
+      "step": 3415
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.79484425349087e-05,
+      "loss": 0.3708,
+      "step": 3416
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.794486215538847e-05,
+      "loss": 0.3964,
+      "step": 3417
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7941281775868244e-05,
+      "loss": 0.3957,
+      "step": 3418
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.793770139634801e-05,
+      "loss": 0.4215,
+      "step": 3419
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7934121016827784e-05,
+      "loss": 0.4461,
+      "step": 3420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.793054063730756e-05,
+      "loss": 0.3529,
+      "step": 3421
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7926960257787324e-05,
+      "loss": 0.504,
+      "step": 3422
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.79233798782671e-05,
+      "loss": 0.3343,
+      "step": 3423
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.791979949874686e-05,
+      "loss": 0.3187,
+      "step": 3424
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.791621911922664e-05,
+      "loss": 0.4127,
+      "step": 3425
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7912638739706416e-05,
+      "loss": 0.3669,
+      "step": 3426
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.790905836018618e-05,
+      "loss": 0.2591,
+      "step": 3427
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7905477980665956e-05,
+      "loss": 0.4755,
+      "step": 3428
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.790189760114572e-05,
+      "loss": 0.3527,
+      "step": 3429
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7898317221625496e-05,
+      "loss": 0.4798,
+      "step": 3430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.789473684210527e-05,
+      "loss": 0.5459,
+      "step": 3431
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.7891156462585035e-05,
+      "loss": 0.4517,
+      "step": 3432
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.788757608306481e-05,
+      "loss": 0.4648,
+      "step": 3433
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7883995703544575e-05,
+      "loss": 0.4397,
+      "step": 3434
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.788041532402435e-05,
+      "loss": 0.4174,
+      "step": 3435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.787683494450412e-05,
+      "loss": 0.3275,
+      "step": 3436
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.787325456498389e-05,
+      "loss": 0.3972,
+      "step": 3437
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.786967418546366e-05,
+      "loss": 0.4135,
+      "step": 3438
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7866093805943434e-05,
+      "loss": 0.452,
+      "step": 3439
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.78625134264232e-05,
+      "loss": 0.4071,
+      "step": 3440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7858933046902974e-05,
+      "loss": 0.362,
+      "step": 3441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.785535266738274e-05,
+      "loss": 0.3887,
+      "step": 3442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.785177228786251e-05,
+      "loss": 0.4508,
+      "step": 3443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7848191908342286e-05,
+      "loss": 0.4284,
+      "step": 3444
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.784461152882205e-05,
+      "loss": 0.4435,
+      "step": 3445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7841031149301826e-05,
+      "loss": 0.303,
+      "step": 3446
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.78374507697816e-05,
+      "loss": 0.4662,
+      "step": 3447
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.783387039026137e-05,
+      "loss": 0.4589,
+      "step": 3448
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7830290010741145e-05,
+      "loss": 0.4251,
+      "step": 3449
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.782670963122091e-05,
+      "loss": 0.3464,
+      "step": 3450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7823129251700685e-05,
+      "loss": 0.3763,
+      "step": 3451
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.781954887218045e-05,
+      "loss": 0.348,
+      "step": 3452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7815968492660225e-05,
+      "loss": 0.4042,
+      "step": 3453
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.781238811314e-05,
+      "loss": 0.4242,
+      "step": 3454
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.7808807733619764e-05,
+      "loss": 0.4571,
+      "step": 3455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780522735409954e-05,
+      "loss": 0.3229,
+      "step": 3456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.780164697457931e-05,
+      "loss": 0.438,
+      "step": 3457
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779806659505908e-05,
+      "loss": 0.2404,
+      "step": 3458
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779448621553885e-05,
+      "loss": 0.3442,
+      "step": 3459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.779090583601862e-05,
+      "loss": 0.3604,
+      "step": 3460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.778732545649839e-05,
+      "loss": 0.4654,
+      "step": 3461
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.778374507697816e-05,
+      "loss": 0.2746,
+      "step": 3462
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.778016469745793e-05,
+      "loss": 0.4712,
+      "step": 3463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.77765843179377e-05,
+      "loss": 0.4103,
+      "step": 3464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7773003938417476e-05,
+      "loss": 0.3763,
+      "step": 3465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.776942355889724e-05,
+      "loss": 0.4783,
+      "step": 3466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7765843179377015e-05,
+      "loss": 0.3929,
+      "step": 3467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.776226279985678e-05,
+      "loss": 0.3934,
+      "step": 3468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7758682420336555e-05,
+      "loss": 0.3789,
+      "step": 3469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.775510204081633e-05,
+      "loss": 0.4667,
+      "step": 3470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7751521661296095e-05,
+      "loss": 0.3816,
+      "step": 3471
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7747941281775875e-05,
+      "loss": 0.469,
+      "step": 3472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.774436090225564e-05,
+      "loss": 0.5766,
+      "step": 3473
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7740780522735414e-05,
+      "loss": 0.4373,
+      "step": 3474
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.773720014321519e-05,
+      "loss": 0.3484,
+      "step": 3475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7733619763694954e-05,
+      "loss": 0.3739,
+      "step": 3476
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.773003938417473e-05,
+      "loss": 0.4988,
+      "step": 3477
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.772645900465449e-05,
+      "loss": 0.4647,
+      "step": 3478
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7722878625134267e-05,
+      "loss": 0.4658,
+      "step": 3479
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.771929824561404e-05,
+      "loss": 0.4425,
+      "step": 3480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7715717866093806e-05,
+      "loss": 0.2208,
+      "step": 3481
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.771213748657358e-05,
+      "loss": 0.3003,
+      "step": 3482
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770855710705335e-05,
+      "loss": 0.4219,
+      "step": 3483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770497672753312e-05,
+      "loss": 0.4344,
+      "step": 3484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.770139634801289e-05,
+      "loss": 0.5045,
+      "step": 3485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.769781596849266e-05,
+      "loss": 0.394,
+      "step": 3486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.769423558897243e-05,
+      "loss": 0.4235,
+      "step": 3487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.7690655209452205e-05,
+      "loss": 0.4263,
+      "step": 3488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 3.768707482993197e-05,
+      "loss": 0.3862,
+      "step": 3489
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7683494450411745e-05,
+      "loss": 0.3264,
+      "step": 3490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.767991407089152e-05,
+      "loss": 0.3681,
+      "step": 3491
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7676333691371284e-05,
+      "loss": 0.5699,
+      "step": 3492
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.767275331185106e-05,
+      "loss": 0.3187,
+      "step": 3493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7669172932330824e-05,
+      "loss": 0.417,
+      "step": 3494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7665592552810604e-05,
+      "loss": 0.3672,
+      "step": 3495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.766201217329037e-05,
+      "loss": 0.4738,
+      "step": 3496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.765843179377014e-05,
+      "loss": 0.4338,
+      "step": 3497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7654851414249916e-05,
+      "loss": 0.4394,
+      "step": 3498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.765127103472968e-05,
+      "loss": 0.4391,
+      "step": 3499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7647690655209456e-05,
+      "loss": 0.4076,
+      "step": 3500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.764411027568923e-05,
+      "loss": 0.4131,
+      "step": 3501
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7640529896168996e-05,
+      "loss": 0.3963,
+      "step": 3502
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.763694951664877e-05,
+      "loss": 0.3942,
+      "step": 3503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7633369137128535e-05,
+      "loss": 0.3964,
+      "step": 3504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.762978875760831e-05,
+      "loss": 0.3408,
+      "step": 3505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.762620837808808e-05,
+      "loss": 0.5589,
+      "step": 3506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.762262799856785e-05,
+      "loss": 0.3662,
+      "step": 3507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.761904761904762e-05,
+      "loss": 0.4305,
+      "step": 3508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7615467239527394e-05,
+      "loss": 0.3576,
+      "step": 3509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.761188686000716e-05,
+      "loss": 0.4502,
+      "step": 3510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7608306480486934e-05,
+      "loss": 0.4446,
+      "step": 3511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.76047261009667e-05,
+      "loss": 0.4019,
+      "step": 3512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7601145721446474e-05,
+      "loss": 0.3516,
+      "step": 3513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.759756534192625e-05,
+      "loss": 0.3723,
+      "step": 3514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.759398496240601e-05,
+      "loss": 0.2936,
+      "step": 3515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.7590404582885786e-05,
+      "loss": 0.3606,
+      "step": 3516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.758682420336555e-05,
+      "loss": 0.4346,
+      "step": 3517
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.758324382384533e-05,
+      "loss": 0.3495,
+      "step": 3518
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7579663444325106e-05,
+      "loss": 0.3396,
+      "step": 3519
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.757608306480487e-05,
+      "loss": 0.4694,
+      "step": 3520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7572502685284646e-05,
+      "loss": 0.5375,
+      "step": 3521
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.756892230576441e-05,
+      "loss": 0.4426,
+      "step": 3522
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7565341926244185e-05,
+      "loss": 0.5406,
+      "step": 3523
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.756176154672396e-05,
+      "loss": 0.3928,
+      "step": 3524
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7558181167203725e-05,
+      "loss": 0.4578,
+      "step": 3525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.75546007876835e-05,
+      "loss": 0.4187,
+      "step": 3526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7551020408163264e-05,
+      "loss": 0.4029,
+      "step": 3527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.754744002864304e-05,
+      "loss": 0.4821,
+      "step": 3528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.754385964912281e-05,
+      "loss": 0.4751,
+      "step": 3529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.754027926960258e-05,
+      "loss": 0.3928,
+      "step": 3530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.753669889008235e-05,
+      "loss": 0.4276,
+      "step": 3531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7533118510562124e-05,
+      "loss": 0.3504,
+      "step": 3532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.752953813104189e-05,
+      "loss": 0.4605,
+      "step": 3533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.752595775152166e-05,
+      "loss": 0.3324,
+      "step": 3534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.752237737200143e-05,
+      "loss": 0.3307,
+      "step": 3535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.75187969924812e-05,
+      "loss": 0.4228,
+      "step": 3536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7515216612960976e-05,
+      "loss": 0.3619,
+      "step": 3537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.751163623344074e-05,
+      "loss": 0.3658,
+      "step": 3538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7508055853920516e-05,
+      "loss": 0.4555,
+      "step": 3539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.750447547440029e-05,
+      "loss": 0.2909,
+      "step": 3540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7500895094880055e-05,
+      "loss": 0.3685,
+      "step": 3541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7497314715359835e-05,
+      "loss": 0.4279,
+      "step": 3542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.74937343358396e-05,
+      "loss": 0.4741,
+      "step": 3543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.7490153956319375e-05,
+      "loss": 0.3947,
+      "step": 3544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 3.748657357679914e-05,
+      "loss": 0.5308,
+      "step": 3545
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7482993197278914e-05,
+      "loss": 0.3818,
+      "step": 3546
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.747941281775869e-05,
+      "loss": 0.5104,
+      "step": 3547
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7475832438238454e-05,
+      "loss": 0.4143,
+      "step": 3548
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.747225205871823e-05,
+      "loss": 0.3262,
+      "step": 3549
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7468671679198e-05,
+      "loss": 0.3948,
+      "step": 3550
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.746509129967777e-05,
+      "loss": 0.4495,
+      "step": 3551
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.746151092015754e-05,
+      "loss": 0.3879,
+      "step": 3552
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7457930540637306e-05,
+      "loss": 0.406,
+      "step": 3553
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.745435016111708e-05,
+      "loss": 0.446,
+      "step": 3554
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.745076978159685e-05,
+      "loss": 0.402,
+      "step": 3555
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.744718940207662e-05,
+      "loss": 0.3587,
+      "step": 3556
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.744360902255639e-05,
+      "loss": 0.3948,
+      "step": 3557
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7440028643036165e-05,
+      "loss": 0.5205,
+      "step": 3558
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.743644826351593e-05,
+      "loss": 0.3877,
+      "step": 3559
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7432867883995705e-05,
+      "loss": 0.3885,
+      "step": 3560
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.742928750447547e-05,
+      "loss": 0.416,
+      "step": 3561
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7425707124955245e-05,
+      "loss": 0.4306,
+      "step": 3562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.742212674543502e-05,
+      "loss": 0.4707,
+      "step": 3563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7418546365914784e-05,
+      "loss": 0.3668,
+      "step": 3564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7414965986394564e-05,
+      "loss": 0.4938,
+      "step": 3565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.741138560687433e-05,
+      "loss": 0.3521,
+      "step": 3566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7407805227354104e-05,
+      "loss": 0.3865,
+      "step": 3567
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.740422484783388e-05,
+      "loss": 0.3365,
+      "step": 3568
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.740064446831364e-05,
+      "loss": 0.4476,
+      "step": 3569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7397064088793417e-05,
+      "loss": 0.404,
+      "step": 3570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.739348370927318e-05,
+      "loss": 0.4158,
+      "step": 3571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.7389903329752956e-05,
+      "loss": 0.3498,
+      "step": 3572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.738632295023273e-05,
+      "loss": 0.4698,
+      "step": 3573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7382742570712496e-05,
+      "loss": 0.4418,
+      "step": 3574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.737916219119227e-05,
+      "loss": 0.3772,
+      "step": 3575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.737558181167204e-05,
+      "loss": 0.3423,
+      "step": 3576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.737200143215181e-05,
+      "loss": 0.3366,
+      "step": 3577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.736842105263158e-05,
+      "loss": 0.395,
+      "step": 3578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.736484067311135e-05,
+      "loss": 0.423,
+      "step": 3579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.736126029359112e-05,
+      "loss": 0.2972,
+      "step": 3580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7357679914070895e-05,
+      "loss": 0.4796,
+      "step": 3581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.735409953455066e-05,
+      "loss": 0.3511,
+      "step": 3582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7350519155030434e-05,
+      "loss": 0.4249,
+      "step": 3583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.734693877551021e-05,
+      "loss": 0.4333,
+      "step": 3584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7343358395989974e-05,
+      "loss": 0.3493,
+      "step": 3585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.733977801646975e-05,
+      "loss": 0.3051,
+      "step": 3586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.733619763694951e-05,
+      "loss": 0.5631,
+      "step": 3587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.733261725742929e-05,
+      "loss": 0.4372,
+      "step": 3588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.732903687790906e-05,
+      "loss": 0.4353,
+      "step": 3589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.732545649838883e-05,
+      "loss": 0.4012,
+      "step": 3590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7321876118868606e-05,
+      "loss": 0.5074,
+      "step": 3591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.731829573934837e-05,
+      "loss": 0.3395,
+      "step": 3592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7314715359828146e-05,
+      "loss": 0.4572,
+      "step": 3593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.731113498030792e-05,
+      "loss": 0.2439,
+      "step": 3594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7307554600787685e-05,
+      "loss": 0.3313,
+      "step": 3595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.730397422126746e-05,
+      "loss": 0.4678,
+      "step": 3596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7300393841747225e-05,
+      "loss": 0.3841,
+      "step": 3597
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.7296813462227e-05,
+      "loss": 0.4121,
+      "step": 3598
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.729323308270677e-05,
+      "loss": 0.3717,
+      "step": 3599
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728965270318654e-05,
+      "loss": 0.3985,
+      "step": 3600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 3.728607232366631e-05,
+      "loss": 0.3352,
+      "step": 3601
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7282491944146084e-05,
+      "loss": 0.3578,
+      "step": 3602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.727891156462585e-05,
+      "loss": 0.3224,
+      "step": 3603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7275331185105624e-05,
+      "loss": 0.3728,
+      "step": 3604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.727175080558539e-05,
+      "loss": 0.3378,
+      "step": 3605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.726817042606516e-05,
+      "loss": 0.5375,
+      "step": 3606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7264590046544936e-05,
+      "loss": 0.3909,
+      "step": 3607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.72610096670247e-05,
+      "loss": 0.3876,
+      "step": 3608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7257429287504476e-05,
+      "loss": 0.3917,
+      "step": 3609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.725384890798424e-05,
+      "loss": 0.343,
+      "step": 3610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7250268528464016e-05,
+      "loss": 0.3774,
+      "step": 3611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7246688148943796e-05,
+      "loss": 0.2827,
+      "step": 3612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.724310776942356e-05,
+      "loss": 0.5204,
+      "step": 3613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7239527389903335e-05,
+      "loss": 0.3212,
+      "step": 3614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.72359470103831e-05,
+      "loss": 0.5228,
+      "step": 3615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7232366630862875e-05,
+      "loss": 0.5728,
+      "step": 3616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.722878625134265e-05,
+      "loss": 0.404,
+      "step": 3617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.7225205871822414e-05,
+      "loss": 0.2913,
+      "step": 3618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.722162549230219e-05,
+      "loss": 0.478,
+      "step": 3619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721804511278196e-05,
+      "loss": 0.3668,
+      "step": 3620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.721446473326173e-05,
+      "loss": 0.3747,
+      "step": 3621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.72108843537415e-05,
+      "loss": 0.3285,
+      "step": 3622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720730397422127e-05,
+      "loss": 0.492,
+      "step": 3623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720372359470104e-05,
+      "loss": 0.3792,
+      "step": 3624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.720014321518081e-05,
+      "loss": 0.3995,
+      "step": 3625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.719656283566058e-05,
+      "loss": 0.2738,
+      "step": 3626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.719298245614035e-05,
+      "loss": 0.4803,
+      "step": 3627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.718940207662012e-05,
+      "loss": 0.3428,
+      "step": 3628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.718582169709989e-05,
+      "loss": 0.4899,
+      "step": 3629
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7182241317579666e-05,
+      "loss": 0.5003,
+      "step": 3630
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.717866093805943e-05,
+      "loss": 0.3922,
+      "step": 3631
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7175080558539205e-05,
+      "loss": 0.3319,
+      "step": 3632
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.717150017901898e-05,
+      "loss": 0.302,
+      "step": 3633
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7167919799498745e-05,
+      "loss": 0.4691,
+      "step": 3634
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7164339419978525e-05,
+      "loss": 0.4088,
+      "step": 3635
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.716075904045829e-05,
+      "loss": 0.3062,
+      "step": 3636
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7157178660938064e-05,
+      "loss": 0.3449,
+      "step": 3637
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.715359828141783e-05,
+      "loss": 0.5088,
+      "step": 3638
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7150017901897604e-05,
+      "loss": 0.4772,
+      "step": 3639
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.714643752237738e-05,
+      "loss": 0.4009,
+      "step": 3640
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7142857142857143e-05,
+      "loss": 0.3543,
+      "step": 3641
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.713927676333692e-05,
+      "loss": 0.2934,
+      "step": 3642
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.713569638381669e-05,
+      "loss": 0.418,
+      "step": 3643
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7132116004296456e-05,
+      "loss": 0.5215,
+      "step": 3644
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.712853562477623e-05,
+      "loss": 0.3071,
+      "step": 3645
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7124955245255996e-05,
+      "loss": 0.4259,
+      "step": 3646
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.712137486573577e-05,
+      "loss": 0.3323,
+      "step": 3647
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.711779448621554e-05,
+      "loss": 0.3756,
+      "step": 3648
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.711421410669531e-05,
+      "loss": 0.4725,
+      "step": 3649
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.711063372717508e-05,
+      "loss": 0.3356,
+      "step": 3650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7107053347654855e-05,
+      "loss": 0.3931,
+      "step": 3651
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.710347296813462e-05,
+      "loss": 0.398,
+      "step": 3652
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7099892588614395e-05,
+      "loss": 0.2207,
+      "step": 3653
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.709631220909416e-05,
+      "loss": 0.371,
+      "step": 3654
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7092731829573934e-05,
+      "loss": 0.3616,
+      "step": 3655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.708915145005371e-05,
+      "loss": 0.4727,
+      "step": 3656
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 3.7085571070533474e-05,
+      "loss": 0.4819,
+      "step": 3657
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.708199069101325e-05,
+      "loss": 0.2781,
+      "step": 3658
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.707841031149302e-05,
+      "loss": 0.275,
+      "step": 3659
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.707482993197279e-05,
+      "loss": 0.3563,
+      "step": 3660
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7071249552452567e-05,
+      "loss": 0.3198,
+      "step": 3661
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.706766917293233e-05,
+      "loss": 0.5788,
+      "step": 3662
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7064088793412106e-05,
+      "loss": 0.3973,
+      "step": 3663
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.706050841389187e-05,
+      "loss": 0.405,
+      "step": 3664
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7056928034371646e-05,
+      "loss": 0.4324,
+      "step": 3665
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.705334765485142e-05,
+      "loss": 0.4963,
+      "step": 3666
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7049767275331185e-05,
+      "loss": 0.4405,
+      "step": 3667
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.704618689581096e-05,
+      "loss": 0.4254,
+      "step": 3668
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.704260651629073e-05,
+      "loss": 0.4896,
+      "step": 3669
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.70390261367705e-05,
+      "loss": 0.3205,
+      "step": 3670
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.703544575725027e-05,
+      "loss": 0.4559,
+      "step": 3671
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.703186537773004e-05,
+      "loss": 0.3951,
+      "step": 3672
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.702828499820981e-05,
+      "loss": 0.446,
+      "step": 3673
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7024704618689584e-05,
+      "loss": 0.4499,
+      "step": 3674
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.702112423916935e-05,
+      "loss": 0.4484,
+      "step": 3675
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7017543859649124e-05,
+      "loss": 0.3609,
+      "step": 3676
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.70139634801289e-05,
+      "loss": 0.3517,
+      "step": 3677
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.701038310060866e-05,
+      "loss": 0.4513,
+      "step": 3678
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.7006802721088437e-05,
+      "loss": 0.4293,
+      "step": 3679
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.70032223415682e-05,
+      "loss": 0.4797,
+      "step": 3680
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6999641962047976e-05,
+      "loss": 0.6419,
+      "step": 3681
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699606158252775e-05,
+      "loss": 0.372,
+      "step": 3682
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.699248120300752e-05,
+      "loss": 0.4464,
+      "step": 3683
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.6988900823487296e-05,
+      "loss": 0.3428,
+      "step": 3684
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.698532044396706e-05,
+      "loss": 0.3586,
+      "step": 3685
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6981740064446835e-05,
+      "loss": 0.3388,
+      "step": 3686
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.697815968492661e-05,
+      "loss": 0.444,
+      "step": 3687
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6974579305406375e-05,
+      "loss": 0.3638,
+      "step": 3688
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.697099892588615e-05,
+      "loss": 0.4062,
+      "step": 3689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6967418546365914e-05,
+      "loss": 0.384,
+      "step": 3690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.696383816684569e-05,
+      "loss": 0.3931,
+      "step": 3691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.696025778732546e-05,
+      "loss": 0.3348,
+      "step": 3692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.695667740780523e-05,
+      "loss": 0.3676,
+      "step": 3693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6953097028285e-05,
+      "loss": 0.3843,
+      "step": 3694
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6949516648764774e-05,
+      "loss": 0.4084,
+      "step": 3695
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.694593626924454e-05,
+      "loss": 0.4736,
+      "step": 3696
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.694235588972431e-05,
+      "loss": 0.3253,
+      "step": 3697
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.693877551020408e-05,
+      "loss": 0.3297,
+      "step": 3698
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.693519513068385e-05,
+      "loss": 0.2494,
+      "step": 3699
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6931614751163626e-05,
+      "loss": 0.2804,
+      "step": 3700
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.692803437164339e-05,
+      "loss": 0.6379,
+      "step": 3701
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6924453992123166e-05,
+      "loss": 0.3571,
+      "step": 3702
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.692087361260293e-05,
+      "loss": 0.459,
+      "step": 3703
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6917293233082705e-05,
+      "loss": 0.3075,
+      "step": 3704
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6913712853562485e-05,
+      "loss": 0.3763,
+      "step": 3705
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.691013247404225e-05,
+      "loss": 0.6304,
+      "step": 3706
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6906552094522025e-05,
+      "loss": 0.4017,
+      "step": 3707
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.690297171500179e-05,
+      "loss": 0.4742,
+      "step": 3708
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6899391335481564e-05,
+      "loss": 0.4171,
+      "step": 3709
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.689581095596134e-05,
+      "loss": 0.4046,
+      "step": 3710
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.6892230576441104e-05,
+      "loss": 0.3181,
+      "step": 3711
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688865019692088e-05,
+      "loss": 0.4916,
+      "step": 3712
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.688506981740065e-05,
+      "loss": 0.3264,
+      "step": 3713
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.688148943788042e-05,
+      "loss": 0.395,
+      "step": 3714
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.687790905836019e-05,
+      "loss": 0.468,
+      "step": 3715
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6874328678839956e-05,
+      "loss": 0.2758,
+      "step": 3716
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.687074829931973e-05,
+      "loss": 0.4254,
+      "step": 3717
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.68671679197995e-05,
+      "loss": 0.381,
+      "step": 3718
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.686358754027927e-05,
+      "loss": 0.375,
+      "step": 3719
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.686000716075904e-05,
+      "loss": 0.3793,
+      "step": 3720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.685642678123881e-05,
+      "loss": 0.4705,
+      "step": 3721
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.685284640171858e-05,
+      "loss": 0.3941,
+      "step": 3722
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6849266022198355e-05,
+      "loss": 0.4521,
+      "step": 3723
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.684568564267812e-05,
+      "loss": 0.3538,
+      "step": 3724
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.4011,
+      "step": 3725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.683852488363767e-05,
+      "loss": 0.5161,
+      "step": 3726
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6834944504117434e-05,
+      "loss": 0.4261,
+      "step": 3727
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.683136412459721e-05,
+      "loss": 0.3712,
+      "step": 3728
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.682778374507698e-05,
+      "loss": 0.3106,
+      "step": 3729
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6824203365556754e-05,
+      "loss": 0.4342,
+      "step": 3730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.682062298603653e-05,
+      "loss": 0.3373,
+      "step": 3731
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6817042606516293e-05,
+      "loss": 0.5482,
+      "step": 3732
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.681346222699607e-05,
+      "loss": 0.5555,
+      "step": 3733
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.680988184747583e-05,
+      "loss": 0.3556,
+      "step": 3734
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6806301467955606e-05,
+      "loss": 0.4559,
+      "step": 3735
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.680272108843538e-05,
+      "loss": 0.3519,
+      "step": 3736
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6799140708915146e-05,
+      "loss": 0.4019,
+      "step": 3737
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.679556032939492e-05,
+      "loss": 0.3909,
+      "step": 3738
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6791979949874685e-05,
+      "loss": 0.4723,
+      "step": 3739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.678839957035446e-05,
+      "loss": 0.48,
+      "step": 3740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.678481919083423e-05,
+      "loss": 0.3543,
+      "step": 3741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 3.6781238811314e-05,
+      "loss": 0.4409,
+      "step": 3742
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.677765843179377e-05,
+      "loss": 0.392,
+      "step": 3743
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6774078052273545e-05,
+      "loss": 0.3966,
+      "step": 3744
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.677049767275331e-05,
+      "loss": 0.2378,
+      "step": 3745
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6766917293233084e-05,
+      "loss": 0.3719,
+      "step": 3746
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.676333691371285e-05,
+      "loss": 0.4091,
+      "step": 3747
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6759756534192624e-05,
+      "loss": 0.366,
+      "step": 3748
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.67561761546724e-05,
+      "loss": 0.4259,
+      "step": 3749
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6752595775152163e-05,
+      "loss": 0.5302,
+      "step": 3750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6749015395631937e-05,
+      "loss": 0.5688,
+      "step": 3751
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.674543501611171e-05,
+      "loss": 0.4548,
+      "step": 3752
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.674185463659148e-05,
+      "loss": 0.3706,
+      "step": 3753
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6738274257071256e-05,
+      "loss": 0.3754,
+      "step": 3754
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.673469387755102e-05,
+      "loss": 0.3529,
+      "step": 3755
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6731113498030796e-05,
+      "loss": 0.3109,
+      "step": 3756
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.672753311851056e-05,
+      "loss": 0.2379,
+      "step": 3757
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6723952738990335e-05,
+      "loss": 0.4057,
+      "step": 3758
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.672037235947011e-05,
+      "loss": 0.3664,
+      "step": 3759
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6716791979949875e-05,
+      "loss": 0.4234,
+      "step": 3760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.671321160042965e-05,
+      "loss": 0.4819,
+      "step": 3761
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670963122090942e-05,
+      "loss": 0.4521,
+      "step": 3762
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670605084138919e-05,
+      "loss": 0.3419,
+      "step": 3763
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.670247046186896e-05,
+      "loss": 0.3123,
+      "step": 3764
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.669889008234873e-05,
+      "loss": 0.2555,
+      "step": 3765
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.66953097028285e-05,
+      "loss": 0.3145,
+      "step": 3766
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6691729323308274e-05,
+      "loss": 0.3974,
+      "step": 3767
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.668814894378804e-05,
+      "loss": 0.3993,
+      "step": 3768
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.668456856426781e-05,
+      "loss": 0.445,
+      "step": 3769
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.6680988184747587e-05,
+      "loss": 0.48,
+      "step": 3770
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.667740780522735e-05,
+      "loss": 0.2865,
+      "step": 3771
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6673827425707126e-05,
+      "loss": 0.3106,
+      "step": 3772
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.667024704618689e-05,
+      "loss": 0.3853,
+      "step": 3773
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6666666666666666e-05,
+      "loss": 0.4461,
+      "step": 3774
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.666308628714644e-05,
+      "loss": 0.4665,
+      "step": 3775
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.665950590762621e-05,
+      "loss": 0.2941,
+      "step": 3776
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6655925528105985e-05,
+      "loss": 0.3057,
+      "step": 3777
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.665234514858575e-05,
+      "loss": 0.5154,
+      "step": 3778
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6648764769065525e-05,
+      "loss": 0.2929,
+      "step": 3779
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.66451843895453e-05,
+      "loss": 0.3636,
+      "step": 3780
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6641604010025064e-05,
+      "loss": 0.4015,
+      "step": 3781
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.663802363050484e-05,
+      "loss": 0.3974,
+      "step": 3782
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6634443250984604e-05,
+      "loss": 0.4341,
+      "step": 3783
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.663086287146438e-05,
+      "loss": 0.4389,
+      "step": 3784
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.662728249194415e-05,
+      "loss": 0.3732,
+      "step": 3785
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.662370211242392e-05,
+      "loss": 0.4326,
+      "step": 3786
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.662012173290369e-05,
+      "loss": 0.3178,
+      "step": 3787
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.661654135338346e-05,
+      "loss": 0.385,
+      "step": 3788
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.661296097386323e-05,
+      "loss": 0.4258,
+      "step": 3789
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6609380594343e-05,
+      "loss": 0.3299,
+      "step": 3790
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.660580021482277e-05,
+      "loss": 0.5319,
+      "step": 3791
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.660221983530254e-05,
+      "loss": 0.429,
+      "step": 3792
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6598639455782316e-05,
+      "loss": 0.2967,
+      "step": 3793
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.659505907626208e-05,
+      "loss": 0.464,
+      "step": 3794
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6591478696741855e-05,
+      "loss": 0.2767,
+      "step": 3795
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658789831722162e-05,
+      "loss": 0.5323,
+      "step": 3796
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.6584317937701395e-05,
+      "loss": 0.3703,
+      "step": 3797
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 3.658073755818117e-05,
+      "loss": 0.3783,
+      "step": 3798
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.657715717866094e-05,
+      "loss": 0.4327,
+      "step": 3799
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6573576799140714e-05,
+      "loss": 0.3169,
+      "step": 3800
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.656999641962048e-05,
+      "loss": 0.4295,
+      "step": 3801
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6566416040100254e-05,
+      "loss": 0.3348,
+      "step": 3802
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.656283566058003e-05,
+      "loss": 0.4095,
+      "step": 3803
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6559255281059794e-05,
+      "loss": 0.4687,
+      "step": 3804
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.655567490153957e-05,
+      "loss": 0.4239,
+      "step": 3805
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.655209452201934e-05,
+      "loss": 0.3441,
+      "step": 3806
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6548514142499106e-05,
+      "loss": 0.4825,
+      "step": 3807
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.654493376297888e-05,
+      "loss": 0.3604,
+      "step": 3808
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6541353383458646e-05,
+      "loss": 0.339,
+      "step": 3809
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.653777300393842e-05,
+      "loss": 0.3915,
+      "step": 3810
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.653419262441819e-05,
+      "loss": 0.2865,
+      "step": 3811
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.653061224489796e-05,
+      "loss": 0.3908,
+      "step": 3812
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.652703186537773e-05,
+      "loss": 0.4835,
+      "step": 3813
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.65234514858575e-05,
+      "loss": 0.3981,
+      "step": 3814
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.651987110633727e-05,
+      "loss": 0.3813,
+      "step": 3815
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6516290726817045e-05,
+      "loss": 0.3623,
+      "step": 3816
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.651271034729681e-05,
+      "loss": 0.3907,
+      "step": 3817
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6509129967776584e-05,
+      "loss": 0.3479,
+      "step": 3818
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.650554958825636e-05,
+      "loss": 0.4197,
+      "step": 3819
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6501969208736124e-05,
+      "loss": 0.3382,
+      "step": 3820
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.64983888292159e-05,
+      "loss": 0.3128,
+      "step": 3821
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.649480844969567e-05,
+      "loss": 0.3593,
+      "step": 3822
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6491228070175443e-05,
+      "loss": 0.4741,
+      "step": 3823
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648764769065522e-05,
+      "loss": 0.4931,
+      "step": 3824
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.648406731113498e-05,
+      "loss": 0.4269,
+      "step": 3825
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.6480486931614756e-05,
+      "loss": 0.3483,
+      "step": 3826
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.647690655209452e-05,
+      "loss": 0.391,
+      "step": 3827
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6473326172574296e-05,
+      "loss": 0.4448,
+      "step": 3828
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646974579305407e-05,
+      "loss": 0.315,
+      "step": 3829
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6466165413533835e-05,
+      "loss": 0.4109,
+      "step": 3830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.646258503401361e-05,
+      "loss": 0.2405,
+      "step": 3831
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6459004654493375e-05,
+      "loss": 0.4198,
+      "step": 3832
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.645542427497315e-05,
+      "loss": 0.2659,
+      "step": 3833
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.645184389545292e-05,
+      "loss": 0.5473,
+      "step": 3834
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.644826351593269e-05,
+      "loss": 0.4541,
+      "step": 3835
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.644468313641246e-05,
+      "loss": 0.3627,
+      "step": 3836
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6441102756892234e-05,
+      "loss": 0.2564,
+      "step": 3837
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6437522377372e-05,
+      "loss": 0.3229,
+      "step": 3838
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6433941997851774e-05,
+      "loss": 0.4753,
+      "step": 3839
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.643036161833154e-05,
+      "loss": 0.4072,
+      "step": 3840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6426781238811313e-05,
+      "loss": 0.28,
+      "step": 3841
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.642320085929109e-05,
+      "loss": 0.4284,
+      "step": 3842
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.641962047977085e-05,
+      "loss": 0.3827,
+      "step": 3843
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6416040100250626e-05,
+      "loss": 0.2877,
+      "step": 3844
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.64124597207304e-05,
+      "loss": 0.3866,
+      "step": 3845
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640887934121017e-05,
+      "loss": 0.3407,
+      "step": 3846
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6405298961689946e-05,
+      "loss": 0.4169,
+      "step": 3847
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.640171858216971e-05,
+      "loss": 0.5063,
+      "step": 3848
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6398138202649485e-05,
+      "loss": 0.3664,
+      "step": 3849
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.639455782312925e-05,
+      "loss": 0.4443,
+      "step": 3850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6390977443609025e-05,
+      "loss": 0.4019,
+      "step": 3851
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.63873970640888e-05,
+      "loss": 0.4708,
+      "step": 3852
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.6383816684568565e-05,
+      "loss": 0.4316,
+      "step": 3853
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 3.638023630504834e-05,
+      "loss": 0.4431,
+      "step": 3854
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637665592552811e-05,
+      "loss": 0.3798,
+      "step": 3855
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.637307554600788e-05,
+      "loss": 0.3312,
+      "step": 3856
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.636949516648765e-05,
+      "loss": 0.4447,
+      "step": 3857
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.636591478696742e-05,
+      "loss": 0.4875,
+      "step": 3858
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.636233440744719e-05,
+      "loss": 0.3833,
+      "step": 3859
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.635875402792696e-05,
+      "loss": 0.3044,
+      "step": 3860
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.635517364840673e-05,
+      "loss": 0.3809,
+      "step": 3861
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.63515932688865e-05,
+      "loss": 0.5122,
+      "step": 3862
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6348012889366276e-05,
+      "loss": 0.619,
+      "step": 3863
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.634443250984604e-05,
+      "loss": 0.4056,
+      "step": 3864
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6340852130325816e-05,
+      "loss": 0.343,
+      "step": 3865
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.633727175080558e-05,
+      "loss": 0.4534,
+      "step": 3866
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6333691371285355e-05,
+      "loss": 0.4284,
+      "step": 3867
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.633011099176513e-05,
+      "loss": 0.3542,
+      "step": 3868
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.63265306122449e-05,
+      "loss": 0.2988,
+      "step": 3869
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6322950232724675e-05,
+      "loss": 0.3447,
+      "step": 3870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.631936985320444e-05,
+      "loss": 0.3437,
+      "step": 3871
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6315789473684214e-05,
+      "loss": 0.3316,
+      "step": 3872
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.631220909416399e-05,
+      "loss": 0.34,
+      "step": 3873
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6308628714643754e-05,
+      "loss": 0.3607,
+      "step": 3874
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.630504833512353e-05,
+      "loss": 0.451,
+      "step": 3875
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6301467955603294e-05,
+      "loss": 0.4361,
+      "step": 3876
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.629788757608307e-05,
+      "loss": 0.3714,
+      "step": 3877
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.629430719656284e-05,
+      "loss": 0.3937,
+      "step": 3878
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.6290726817042606e-05,
+      "loss": 0.3218,
+      "step": 3879
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.628714643752238e-05,
+      "loss": 0.468,
+      "step": 3880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.628356605800215e-05,
+      "loss": 0.4596,
+      "step": 3881
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.627998567848192e-05,
+      "loss": 0.4626,
+      "step": 3882
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.627640529896169e-05,
+      "loss": 0.3335,
+      "step": 3883
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.627282491944146e-05,
+      "loss": 0.2678,
+      "step": 3884
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.626924453992123e-05,
+      "loss": 0.4248,
+      "step": 3885
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6265664160401005e-05,
+      "loss": 0.4809,
+      "step": 3886
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.626208378088077e-05,
+      "loss": 0.3752,
+      "step": 3887
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6258503401360545e-05,
+      "loss": 0.4396,
+      "step": 3888
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.625492302184032e-05,
+      "loss": 0.4159,
+      "step": 3889
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6251342642320084e-05,
+      "loss": 0.4831,
+      "step": 3890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624776226279986e-05,
+      "loss": 0.338,
+      "step": 3891
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.624418188327963e-05,
+      "loss": 0.4745,
+      "step": 3892
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6240601503759404e-05,
+      "loss": 0.452,
+      "step": 3893
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.623702112423917e-05,
+      "loss": 0.389,
+      "step": 3894
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6233440744718944e-05,
+      "loss": 0.4548,
+      "step": 3895
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622986036519872e-05,
+      "loss": 0.2631,
+      "step": 3896
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.622627998567848e-05,
+      "loss": 0.4133,
+      "step": 3897
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6222699606158256e-05,
+      "loss": 0.4044,
+      "step": 3898
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621911922663803e-05,
+      "loss": 0.3847,
+      "step": 3899
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6215538847117796e-05,
+      "loss": 0.2864,
+      "step": 3900
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.621195846759757e-05,
+      "loss": 0.348,
+      "step": 3901
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6208378088077336e-05,
+      "loss": 0.3684,
+      "step": 3902
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620479770855711e-05,
+      "loss": 0.3219,
+      "step": 3903
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.620121732903688e-05,
+      "loss": 0.4448,
+      "step": 3904
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619763694951665e-05,
+      "loss": 0.3616,
+      "step": 3905
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619405656999642e-05,
+      "loss": 0.3504,
+      "step": 3906
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.619047619047619e-05,
+      "loss": 0.3548,
+      "step": 3907
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.618689581095596e-05,
+      "loss": 0.4254,
+      "step": 3908
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.6183315431435734e-05,
+      "loss": 0.3908,
+      "step": 3909
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 3.61797350519155e-05,
+      "loss": 0.3908,
+      "step": 3910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6176154672395274e-05,
+      "loss": 0.6092,
+      "step": 3911
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.617257429287505e-05,
+      "loss": 0.2768,
+      "step": 3912
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6168993913354814e-05,
+      "loss": 0.298,
+      "step": 3913
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616541353383459e-05,
+      "loss": 0.3999,
+      "step": 3914
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.616183315431436e-05,
+      "loss": 0.3582,
+      "step": 3915
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.615825277479413e-05,
+      "loss": 0.3434,
+      "step": 3916
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6154672395273906e-05,
+      "loss": 0.2055,
+      "step": 3917
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.615109201575367e-05,
+      "loss": 0.3434,
+      "step": 3918
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6147511636233446e-05,
+      "loss": 0.392,
+      "step": 3919
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.614393125671321e-05,
+      "loss": 0.3651,
+      "step": 3920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6140350877192985e-05,
+      "loss": 0.3914,
+      "step": 3921
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.613677049767276e-05,
+      "loss": 0.4494,
+      "step": 3922
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6133190118152525e-05,
+      "loss": 0.4886,
+      "step": 3923
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.61296097386323e-05,
+      "loss": 0.4364,
+      "step": 3924
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6126029359112065e-05,
+      "loss": 0.499,
+      "step": 3925
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.612244897959184e-05,
+      "loss": 0.4373,
+      "step": 3926
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611886860007161e-05,
+      "loss": 0.5013,
+      "step": 3927
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611528822055138e-05,
+      "loss": 0.3467,
+      "step": 3928
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.611170784103115e-05,
+      "loss": 0.4134,
+      "step": 3929
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6108127461510924e-05,
+      "loss": 0.348,
+      "step": 3930
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.610454708199069e-05,
+      "loss": 0.4166,
+      "step": 3931
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6100966702470463e-05,
+      "loss": 0.3654,
+      "step": 3932
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609738632295023e-05,
+      "loss": 0.396,
+      "step": 3933
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.609380594343e-05,
+      "loss": 0.3873,
+      "step": 3934
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6090225563909776e-05,
+      "loss": 0.4438,
+      "step": 3935
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.608664518438954e-05,
+      "loss": 0.4685,
+      "step": 3936
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.6083064804869316e-05,
+      "loss": 0.2913,
+      "step": 3937
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.607948442534909e-05,
+      "loss": 0.5729,
+      "step": 3938
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.607590404582886e-05,
+      "loss": 0.3805,
+      "step": 3939
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6072323666308635e-05,
+      "loss": 0.3941,
+      "step": 3940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.60687432867884e-05,
+      "loss": 0.4424,
+      "step": 3941
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6065162907268175e-05,
+      "loss": 0.4563,
+      "step": 3942
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.606158252774794e-05,
+      "loss": 0.4284,
+      "step": 3943
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6058002148227715e-05,
+      "loss": 0.3501,
+      "step": 3944
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.605442176870749e-05,
+      "loss": 0.4553,
+      "step": 3945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6050841389187254e-05,
+      "loss": 0.3817,
+      "step": 3946
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.604726100966703e-05,
+      "loss": 0.3553,
+      "step": 3947
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.60436806301468e-05,
+      "loss": 0.4488,
+      "step": 3948
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.604010025062657e-05,
+      "loss": 0.3902,
+      "step": 3949
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.603651987110634e-05,
+      "loss": 0.334,
+      "step": 3950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6032939491586107e-05,
+      "loss": 0.3572,
+      "step": 3951
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.602935911206588e-05,
+      "loss": 0.4677,
+      "step": 3952
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.602577873254565e-05,
+      "loss": 0.3164,
+      "step": 3953
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.602219835302542e-05,
+      "loss": 0.3077,
+      "step": 3954
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.601861797350519e-05,
+      "loss": 0.4094,
+      "step": 3955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6015037593984966e-05,
+      "loss": 0.3993,
+      "step": 3956
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.601145721446473e-05,
+      "loss": 0.424,
+      "step": 3957
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6007876834944505e-05,
+      "loss": 0.3184,
+      "step": 3958
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.600429645542427e-05,
+      "loss": 0.3629,
+      "step": 3959
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.6000716075904045e-05,
+      "loss": 0.4961,
+      "step": 3960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599713569638382e-05,
+      "loss": 0.4238,
+      "step": 3961
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.599355531686359e-05,
+      "loss": 0.5334,
+      "step": 3962
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5989974937343364e-05,
+      "loss": 0.4519,
+      "step": 3963
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.598639455782313e-05,
+      "loss": 0.4501,
+      "step": 3964
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.5982814178302904e-05,
+      "loss": 0.4105,
+      "step": 3965
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 3.597923379878268e-05,
+      "loss": 0.5132,
+      "step": 3966
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5975653419262444e-05,
+      "loss": 0.4348,
+      "step": 3967
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.597207303974222e-05,
+      "loss": 0.3482,
+      "step": 3968
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.596849266022198e-05,
+      "loss": 0.2705,
+      "step": 3969
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5964912280701756e-05,
+      "loss": 0.3168,
+      "step": 3970
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.596133190118153e-05,
+      "loss": 0.4775,
+      "step": 3971
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5957751521661296e-05,
+      "loss": 0.4088,
+      "step": 3972
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.595417114214107e-05,
+      "loss": 0.4213,
+      "step": 3973
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.595059076262084e-05,
+      "loss": 0.3488,
+      "step": 3974
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.594701038310061e-05,
+      "loss": 0.429,
+      "step": 3975
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.594343000358038e-05,
+      "loss": 0.4404,
+      "step": 3976
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.593984962406015e-05,
+      "loss": 0.3359,
+      "step": 3977
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.593626924453992e-05,
+      "loss": 0.4096,
+      "step": 3978
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5932688865019695e-05,
+      "loss": 0.4358,
+      "step": 3979
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.592910848549946e-05,
+      "loss": 0.472,
+      "step": 3980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5925528105979234e-05,
+      "loss": 0.3009,
+      "step": 3981
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.592194772645901e-05,
+      "loss": 0.5103,
+      "step": 3982
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5918367346938774e-05,
+      "loss": 0.4804,
+      "step": 3983
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.591478696741855e-05,
+      "loss": 0.3355,
+      "step": 3984
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.591120658789832e-05,
+      "loss": 0.4158,
+      "step": 3985
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5907626208378094e-05,
+      "loss": 0.4538,
+      "step": 3986
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.590404582885786e-05,
+      "loss": 0.4568,
+      "step": 3987
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.590046544933763e-05,
+      "loss": 0.2783,
+      "step": 3988
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5896885069817406e-05,
+      "loss": 0.2969,
+      "step": 3989
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.589330469029717e-05,
+      "loss": 0.4091,
+      "step": 3990
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5889724310776946e-05,
+      "loss": 0.5281,
+      "step": 3991
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.588614393125672e-05,
+      "loss": 0.4912,
+      "step": 3992
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.5882563551736486e-05,
+      "loss": 0.3117,
+      "step": 3993
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.587898317221626e-05,
+      "loss": 0.3777,
+      "step": 3994
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5875402792696025e-05,
+      "loss": 0.4127,
+      "step": 3995
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.58718224131758e-05,
+      "loss": 0.4186,
+      "step": 3996
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.586824203365557e-05,
+      "loss": 0.4126,
+      "step": 3997
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.586466165413534e-05,
+      "loss": 0.3701,
+      "step": 3998
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.586108127461511e-05,
+      "loss": 0.3879,
+      "step": 3999
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5857500895094884e-05,
+      "loss": 0.4041,
+      "step": 4000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.585392051557465e-05,
+      "loss": 0.38,
+      "step": 4001
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5850340136054424e-05,
+      "loss": 0.5023,
+      "step": 4002
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.584675975653419e-05,
+      "loss": 0.4961,
+      "step": 4003
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5843179377013964e-05,
+      "loss": 0.3151,
+      "step": 4004
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.583959899749374e-05,
+      "loss": 0.4367,
+      "step": 4005
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.58360186179735e-05,
+      "loss": 0.4719,
+      "step": 4006
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5832438238453276e-05,
+      "loss": 0.3592,
+      "step": 4007
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.582885785893304e-05,
+      "loss": 0.3513,
+      "step": 4008
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.582527747941282e-05,
+      "loss": 0.3857,
+      "step": 4009
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5821697099892596e-05,
+      "loss": 0.3928,
+      "step": 4010
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.581811672037236e-05,
+      "loss": 0.4036,
+      "step": 4011
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5814536340852135e-05,
+      "loss": 0.2658,
+      "step": 4012
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.58109559613319e-05,
+      "loss": 0.3126,
+      "step": 4013
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5807375581811675e-05,
+      "loss": 0.3048,
+      "step": 4014
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.580379520229145e-05,
+      "loss": 0.3661,
+      "step": 4015
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5800214822771215e-05,
+      "loss": 0.4323,
+      "step": 4016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.579663444325099e-05,
+      "loss": 0.4042,
+      "step": 4017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.5793054063730754e-05,
+      "loss": 0.3858,
+      "step": 4018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578947368421053e-05,
+      "loss": 0.4565,
+      "step": 4019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.57858933046903e-05,
+      "loss": 0.3167,
+      "step": 4020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.578231292517007e-05,
+      "loss": 0.4361,
+      "step": 4021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.577873254564984e-05,
+      "loss": 0.3345,
+      "step": 4022
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5775152166129613e-05,
+      "loss": 0.4227,
+      "step": 4023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.577157178660938e-05,
+      "loss": 0.3699,
+      "step": 4024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.576799140708915e-05,
+      "loss": 0.3779,
+      "step": 4025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.576441102756892e-05,
+      "loss": 0.4934,
+      "step": 4026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.576083064804869e-05,
+      "loss": 0.3862,
+      "step": 4027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5757250268528466e-05,
+      "loss": 0.6277,
+      "step": 4028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.575366988900823e-05,
+      "loss": 0.3674,
+      "step": 4029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5750089509488005e-05,
+      "loss": 0.4938,
+      "step": 4030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.574650912996778e-05,
+      "loss": 0.3425,
+      "step": 4031
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.574292875044755e-05,
+      "loss": 0.3134,
+      "step": 4032
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5739348370927325e-05,
+      "loss": 0.3582,
+      "step": 4033
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.573576799140709e-05,
+      "loss": 0.4611,
+      "step": 4034
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5732187611886865e-05,
+      "loss": 0.427,
+      "step": 4035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.572860723236663e-05,
+      "loss": 0.3071,
+      "step": 4036
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5725026852846404e-05,
+      "loss": 0.411,
+      "step": 4037
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.572144647332618e-05,
+      "loss": 0.3526,
+      "step": 4038
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5717866093805944e-05,
+      "loss": 0.413,
+      "step": 4039
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3874,
+      "step": 4040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.571070533476549e-05,
+      "loss": 0.4509,
+      "step": 4041
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5707124955245257e-05,
+      "loss": 0.4271,
+      "step": 4042
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.570354457572503e-05,
+      "loss": 0.3819,
+      "step": 4043
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5699964196204796e-05,
+      "loss": 0.3345,
+      "step": 4044
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569638381668457e-05,
+      "loss": 0.5108,
+      "step": 4045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.569280343716434e-05,
+      "loss": 0.3223,
+      "step": 4046
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568922305764411e-05,
+      "loss": 0.406,
+      "step": 4047
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.568564267812388e-05,
+      "loss": 0.3223,
+      "step": 4048
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.5682062298603655e-05,
+      "loss": 0.318,
+      "step": 4049
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 3.567848191908342e-05,
+      "loss": 0.35,
+      "step": 4050
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5674901539563195e-05,
+      "loss": 0.2959,
+      "step": 4051
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.567132116004296e-05,
+      "loss": 0.3234,
+      "step": 4052
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5667740780522735e-05,
+      "loss": 0.3439,
+      "step": 4053
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.566416040100251e-05,
+      "loss": 0.3667,
+      "step": 4054
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5660580021482274e-05,
+      "loss": 0.5333,
+      "step": 4055
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5656999641962054e-05,
+      "loss": 0.4639,
+      "step": 4056
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.565341926244182e-05,
+      "loss": 0.3775,
+      "step": 4057
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5649838882921594e-05,
+      "loss": 0.2739,
+      "step": 4058
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.564625850340137e-05,
+      "loss": 0.369,
+      "step": 4059
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.564267812388113e-05,
+      "loss": 0.4181,
+      "step": 4060
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5639097744360906e-05,
+      "loss": 0.4611,
+      "step": 4061
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.563551736484067e-05,
+      "loss": 0.5259,
+      "step": 4062
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5631936985320446e-05,
+      "loss": 0.4285,
+      "step": 4063
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.562835660580022e-05,
+      "loss": 0.3329,
+      "step": 4064
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5624776226279986e-05,
+      "loss": 0.3236,
+      "step": 4065
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.562119584675976e-05,
+      "loss": 0.3397,
+      "step": 4066
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.561761546723953e-05,
+      "loss": 0.3459,
+      "step": 4067
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.56140350877193e-05,
+      "loss": 0.2788,
+      "step": 4068
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.561045470819907e-05,
+      "loss": 0.2239,
+      "step": 4069
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.560687432867884e-05,
+      "loss": 0.3839,
+      "step": 4070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.560329394915861e-05,
+      "loss": 0.3243,
+      "step": 4071
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5599713569638384e-05,
+      "loss": 0.3058,
+      "step": 4072
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.559613319011815e-05,
+      "loss": 0.391,
+      "step": 4073
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5592552810597924e-05,
+      "loss": 0.3189,
+      "step": 4074
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.55889724310777e-05,
+      "loss": 0.2732,
+      "step": 4075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5585392051557464e-05,
+      "loss": 0.5734,
+      "step": 4076
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.558181167203724e-05,
+      "loss": 0.5069,
+      "step": 4077
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.5578231292517e-05,
+      "loss": 0.3204,
+      "step": 4078
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.557465091299678e-05,
+      "loss": 0.396,
+      "step": 4079
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.557107053347655e-05,
+      "loss": 0.303,
+      "step": 4080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.556749015395632e-05,
+      "loss": 0.2489,
+      "step": 4081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5563909774436096e-05,
+      "loss": 0.4153,
+      "step": 4082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.556032939491586e-05,
+      "loss": 0.3841,
+      "step": 4083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5556749015395636e-05,
+      "loss": 0.5086,
+      "step": 4084
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.555316863587541e-05,
+      "loss": 0.3061,
+      "step": 4085
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5549588256355175e-05,
+      "loss": 0.336,
+      "step": 4086
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.554600787683495e-05,
+      "loss": 0.515,
+      "step": 4087
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5542427497314715e-05,
+      "loss": 0.3472,
+      "step": 4088
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553884711779449e-05,
+      "loss": 0.5236,
+      "step": 4089
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553526673827426e-05,
+      "loss": 0.4569,
+      "step": 4090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.553168635875403e-05,
+      "loss": 0.5146,
+      "step": 4091
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.55281059792338e-05,
+      "loss": 0.4214,
+      "step": 4092
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5524525599713574e-05,
+      "loss": 0.3503,
+      "step": 4093
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.552094522019334e-05,
+      "loss": 0.3669,
+      "step": 4094
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5517364840673114e-05,
+      "loss": 0.4289,
+      "step": 4095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.551378446115288e-05,
+      "loss": 0.4594,
+      "step": 4096
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.551020408163265e-05,
+      "loss": 0.2887,
+      "step": 4097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5506623702112426e-05,
+      "loss": 0.3181,
+      "step": 4098
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.550304332259219e-05,
+      "loss": 0.2992,
+      "step": 4099
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5499462943071966e-05,
+      "loss": 0.4714,
+      "step": 4100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549588256355173e-05,
+      "loss": 0.3318,
+      "step": 4101
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.549230218403151e-05,
+      "loss": 0.3781,
+      "step": 4102
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5488721804511285e-05,
+      "loss": 0.3104,
+      "step": 4103
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.548514142499105e-05,
+      "loss": 0.3501,
+      "step": 4104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.5481561045470825e-05,
+      "loss": 0.3537,
+      "step": 4105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 3.547798066595059e-05,
+      "loss": 0.3654,
+      "step": 4106
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5474400286430365e-05,
+      "loss": 0.3488,
+      "step": 4107
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.547081990691014e-05,
+      "loss": 0.411,
+      "step": 4108
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5467239527389904e-05,
+      "loss": 0.5381,
+      "step": 4109
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.546365914786968e-05,
+      "loss": 0.4811,
+      "step": 4110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.546007876834945e-05,
+      "loss": 0.4437,
+      "step": 4111
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.545649838882922e-05,
+      "loss": 0.4345,
+      "step": 4112
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.545291800930899e-05,
+      "loss": 0.4259,
+      "step": 4113
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.544933762978876e-05,
+      "loss": 0.2813,
+      "step": 4114
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.544575725026853e-05,
+      "loss": 0.3723,
+      "step": 4115
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.54421768707483e-05,
+      "loss": 0.4947,
+      "step": 4116
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.543859649122807e-05,
+      "loss": 0.3466,
+      "step": 4117
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.543501611170784e-05,
+      "loss": 0.4974,
+      "step": 4118
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.543143573218761e-05,
+      "loss": 0.357,
+      "step": 4119
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.542785535266738e-05,
+      "loss": 0.379,
+      "step": 4120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5424274973147155e-05,
+      "loss": 0.366,
+      "step": 4121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.542069459362692e-05,
+      "loss": 0.3989,
+      "step": 4122
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5417114214106695e-05,
+      "loss": 0.538,
+      "step": 4123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.541353383458647e-05,
+      "loss": 0.3091,
+      "step": 4124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5409953455066235e-05,
+      "loss": 0.3453,
+      "step": 4125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5406373075546015e-05,
+      "loss": 0.384,
+      "step": 4126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.540279269602578e-05,
+      "loss": 0.532,
+      "step": 4127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5399212316505554e-05,
+      "loss": 0.4128,
+      "step": 4128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.539563193698532e-05,
+      "loss": 0.4008,
+      "step": 4129
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5392051557465094e-05,
+      "loss": 0.5186,
+      "step": 4130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.538847117794487e-05,
+      "loss": 0.4138,
+      "step": 4131
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.538489079842463e-05,
+      "loss": 0.4476,
+      "step": 4132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.5381310418904407e-05,
+      "loss": 0.3338,
+      "step": 4133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.537773003938418e-05,
+      "loss": 0.3224,
+      "step": 4134
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5374149659863946e-05,
+      "loss": 0.4115,
+      "step": 4135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.537056928034372e-05,
+      "loss": 0.2988,
+      "step": 4136
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5366988900823486e-05,
+      "loss": 0.3711,
+      "step": 4137
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.536340852130326e-05,
+      "loss": 0.4803,
+      "step": 4138
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.535982814178303e-05,
+      "loss": 0.4241,
+      "step": 4139
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.53562477622628e-05,
+      "loss": 0.4224,
+      "step": 4140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.535266738274257e-05,
+      "loss": 0.3726,
+      "step": 4141
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5349087003222345e-05,
+      "loss": 0.3275,
+      "step": 4142
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.534550662370211e-05,
+      "loss": 0.3435,
+      "step": 4143
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5341926244181885e-05,
+      "loss": 0.3381,
+      "step": 4144
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.533834586466165e-05,
+      "loss": 0.3571,
+      "step": 4145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5334765485141424e-05,
+      "loss": 0.5619,
+      "step": 4146
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.53311851056212e-05,
+      "loss": 0.2963,
+      "step": 4147
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5327604726100964e-05,
+      "loss": 0.2948,
+      "step": 4148
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5324024346580744e-05,
+      "loss": 0.3002,
+      "step": 4149
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.532044396706051e-05,
+      "loss": 0.4214,
+      "step": 4150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.531686358754028e-05,
+      "loss": 0.3685,
+      "step": 4151
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5313283208020056e-05,
+      "loss": 0.385,
+      "step": 4152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530970282849982e-05,
+      "loss": 0.4567,
+      "step": 4153
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5306122448979596e-05,
+      "loss": 0.4148,
+      "step": 4154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.530254206945936e-05,
+      "loss": 0.5082,
+      "step": 4155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5298961689939136e-05,
+      "loss": 0.3573,
+      "step": 4156
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.529538131041891e-05,
+      "loss": 0.5192,
+      "step": 4157
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.5291800930898675e-05,
+      "loss": 0.2832,
+      "step": 4158
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528822055137845e-05,
+      "loss": 0.2545,
+      "step": 4159
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528464017185822e-05,
+      "loss": 0.461,
+      "step": 4160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.528105979233799e-05,
+      "loss": 0.4142,
+      "step": 4161
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 3.527747941281776e-05,
+      "loss": 0.3938,
+      "step": 4162
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.527389903329753e-05,
+      "loss": 0.4142,
+      "step": 4163
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.52703186537773e-05,
+      "loss": 0.3159,
+      "step": 4164
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5266738274257074e-05,
+      "loss": 0.3556,
+      "step": 4165
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.526315789473684e-05,
+      "loss": 0.3125,
+      "step": 4166
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5259577515216614e-05,
+      "loss": 0.3504,
+      "step": 4167
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.525599713569639e-05,
+      "loss": 0.3679,
+      "step": 4168
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.525241675617615e-05,
+      "loss": 0.4236,
+      "step": 4169
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5248836376655926e-05,
+      "loss": 0.5174,
+      "step": 4170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.524525599713569e-05,
+      "loss": 0.3422,
+      "step": 4171
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.524167561761547e-05,
+      "loss": 0.3281,
+      "step": 4172
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.523809523809524e-05,
+      "loss": 0.3679,
+      "step": 4173
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.523451485857501e-05,
+      "loss": 0.3826,
+      "step": 4174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5230934479054786e-05,
+      "loss": 0.384,
+      "step": 4175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.522735409953455e-05,
+      "loss": 0.439,
+      "step": 4176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5223773720014325e-05,
+      "loss": 0.3585,
+      "step": 4177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.52201933404941e-05,
+      "loss": 0.3432,
+      "step": 4178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5216612960973865e-05,
+      "loss": 0.3466,
+      "step": 4179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.521303258145364e-05,
+      "loss": 0.3592,
+      "step": 4180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5209452201933404e-05,
+      "loss": 0.3329,
+      "step": 4181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520587182241318e-05,
+      "loss": 0.3612,
+      "step": 4182
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.520229144289295e-05,
+      "loss": 0.4394,
+      "step": 4183
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.519871106337272e-05,
+      "loss": 0.5106,
+      "step": 4184
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.519513068385249e-05,
+      "loss": 0.312,
+      "step": 4185
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.5191550304332264e-05,
+      "loss": 0.3978,
+      "step": 4186
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518796992481203e-05,
+      "loss": 0.3715,
+      "step": 4187
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.51843895452918e-05,
+      "loss": 0.32,
+      "step": 4188
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.518080916577157e-05,
+      "loss": 0.3605,
+      "step": 4189
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.517722878625134e-05,
+      "loss": 0.3819,
+      "step": 4190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5173648406731116e-05,
+      "loss": 0.4195,
+      "step": 4191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.517006802721088e-05,
+      "loss": 0.4546,
+      "step": 4192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5166487647690656e-05,
+      "loss": 0.3835,
+      "step": 4193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.516290726817042e-05,
+      "loss": 0.3984,
+      "step": 4194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5159326888650195e-05,
+      "loss": 0.3851,
+      "step": 4195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5155746509129975e-05,
+      "loss": 0.4312,
+      "step": 4196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.515216612960974e-05,
+      "loss": 0.3716,
+      "step": 4197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5148585750089515e-05,
+      "loss": 0.2759,
+      "step": 4198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.514500537056928e-05,
+      "loss": 0.3802,
+      "step": 4199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5141424991049054e-05,
+      "loss": 0.4493,
+      "step": 4200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.513784461152883e-05,
+      "loss": 0.4525,
+      "step": 4201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5134264232008594e-05,
+      "loss": 0.5171,
+      "step": 4202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.513068385248837e-05,
+      "loss": 0.4653,
+      "step": 4203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.512710347296814e-05,
+      "loss": 0.3281,
+      "step": 4204
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.512352309344791e-05,
+      "loss": 0.4964,
+      "step": 4205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.511994271392768e-05,
+      "loss": 0.3917,
+      "step": 4206
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5116362334407446e-05,
+      "loss": 0.3486,
+      "step": 4207
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.511278195488722e-05,
+      "loss": 0.3531,
+      "step": 4208
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510920157536699e-05,
+      "loss": 0.3051,
+      "step": 4209
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510562119584676e-05,
+      "loss": 0.3768,
+      "step": 4210
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.510204081632653e-05,
+      "loss": 0.3373,
+      "step": 4211
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.50984604368063e-05,
+      "loss": 0.3622,
+      "step": 4212
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.509488005728607e-05,
+      "loss": 0.4021,
+      "step": 4213
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5091299677765845e-05,
+      "loss": 0.3118,
+      "step": 4214
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.3928,
+      "step": 4215
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5084138918725385e-05,
+      "loss": 0.4599,
+      "step": 4216
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.508055853920516e-05,
+      "loss": 0.4247,
+      "step": 4217
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 3.5076978159684924e-05,
+      "loss": 0.4507,
+      "step": 4218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5073397780164704e-05,
+      "loss": 0.4271,
+      "step": 4219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.506981740064447e-05,
+      "loss": 0.4804,
+      "step": 4220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5066237021124244e-05,
+      "loss": 0.3693,
+      "step": 4221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.506265664160402e-05,
+      "loss": 0.4298,
+      "step": 4222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.505907626208378e-05,
+      "loss": 0.3814,
+      "step": 4223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5055495882563557e-05,
+      "loss": 0.5272,
+      "step": 4224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.505191550304332e-05,
+      "loss": 0.3935,
+      "step": 4225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5048335123523096e-05,
+      "loss": 0.5009,
+      "step": 4226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.504475474400287e-05,
+      "loss": 0.3905,
+      "step": 4227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5041174364482636e-05,
+      "loss": 0.3603,
+      "step": 4228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.503759398496241e-05,
+      "loss": 0.3515,
+      "step": 4229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5034013605442175e-05,
+      "loss": 0.4371,
+      "step": 4230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.503043322592195e-05,
+      "loss": 0.3633,
+      "step": 4231
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.502685284640172e-05,
+      "loss": 0.3542,
+      "step": 4232
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.502327246688149e-05,
+      "loss": 0.4645,
+      "step": 4233
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.501969208736126e-05,
+      "loss": 0.3689,
+      "step": 4234
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5016111707841035e-05,
+      "loss": 0.2909,
+      "step": 4235
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.50125313283208e-05,
+      "loss": 0.4023,
+      "step": 4236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5008950948800574e-05,
+      "loss": 0.4248,
+      "step": 4237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.500537056928034e-05,
+      "loss": 0.3987,
+      "step": 4238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.5001790189760114e-05,
+      "loss": 0.3328,
+      "step": 4239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499820981023989e-05,
+      "loss": 0.4453,
+      "step": 4240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499462943071965e-05,
+      "loss": 0.3632,
+      "step": 4241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.499104905119943e-05,
+      "loss": 0.2811,
+      "step": 4242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.49874686716792e-05,
+      "loss": 0.3884,
+      "step": 4243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.498388829215897e-05,
+      "loss": 0.4205,
+      "step": 4244
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.4980307912638746e-05,
+      "loss": 0.3786,
+      "step": 4245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.497672753311851e-05,
+      "loss": 0.3813,
+      "step": 4246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4973147153598286e-05,
+      "loss": 0.4665,
+      "step": 4247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.496956677407805e-05,
+      "loss": 0.4235,
+      "step": 4248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4965986394557825e-05,
+      "loss": 0.5009,
+      "step": 4249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.49624060150376e-05,
+      "loss": 0.3459,
+      "step": 4250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4958825635517365e-05,
+      "loss": 0.3926,
+      "step": 4251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.495524525599714e-05,
+      "loss": 0.3113,
+      "step": 4252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.495166487647691e-05,
+      "loss": 0.5109,
+      "step": 4253
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.494808449695668e-05,
+      "loss": 0.306,
+      "step": 4254
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.494450411743645e-05,
+      "loss": 0.4772,
+      "step": 4255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.494092373791622e-05,
+      "loss": 0.3414,
+      "step": 4256
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.493734335839599e-05,
+      "loss": 0.4273,
+      "step": 4257
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4933762978875764e-05,
+      "loss": 0.3767,
+      "step": 4258
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.493018259935553e-05,
+      "loss": 0.3359,
+      "step": 4259
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.49266022198353e-05,
+      "loss": 0.3041,
+      "step": 4260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4923021840315076e-05,
+      "loss": 0.4014,
+      "step": 4261
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.491944146079484e-05,
+      "loss": 0.3317,
+      "step": 4262
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4915861081274616e-05,
+      "loss": 0.4275,
+      "step": 4263
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.491228070175438e-05,
+      "loss": 0.3069,
+      "step": 4264
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4908700322234156e-05,
+      "loss": 0.4295,
+      "step": 4265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.490511994271393e-05,
+      "loss": 0.2803,
+      "step": 4266
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.49015395631937e-05,
+      "loss": 0.4739,
+      "step": 4267
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4897959183673475e-05,
+      "loss": 0.4407,
+      "step": 4268
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.489437880415324e-05,
+      "loss": 0.4522,
+      "step": 4269
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4890798424633015e-05,
+      "loss": 0.2631,
+      "step": 4270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488721804511279e-05,
+      "loss": 0.3077,
+      "step": 4271
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4883637665592554e-05,
+      "loss": 0.3203,
+      "step": 4272
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.488005728607233e-05,
+      "loss": 0.3615,
+      "step": 4273
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.4876476906552094e-05,
+      "loss": 0.5304,
+      "step": 4274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.487289652703187e-05,
+      "loss": 0.3384,
+      "step": 4275
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.486931614751164e-05,
+      "loss": 0.2076,
+      "step": 4276
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.486573576799141e-05,
+      "loss": 0.3994,
+      "step": 4277
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.486215538847118e-05,
+      "loss": 0.2899,
+      "step": 4278
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.485857500895095e-05,
+      "loss": 0.3231,
+      "step": 4279
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.485499462943072e-05,
+      "loss": 0.2848,
+      "step": 4280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.485141424991049e-05,
+      "loss": 0.6507,
+      "step": 4281
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.484783387039026e-05,
+      "loss": 0.4349,
+      "step": 4282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.484425349087003e-05,
+      "loss": 0.3193,
+      "step": 4283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4840673111349806e-05,
+      "loss": 0.3426,
+      "step": 4284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.483709273182957e-05,
+      "loss": 0.3813,
+      "step": 4285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4833512352309345e-05,
+      "loss": 0.3661,
+      "step": 4286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.482993197278911e-05,
+      "loss": 0.4663,
+      "step": 4287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4826351593268885e-05,
+      "loss": 0.4015,
+      "step": 4288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4822771213748665e-05,
+      "loss": 0.3638,
+      "step": 4289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.481919083422843e-05,
+      "loss": 0.3399,
+      "step": 4290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4815610454708204e-05,
+      "loss": 0.3622,
+      "step": 4291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.481203007518797e-05,
+      "loss": 0.2832,
+      "step": 4292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4808449695667744e-05,
+      "loss": 0.4841,
+      "step": 4293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.480486931614752e-05,
+      "loss": 0.2658,
+      "step": 4294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4801288936627283e-05,
+      "loss": 0.2867,
+      "step": 4295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479770855710706e-05,
+      "loss": 0.3951,
+      "step": 4296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.479412817758683e-05,
+      "loss": 0.4752,
+      "step": 4297
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4790547798066596e-05,
+      "loss": 0.337,
+      "step": 4298
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.478696741854637e-05,
+      "loss": 0.5362,
+      "step": 4299
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.4783387039026136e-05,
+      "loss": 0.3846,
+      "step": 4300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477980665950591e-05,
+      "loss": 0.3089,
+      "step": 4301
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 3.477622627998568e-05,
+      "loss": 0.3052,
+      "step": 4302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.477264590046545e-05,
+      "loss": 0.4461,
+      "step": 4303
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.476906552094522e-05,
+      "loss": 0.4469,
+      "step": 4304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.476548514142499e-05,
+      "loss": 0.3062,
+      "step": 4305
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.476190476190476e-05,
+      "loss": 0.514,
+      "step": 4306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4758324382384535e-05,
+      "loss": 0.3739,
+      "step": 4307
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.47547440028643e-05,
+      "loss": 0.4063,
+      "step": 4308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4751163623344074e-05,
+      "loss": 0.4242,
+      "step": 4309
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.474758324382385e-05,
+      "loss": 0.4869,
+      "step": 4310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4744002864303614e-05,
+      "loss": 0.3283,
+      "step": 4311
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4740422484783394e-05,
+      "loss": 0.3913,
+      "step": 4312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.473684210526316e-05,
+      "loss": 0.4415,
+      "step": 4313
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4733261725742933e-05,
+      "loss": 0.3217,
+      "step": 4314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4729681346222707e-05,
+      "loss": 0.4964,
+      "step": 4315
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.472610096670247e-05,
+      "loss": 0.3527,
+      "step": 4316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4722520587182246e-05,
+      "loss": 0.4613,
+      "step": 4317
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.471894020766201e-05,
+      "loss": 0.33,
+      "step": 4318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4715359828141786e-05,
+      "loss": 0.3376,
+      "step": 4319
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.471177944862156e-05,
+      "loss": 0.3942,
+      "step": 4320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4708199069101325e-05,
+      "loss": 0.3562,
+      "step": 4321
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.47046186895811e-05,
+      "loss": 0.317,
+      "step": 4322
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4701038310060865e-05,
+      "loss": 0.4274,
+      "step": 4323
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469745793054064e-05,
+      "loss": 0.4549,
+      "step": 4324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469387755102041e-05,
+      "loss": 0.4849,
+      "step": 4325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.469029717150018e-05,
+      "loss": 0.3825,
+      "step": 4326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.468671679197995e-05,
+      "loss": 0.2674,
+      "step": 4327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4683136412459724e-05,
+      "loss": 0.3916,
+      "step": 4328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.467955603293949e-05,
+      "loss": 0.2953,
+      "step": 4329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.4675975653419264e-05,
+      "loss": 0.4289,
+      "step": 4330
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.467239527389903e-05,
+      "loss": 0.3775,
+      "step": 4331
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.46688148943788e-05,
+      "loss": 0.369,
+      "step": 4332
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4665234514858577e-05,
+      "loss": 0.3655,
+      "step": 4333
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.466165413533834e-05,
+      "loss": 0.3751,
+      "step": 4334
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4658073755818116e-05,
+      "loss": 0.4591,
+      "step": 4335
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.465449337629789e-05,
+      "loss": 0.2895,
+      "step": 4336
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.465091299677766e-05,
+      "loss": 0.4643,
+      "step": 4337
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4647332617257436e-05,
+      "loss": 0.3689,
+      "step": 4338
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.46437522377372e-05,
+      "loss": 0.2866,
+      "step": 4339
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4640171858216975e-05,
+      "loss": 0.3728,
+      "step": 4340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.463659147869674e-05,
+      "loss": 0.4708,
+      "step": 4341
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4633011099176515e-05,
+      "loss": 0.3385,
+      "step": 4342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.462943071965629e-05,
+      "loss": 0.35,
+      "step": 4343
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4625850340136054e-05,
+      "loss": 0.3547,
+      "step": 4344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.462226996061583e-05,
+      "loss": 0.2357,
+      "step": 4345
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.46186895810956e-05,
+      "loss": 0.3368,
+      "step": 4346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.461510920157537e-05,
+      "loss": 0.3691,
+      "step": 4347
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.461152882205514e-05,
+      "loss": 0.3189,
+      "step": 4348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460794844253491e-05,
+      "loss": 0.4404,
+      "step": 4349
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460436806301468e-05,
+      "loss": 0.3682,
+      "step": 4350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.460078768349445e-05,
+      "loss": 0.4261,
+      "step": 4351
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459720730397422e-05,
+      "loss": 0.3463,
+      "step": 4352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.459362692445399e-05,
+      "loss": 0.373,
+      "step": 4353
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4590046544933766e-05,
+      "loss": 0.6169,
+      "step": 4354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.458646616541353e-05,
+      "loss": 0.4181,
+      "step": 4355
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4582885785893306e-05,
+      "loss": 0.3908,
+      "step": 4356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.457930540637307e-05,
+      "loss": 0.3165,
+      "step": 4357
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 3.4575725026852845e-05,
+      "loss": 0.3943,
+      "step": 4358
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.457214464733262e-05,
+      "loss": 0.3095,
+      "step": 4359
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.456856426781239e-05,
+      "loss": 0.3663,
+      "step": 4360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4564983888292165e-05,
+      "loss": 0.5264,
+      "step": 4361
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.456140350877193e-05,
+      "loss": 0.3247,
+      "step": 4362
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4557823129251704e-05,
+      "loss": 0.3859,
+      "step": 4363
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.455424274973148e-05,
+      "loss": 0.3998,
+      "step": 4364
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4550662370211244e-05,
+      "loss": 0.3257,
+      "step": 4365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.454708199069102e-05,
+      "loss": 0.3608,
+      "step": 4366
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4543501611170784e-05,
+      "loss": 0.5305,
+      "step": 4367
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.453992123165056e-05,
+      "loss": 0.4251,
+      "step": 4368
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.453634085213033e-05,
+      "loss": 0.3619,
+      "step": 4369
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4532760472610096e-05,
+      "loss": 0.3739,
+      "step": 4370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.452918009308987e-05,
+      "loss": 0.2799,
+      "step": 4371
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.452559971356964e-05,
+      "loss": 0.3923,
+      "step": 4372
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.452201933404941e-05,
+      "loss": 0.3939,
+      "step": 4373
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.451843895452918e-05,
+      "loss": 0.3088,
+      "step": 4374
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.451485857500895e-05,
+      "loss": 0.2937,
+      "step": 4375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.451127819548872e-05,
+      "loss": 0.4657,
+      "step": 4376
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4507697815968495e-05,
+      "loss": 0.4174,
+      "step": 4377
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.450411743644826e-05,
+      "loss": 0.412,
+      "step": 4378
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4500537056928035e-05,
+      "loss": 0.3803,
+      "step": 4379
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.449695667740781e-05,
+      "loss": 0.3122,
+      "step": 4380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4493376297887574e-05,
+      "loss": 0.2759,
+      "step": 4381
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4489795918367354e-05,
+      "loss": 0.2778,
+      "step": 4382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.448621553884712e-05,
+      "loss": 0.3018,
+      "step": 4383
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4482635159326894e-05,
+      "loss": 0.3158,
+      "step": 4384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.447905477980666e-05,
+      "loss": 0.4842,
+      "step": 4385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.4475474400286433e-05,
+      "loss": 0.4593,
+      "step": 4386
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.447189402076621e-05,
+      "loss": 0.455,
+      "step": 4387
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.446831364124597e-05,
+      "loss": 0.3711,
+      "step": 4388
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4464733261725746e-05,
+      "loss": 0.2712,
+      "step": 4389
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.446115288220552e-05,
+      "loss": 0.5907,
+      "step": 4390
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4457572502685286e-05,
+      "loss": 0.4175,
+      "step": 4391
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.445399212316506e-05,
+      "loss": 0.469,
+      "step": 4392
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4450411743644825e-05,
+      "loss": 0.368,
+      "step": 4393
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.44468313641246e-05,
+      "loss": 0.3336,
+      "step": 4394
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.444325098460437e-05,
+      "loss": 0.3832,
+      "step": 4395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.443967060508414e-05,
+      "loss": 0.4351,
+      "step": 4396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.443609022556391e-05,
+      "loss": 0.3568,
+      "step": 4397
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.443250984604368e-05,
+      "loss": 0.3199,
+      "step": 4398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.442892946652345e-05,
+      "loss": 0.5582,
+      "step": 4399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4425349087003224e-05,
+      "loss": 0.3731,
+      "step": 4400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.442176870748299e-05,
+      "loss": 0.4038,
+      "step": 4401
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4418188327962764e-05,
+      "loss": 0.3791,
+      "step": 4402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.441460794844254e-05,
+      "loss": 0.2735,
+      "step": 4403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4411027568922303e-05,
+      "loss": 0.3548,
+      "step": 4404
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440744718940208e-05,
+      "loss": 0.4269,
+      "step": 4405
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440386680988185e-05,
+      "loss": 0.4312,
+      "step": 4406
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.440028643036162e-05,
+      "loss": 0.3713,
+      "step": 4407
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4396706050841396e-05,
+      "loss": 0.334,
+      "step": 4408
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.439312567132116e-05,
+      "loss": 0.3939,
+      "step": 4409
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4389545291800936e-05,
+      "loss": 0.3103,
+      "step": 4410
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.43859649122807e-05,
+      "loss": 0.3517,
+      "step": 4411
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4382384532760475e-05,
+      "loss": 0.341,
+      "step": 4412
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.437880415324025e-05,
+      "loss": 0.3693,
+      "step": 4413
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 3.4375223773720015e-05,
+      "loss": 0.3469,
+      "step": 4414
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.437164339419979e-05,
+      "loss": 0.3815,
+      "step": 4415
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4368063014679555e-05,
+      "loss": 0.3072,
+      "step": 4416
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.436448263515933e-05,
+      "loss": 0.5265,
+      "step": 4417
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.43609022556391e-05,
+      "loss": 0.3368,
+      "step": 4418
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.435732187611887e-05,
+      "loss": 0.352,
+      "step": 4419
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.435374149659864e-05,
+      "loss": 0.3139,
+      "step": 4420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4350161117078414e-05,
+      "loss": 0.3867,
+      "step": 4421
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.434658073755818e-05,
+      "loss": 0.4372,
+      "step": 4422
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.434300035803795e-05,
+      "loss": 0.3118,
+      "step": 4423
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.433941997851772e-05,
+      "loss": 0.2724,
+      "step": 4424
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.433583959899749e-05,
+      "loss": 0.4146,
+      "step": 4425
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4332259219477266e-05,
+      "loss": 0.3802,
+      "step": 4426
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.432867883995703e-05,
+      "loss": 0.3678,
+      "step": 4427
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4325098460436806e-05,
+      "loss": 0.2553,
+      "step": 4428
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.432151808091658e-05,
+      "loss": 0.2652,
+      "step": 4429
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.431793770139635e-05,
+      "loss": 0.3241,
+      "step": 4430
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4314357321876125e-05,
+      "loss": 0.2732,
+      "step": 4431
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.431077694235589e-05,
+      "loss": 0.2512,
+      "step": 4432
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4307196562835665e-05,
+      "loss": 0.5289,
+      "step": 4433
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.430361618331543e-05,
+      "loss": 0.3918,
+      "step": 4434
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4300035803795204e-05,
+      "loss": 0.2427,
+      "step": 4435
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.429645542427498e-05,
+      "loss": 0.4346,
+      "step": 4436
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4292875044754744e-05,
+      "loss": 0.3502,
+      "step": 4437
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.428929466523452e-05,
+      "loss": 0.4567,
+      "step": 4438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.428571428571429e-05,
+      "loss": 0.2848,
+      "step": 4439
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.428213390619406e-05,
+      "loss": 0.2851,
+      "step": 4440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.427855352667383e-05,
+      "loss": 0.3555,
+      "step": 4441
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.4274973147153596e-05,
+      "loss": 0.2781,
+      "step": 4442
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.427139276763337e-05,
+      "loss": 0.3869,
+      "step": 4443
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.426781238811314e-05,
+      "loss": 0.4517,
+      "step": 4444
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.426423200859291e-05,
+      "loss": 0.4009,
+      "step": 4445
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.426065162907268e-05,
+      "loss": 0.4547,
+      "step": 4446
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4257071249552456e-05,
+      "loss": 0.4354,
+      "step": 4447
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.425349087003222e-05,
+      "loss": 0.4018,
+      "step": 4448
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4249910490511995e-05,
+      "loss": 0.5062,
+      "step": 4449
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.424633011099176e-05,
+      "loss": 0.4269,
+      "step": 4450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4242749731471535e-05,
+      "loss": 0.4067,
+      "step": 4451
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.423916935195131e-05,
+      "loss": 0.4883,
+      "step": 4452
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.423558897243108e-05,
+      "loss": 0.1994,
+      "step": 4453
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4232008592910854e-05,
+      "loss": 0.3829,
+      "step": 4454
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.422842821339062e-05,
+      "loss": 0.1988,
+      "step": 4455
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4224847833870394e-05,
+      "loss": 0.3153,
+      "step": 4456
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.422126745435017e-05,
+      "loss": 0.3445,
+      "step": 4457
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4217687074829934e-05,
+      "loss": 0.4742,
+      "step": 4458
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.421410669530971e-05,
+      "loss": 0.4958,
+      "step": 4459
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.421052631578947e-05,
+      "loss": 0.3585,
+      "step": 4460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4206945936269246e-05,
+      "loss": 0.3297,
+      "step": 4461
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.420336555674902e-05,
+      "loss": 0.4544,
+      "step": 4462
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4199785177228786e-05,
+      "loss": 0.4282,
+      "step": 4463
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.419620479770856e-05,
+      "loss": 0.3497,
+      "step": 4464
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.419262441818833e-05,
+      "loss": 0.4662,
+      "step": 4465
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.41890440386681e-05,
+      "loss": 0.271,
+      "step": 4466
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.418546365914787e-05,
+      "loss": 0.4729,
+      "step": 4467
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.418188327962764e-05,
+      "loss": 0.461,
+      "step": 4468
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.417830290010741e-05,
+      "loss": 0.357,
+      "step": 4469
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.4174722520587185e-05,
+      "loss": 0.4426,
+      "step": 4470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.417114214106695e-05,
+      "loss": 0.4116,
+      "step": 4471
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4167561761546724e-05,
+      "loss": 0.449,
+      "step": 4472
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.41639813820265e-05,
+      "loss": 0.3792,
+      "step": 4473
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4160401002506264e-05,
+      "loss": 0.3776,
+      "step": 4474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.415682062298604e-05,
+      "loss": 0.3502,
+      "step": 4475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.415324024346581e-05,
+      "loss": 0.3678,
+      "step": 4476
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4149659863945583e-05,
+      "loss": 0.2495,
+      "step": 4477
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.414607948442535e-05,
+      "loss": 0.3596,
+      "step": 4478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.414249910490512e-05,
+      "loss": 0.2974,
+      "step": 4479
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4138918725384896e-05,
+      "loss": 0.4601,
+      "step": 4480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.413533834586466e-05,
+      "loss": 0.371,
+      "step": 4481
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4131757966344436e-05,
+      "loss": 0.3726,
+      "step": 4482
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.412817758682421e-05,
+      "loss": 0.3868,
+      "step": 4483
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4124597207303975e-05,
+      "loss": 0.4187,
+      "step": 4484
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.412101682778375e-05,
+      "loss": 0.366,
+      "step": 4485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4117436448263515e-05,
+      "loss": 0.372,
+      "step": 4486
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.411385606874329e-05,
+      "loss": 0.3365,
+      "step": 4487
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.411027568922306e-05,
+      "loss": 0.381,
+      "step": 4488
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.410669530970283e-05,
+      "loss": 0.3635,
+      "step": 4489
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.41031149301826e-05,
+      "loss": 0.3068,
+      "step": 4490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4099534550662374e-05,
+      "loss": 0.3575,
+      "step": 4491
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.409595417114214e-05,
+      "loss": 0.3436,
+      "step": 4492
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4092373791621914e-05,
+      "loss": 0.2329,
+      "step": 4493
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.408879341210168e-05,
+      "loss": 0.2479,
+      "step": 4494
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4085213032581453e-05,
+      "loss": 0.2557,
+      "step": 4495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.408163265306123e-05,
+      "loss": 0.38,
+      "step": 4496
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.407805227354099e-05,
+      "loss": 0.6363,
+      "step": 4497
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.4074471894020766e-05,
+      "loss": 0.3905,
+      "step": 4498
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.407089151450054e-05,
+      "loss": 0.2991,
+      "step": 4499
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.406731113498031e-05,
+      "loss": 0.2584,
+      "step": 4500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4063730755460086e-05,
+      "loss": 0.4525,
+      "step": 4501
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.406015037593985e-05,
+      "loss": 0.3398,
+      "step": 4502
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4056569996419625e-05,
+      "loss": 0.3899,
+      "step": 4503
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.405298961689939e-05,
+      "loss": 0.407,
+      "step": 4504
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4049409237379165e-05,
+      "loss": 0.5002,
+      "step": 4505
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.404582885785894e-05,
+      "loss": 0.2622,
+      "step": 4506
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.4042248478338705e-05,
+      "loss": 0.3018,
+      "step": 4507
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.403866809881848e-05,
+      "loss": 0.17,
+      "step": 4508
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.403508771929825e-05,
+      "loss": 0.3963,
+      "step": 4509
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.403150733977802e-05,
+      "loss": 0.3447,
+      "step": 4510
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.402792696025779e-05,
+      "loss": 0.3435,
+      "step": 4511
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.402434658073756e-05,
+      "loss": 0.3407,
+      "step": 4512
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.402076620121733e-05,
+      "loss": 0.4172,
+      "step": 4513
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.40171858216971e-05,
+      "loss": 0.4596,
+      "step": 4514
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.401360544217687e-05,
+      "loss": 0.2017,
+      "step": 4515
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.401002506265664e-05,
+      "loss": 0.4443,
+      "step": 4516
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400644468313641e-05,
+      "loss": 0.4053,
+      "step": 4517
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.400286430361618e-05,
+      "loss": 0.3857,
+      "step": 4518
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3999283924095956e-05,
+      "loss": 0.4806,
+      "step": 4519
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.399570354457572e-05,
+      "loss": 0.517,
+      "step": 4520
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3992123165055495e-05,
+      "loss": 0.3433,
+      "step": 4521
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398854278553527e-05,
+      "loss": 0.4286,
+      "step": 4522
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.398496240601504e-05,
+      "loss": 0.3075,
+      "step": 4523
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3981382026494815e-05,
+      "loss": 0.407,
+      "step": 4524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.397780164697458e-05,
+      "loss": 0.3744,
+      "step": 4525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.3974221267454354e-05,
+      "loss": 0.2968,
+      "step": 4526
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.397064088793412e-05,
+      "loss": 0.3226,
+      "step": 4527
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3967060508413894e-05,
+      "loss": 0.2817,
+      "step": 4528
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.396348012889367e-05,
+      "loss": 0.3533,
+      "step": 4529
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3959899749373434e-05,
+      "loss": 0.3662,
+      "step": 4530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.395631936985321e-05,
+      "loss": 0.3335,
+      "step": 4531
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.395273899033298e-05,
+      "loss": 0.4681,
+      "step": 4532
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3949158610812746e-05,
+      "loss": 0.2908,
+      "step": 4533
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.394557823129252e-05,
+      "loss": 0.256,
+      "step": 4534
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3941997851772286e-05,
+      "loss": 0.5005,
+      "step": 4535
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.393841747225206e-05,
+      "loss": 0.2934,
+      "step": 4536
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.393483709273183e-05,
+      "loss": 0.2397,
+      "step": 4537
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.39312567132116e-05,
+      "loss": 0.2886,
+      "step": 4538
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.392767633369137e-05,
+      "loss": 0.4329,
+      "step": 4539
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3924095954171145e-05,
+      "loss": 0.5054,
+      "step": 4540
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.392051557465091e-05,
+      "loss": 0.4921,
+      "step": 4541
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3916935195130685e-05,
+      "loss": 0.3941,
+      "step": 4542
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.391335481561045e-05,
+      "loss": 0.4301,
+      "step": 4543
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3909774436090224e-05,
+      "loss": 0.3489,
+      "step": 4544
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390619405657e-05,
+      "loss": 0.3257,
+      "step": 4545
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.390261367704977e-05,
+      "loss": 0.2508,
+      "step": 4546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3899033297529544e-05,
+      "loss": 0.268,
+      "step": 4547
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.389545291800931e-05,
+      "loss": 0.4411,
+      "step": 4548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3891872538489084e-05,
+      "loss": 0.5142,
+      "step": 4549
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.388829215896886e-05,
+      "loss": 0.3978,
+      "step": 4550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.388471177944862e-05,
+      "loss": 0.3761,
+      "step": 4551
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3881131399928396e-05,
+      "loss": 0.4543,
+      "step": 4552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.387755102040816e-05,
+      "loss": 0.4014,
+      "step": 4553
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.3873970640887936e-05,
+      "loss": 0.4398,
+      "step": 4554
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.387039026136771e-05,
+      "loss": 0.4592,
+      "step": 4555
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3866809881847476e-05,
+      "loss": 0.3955,
+      "step": 4556
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.386322950232725e-05,
+      "loss": 0.3432,
+      "step": 4557
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.385964912280702e-05,
+      "loss": 0.6528,
+      "step": 4558
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.385606874328679e-05,
+      "loss": 0.3504,
+      "step": 4559
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.385248836376656e-05,
+      "loss": 0.3236,
+      "step": 4560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.384890798424633e-05,
+      "loss": 0.4379,
+      "step": 4561
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.38453276047261e-05,
+      "loss": 0.5673,
+      "step": 4562
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3841747225205874e-05,
+      "loss": 0.4252,
+      "step": 4563
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.383816684568564e-05,
+      "loss": 0.3614,
+      "step": 4564
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3834586466165414e-05,
+      "loss": 0.3803,
+      "step": 4565
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.383100608664519e-05,
+      "loss": 0.4516,
+      "step": 4566
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3827425707124954e-05,
+      "loss": 0.4401,
+      "step": 4567
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.382384532760473e-05,
+      "loss": 0.4819,
+      "step": 4568
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.38202649480845e-05,
+      "loss": 0.3299,
+      "step": 4569
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.381668456856427e-05,
+      "loss": 0.4019,
+      "step": 4570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.381310418904404e-05,
+      "loss": 0.3422,
+      "step": 4571
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.380952380952381e-05,
+      "loss": 0.2569,
+      "step": 4572
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3805943430003586e-05,
+      "loss": 0.4795,
+      "step": 4573
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.380236305048335e-05,
+      "loss": 0.4736,
+      "step": 4574
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3798782670963125e-05,
+      "loss": 0.3915,
+      "step": 4575
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.37952022914429e-05,
+      "loss": 0.4887,
+      "step": 4576
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3791621911922665e-05,
+      "loss": 0.4645,
+      "step": 4577
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.378804153240244e-05,
+      "loss": 0.4188,
+      "step": 4578
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.3784461152882205e-05,
+      "loss": 0.3878,
+      "step": 4579
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.378088077336198e-05,
+      "loss": 0.3612,
+      "step": 4580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377730039384175e-05,
+      "loss": 0.3891,
+      "step": 4581
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.377372001432152e-05,
+      "loss": 0.4138,
+      "step": 4582
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.377013963480129e-05,
+      "loss": 0.3625,
+      "step": 4583
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3766559255281064e-05,
+      "loss": 0.3722,
+      "step": 4584
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.376297887576083e-05,
+      "loss": 0.3696,
+      "step": 4585
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3759398496240603e-05,
+      "loss": 0.2342,
+      "step": 4586
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.375581811672037e-05,
+      "loss": 0.4244,
+      "step": 4587
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.375223773720014e-05,
+      "loss": 0.273,
+      "step": 4588
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3748657357679916e-05,
+      "loss": 0.2993,
+      "step": 4589
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.374507697815968e-05,
+      "loss": 0.3101,
+      "step": 4590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3741496598639456e-05,
+      "loss": 0.3793,
+      "step": 4591
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.373791621911922e-05,
+      "loss": 0.3725,
+      "step": 4592
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3734335839599e-05,
+      "loss": 0.4079,
+      "step": 4593
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3730755460078775e-05,
+      "loss": 0.3423,
+      "step": 4594
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.372717508055854e-05,
+      "loss": 0.3224,
+      "step": 4595
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3723594701038315e-05,
+      "loss": 0.4415,
+      "step": 4596
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.372001432151808e-05,
+      "loss": 0.3953,
+      "step": 4597
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3716433941997855e-05,
+      "loss": 0.3141,
+      "step": 4598
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.371285356247763e-05,
+      "loss": 0.3117,
+      "step": 4599
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3709273182957394e-05,
+      "loss": 0.291,
+      "step": 4600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370569280343717e-05,
+      "loss": 0.5576,
+      "step": 4601
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.370211242391694e-05,
+      "loss": 0.4531,
+      "step": 4602
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369853204439671e-05,
+      "loss": 0.4805,
+      "step": 4603
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.369495166487648e-05,
+      "loss": 0.3635,
+      "step": 4604
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.3691371285356247e-05,
+      "loss": 0.3656,
+      "step": 4605
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368779090583602e-05,
+      "loss": 0.4327,
+      "step": 4606
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368421052631579e-05,
+      "loss": 0.4103,
+      "step": 4607
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.368063014679556e-05,
+      "loss": 0.3063,
+      "step": 4608
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.367704976727533e-05,
+      "loss": 0.3221,
+      "step": 4609
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.36734693877551e-05,
+      "loss": 0.3089,
+      "step": 4610
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.366988900823487e-05,
+      "loss": 0.4028,
+      "step": 4611
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3666308628714645e-05,
+      "loss": 0.3112,
+      "step": 4612
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.366272824919441e-05,
+      "loss": 0.3981,
+      "step": 4613
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3659147869674185e-05,
+      "loss": 0.3626,
+      "step": 4614
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.365556749015396e-05,
+      "loss": 0.3918,
+      "step": 4615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.365198711063373e-05,
+      "loss": 0.4775,
+      "step": 4616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3648406731113504e-05,
+      "loss": 0.3502,
+      "step": 4617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.364482635159327e-05,
+      "loss": 0.3469,
+      "step": 4618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3641245972073044e-05,
+      "loss": 0.3527,
+      "step": 4619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.363766559255282e-05,
+      "loss": 0.3068,
+      "step": 4620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3634085213032584e-05,
+      "loss": 0.4117,
+      "step": 4621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.363050483351236e-05,
+      "loss": 0.2983,
+      "step": 4622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.362692445399212e-05,
+      "loss": 0.4207,
+      "step": 4623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3623344074471896e-05,
+      "loss": 0.3934,
+      "step": 4624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.361976369495167e-05,
+      "loss": 0.394,
+      "step": 4625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3616183315431436e-05,
+      "loss": 0.4493,
+      "step": 4626
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.361260293591121e-05,
+      "loss": 0.3601,
+      "step": 4627
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3609022556390976e-05,
+      "loss": 0.3456,
+      "step": 4628
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360544217687075e-05,
+      "loss": 0.3113,
+      "step": 4629
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.360186179735052e-05,
+      "loss": 0.2344,
+      "step": 4630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359828141783029e-05,
+      "loss": 0.5042,
+      "step": 4631
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.359470103831006e-05,
+      "loss": 0.4363,
+      "step": 4632
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3591120658789835e-05,
+      "loss": 0.3217,
+      "step": 4633
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.35875402792696e-05,
+      "loss": 0.3541,
+      "step": 4634
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3583959899749374e-05,
+      "loss": 0.4786,
+      "step": 4635
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.358037952022914e-05,
+      "loss": 0.4404,
+      "step": 4636
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.3576799140708914e-05,
+      "loss": 0.3289,
+      "step": 4637
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.357321876118869e-05,
+      "loss": 0.4197,
+      "step": 4638
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.356963838166846e-05,
+      "loss": 0.4883,
+      "step": 4639
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3566058002148234e-05,
+      "loss": 0.4886,
+      "step": 4640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3562477622628e-05,
+      "loss": 0.4406,
+      "step": 4641
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.355889724310777e-05,
+      "loss": 0.3244,
+      "step": 4642
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3555316863587546e-05,
+      "loss": 0.3688,
+      "step": 4643
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.355173648406731e-05,
+      "loss": 0.334,
+      "step": 4644
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3548156104547086e-05,
+      "loss": 0.326,
+      "step": 4645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.354457572502685e-05,
+      "loss": 0.3519,
+      "step": 4646
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3540995345506626e-05,
+      "loss": 0.4675,
+      "step": 4647
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.35374149659864e-05,
+      "loss": 0.3229,
+      "step": 4648
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3533834586466165e-05,
+      "loss": 0.2546,
+      "step": 4649
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.353025420694594e-05,
+      "loss": 0.2754,
+      "step": 4650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.352667382742571e-05,
+      "loss": 0.2927,
+      "step": 4651
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.352309344790548e-05,
+      "loss": 0.3271,
+      "step": 4652
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.351951306838525e-05,
+      "loss": 0.3389,
+      "step": 4653
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.351593268886502e-05,
+      "loss": 0.3422,
+      "step": 4654
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.351235230934479e-05,
+      "loss": 0.3626,
+      "step": 4655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3508771929824564e-05,
+      "loss": 0.4067,
+      "step": 4656
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.350519155030433e-05,
+      "loss": 0.3738,
+      "step": 4657
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3501611170784104e-05,
+      "loss": 0.2605,
+      "step": 4658
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349803079126388e-05,
+      "loss": 0.3346,
+      "step": 4659
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.349445041174364e-05,
+      "loss": 0.357,
+      "step": 4660
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3490870032223416e-05,
+      "loss": 0.454,
+      "step": 4661
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.348728965270318e-05,
+      "loss": 0.4005,
+      "step": 4662
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.348370927318296e-05,
+      "loss": 0.3352,
+      "step": 4663
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.348012889366273e-05,
+      "loss": 0.2826,
+      "step": 4664
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.34765485141425e-05,
+      "loss": 0.4547,
+      "step": 4665
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.3472968134622275e-05,
+      "loss": 0.3487,
+      "step": 4666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.346938775510204e-05,
+      "loss": 0.4337,
+      "step": 4667
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3465807375581815e-05,
+      "loss": 0.4363,
+      "step": 4668
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.346222699606159e-05,
+      "loss": 0.232,
+      "step": 4669
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3458646616541355e-05,
+      "loss": 0.4089,
+      "step": 4670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.345506623702113e-05,
+      "loss": 0.4738,
+      "step": 4671
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3451485857500894e-05,
+      "loss": 0.4057,
+      "step": 4672
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.344790547798067e-05,
+      "loss": 0.4653,
+      "step": 4673
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.344432509846044e-05,
+      "loss": 0.558,
+      "step": 4674
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.344074471894021e-05,
+      "loss": 0.3422,
+      "step": 4675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.343716433941998e-05,
+      "loss": 0.4618,
+      "step": 4676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3433583959899753e-05,
+      "loss": 0.3061,
+      "step": 4677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.343000358037952e-05,
+      "loss": 0.328,
+      "step": 4678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.342642320085929e-05,
+      "loss": 0.2681,
+      "step": 4679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.342284282133906e-05,
+      "loss": 0.4662,
+      "step": 4680
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.341926244181883e-05,
+      "loss": 0.3861,
+      "step": 4681
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3415682062298606e-05,
+      "loss": 0.4082,
+      "step": 4682
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.341210168277837e-05,
+      "loss": 0.4241,
+      "step": 4683
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3408521303258145e-05,
+      "loss": 0.4521,
+      "step": 4684
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340494092373791e-05,
+      "loss": 0.4057,
+      "step": 4685
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.340136054421769e-05,
+      "loss": 0.2593,
+      "step": 4686
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3397780164697465e-05,
+      "loss": 0.3815,
+      "step": 4687
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.339419978517723e-05,
+      "loss": 0.3301,
+      "step": 4688
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3390619405657005e-05,
+      "loss": 0.3185,
+      "step": 4689
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.338703902613677e-05,
+      "loss": 0.3859,
+      "step": 4690
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3383458646616544e-05,
+      "loss": 0.4471,
+      "step": 4691
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.337987826709632e-05,
+      "loss": 0.425,
+      "step": 4692
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.3376297887576084e-05,
+      "loss": 0.3254,
+      "step": 4693
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.337271750805586e-05,
+      "loss": 0.417,
+      "step": 4694
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.336913712853563e-05,
+      "loss": 0.3764,
+      "step": 4695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3365556749015397e-05,
+      "loss": 0.3246,
+      "step": 4696
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.336197636949517e-05,
+      "loss": 0.3298,
+      "step": 4697
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3358395989974936e-05,
+      "loss": 0.298,
+      "step": 4698
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.335481561045471e-05,
+      "loss": 0.3466,
+      "step": 4699
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.335123523093448e-05,
+      "loss": 0.3975,
+      "step": 4700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.334765485141425e-05,
+      "loss": 0.3256,
+      "step": 4701
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.334407447189402e-05,
+      "loss": 0.3964,
+      "step": 4702
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.334049409237379e-05,
+      "loss": 0.3263,
+      "step": 4703
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.333691371285356e-05,
+      "loss": 0.5412,
+      "step": 4704
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.485,
+      "step": 4705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.33297529538131e-05,
+      "loss": 0.2991,
+      "step": 4706
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3326172574292875e-05,
+      "loss": 0.3788,
+      "step": 4707
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.332259219477265e-05,
+      "loss": 0.3025,
+      "step": 4708
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.331901181525242e-05,
+      "loss": 0.3143,
+      "step": 4709
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3315431435732194e-05,
+      "loss": 0.3452,
+      "step": 4710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.331185105621196e-05,
+      "loss": 0.5391,
+      "step": 4711
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3308270676691734e-05,
+      "loss": 0.2612,
+      "step": 4712
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.330469029717151e-05,
+      "loss": 0.3678,
+      "step": 4713
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.330110991765127e-05,
+      "loss": 0.284,
+      "step": 4714
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3297529538131046e-05,
+      "loss": 0.4819,
+      "step": 4715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.329394915861081e-05,
+      "loss": 0.4752,
+      "step": 4716
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3290368779090586e-05,
+      "loss": 0.41,
+      "step": 4717
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.328678839957036e-05,
+      "loss": 0.3073,
+      "step": 4718
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3283208020050126e-05,
+      "loss": 0.4755,
+      "step": 4719
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.32796276405299e-05,
+      "loss": 0.4348,
+      "step": 4720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.3276047261009665e-05,
+      "loss": 0.2778,
+      "step": 4721
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.327246688148944e-05,
+      "loss": 0.4167,
+      "step": 4722
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.326888650196921e-05,
+      "loss": 0.362,
+      "step": 4723
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.326530612244898e-05,
+      "loss": 0.4361,
+      "step": 4724
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.326172574292875e-05,
+      "loss": 0.3935,
+      "step": 4725
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3258145363408524e-05,
+      "loss": 0.3116,
+      "step": 4726
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.325456498388829e-05,
+      "loss": 0.2996,
+      "step": 4727
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3250984604368064e-05,
+      "loss": 0.4152,
+      "step": 4728
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.324740422484783e-05,
+      "loss": 0.278,
+      "step": 4729
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3243823845327604e-05,
+      "loss": 0.4036,
+      "step": 4730
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.324024346580738e-05,
+      "loss": 0.404,
+      "step": 4731
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.323666308628714e-05,
+      "loss": 0.3091,
+      "step": 4732
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.323308270676692e-05,
+      "loss": 0.4139,
+      "step": 4733
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.322950232724669e-05,
+      "loss": 0.4471,
+      "step": 4734
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.322592194772646e-05,
+      "loss": 0.4234,
+      "step": 4735
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3222341568206236e-05,
+      "loss": 0.2808,
+      "step": 4736
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3218761188686e-05,
+      "loss": 0.2377,
+      "step": 4737
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3215180809165776e-05,
+      "loss": 0.3919,
+      "step": 4738
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.321160042964554e-05,
+      "loss": 0.3909,
+      "step": 4739
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3208020050125315e-05,
+      "loss": 0.4623,
+      "step": 4740
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.320443967060509e-05,
+      "loss": 0.5142,
+      "step": 4741
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3200859291084855e-05,
+      "loss": 0.4915,
+      "step": 4742
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319727891156463e-05,
+      "loss": 0.3868,
+      "step": 4743
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.31936985320444e-05,
+      "loss": 0.2565,
+      "step": 4744
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.319011815252417e-05,
+      "loss": 0.261,
+      "step": 4745
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318653777300394e-05,
+      "loss": 0.3172,
+      "step": 4746
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.318295739348371e-05,
+      "loss": 0.354,
+      "step": 4747
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.317937701396348e-05,
+      "loss": 0.5211,
+      "step": 4748
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.3175796634443254e-05,
+      "loss": 0.396,
+      "step": 4749
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.317221625492302e-05,
+      "loss": 0.3894,
+      "step": 4750
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.316863587540279e-05,
+      "loss": 0.3398,
+      "step": 4751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3165055495882566e-05,
+      "loss": 0.3402,
+      "step": 4752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.316147511636233e-05,
+      "loss": 0.3644,
+      "step": 4753
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3157894736842106e-05,
+      "loss": 0.3236,
+      "step": 4754
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.315431435732187e-05,
+      "loss": 0.3048,
+      "step": 4755
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.315073397780165e-05,
+      "loss": 0.2815,
+      "step": 4756
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.314715359828142e-05,
+      "loss": 0.5172,
+      "step": 4757
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.314357321876119e-05,
+      "loss": 0.3255,
+      "step": 4758
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3139992839240965e-05,
+      "loss": 0.4464,
+      "step": 4759
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.313641245972073e-05,
+      "loss": 0.4753,
+      "step": 4760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3132832080200505e-05,
+      "loss": 0.4166,
+      "step": 4761
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.312925170068028e-05,
+      "loss": 0.3157,
+      "step": 4762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3125671321160044e-05,
+      "loss": 0.4213,
+      "step": 4763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.312209094163982e-05,
+      "loss": 0.3807,
+      "step": 4764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3118510562119584e-05,
+      "loss": 0.3506,
+      "step": 4765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.311493018259936e-05,
+      "loss": 0.1828,
+      "step": 4766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.311134980307913e-05,
+      "loss": 0.3429,
+      "step": 4767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.31077694235589e-05,
+      "loss": 0.4023,
+      "step": 4768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.310418904403867e-05,
+      "loss": 0.4118,
+      "step": 4769
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.310060866451844e-05,
+      "loss": 0.3207,
+      "step": 4770
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309702828499821e-05,
+      "loss": 0.307,
+      "step": 4771
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.309344790547798e-05,
+      "loss": 0.3319,
+      "step": 4772
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.308986752595775e-05,
+      "loss": 0.4307,
+      "step": 4773
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.308628714643752e-05,
+      "loss": 0.3304,
+      "step": 4774
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3082706766917295e-05,
+      "loss": 0.3509,
+      "step": 4775
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.307912638739706e-05,
+      "loss": 0.3462,
+      "step": 4776
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.3075546007876835e-05,
+      "loss": 0.3087,
+      "step": 4777
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.307196562835661e-05,
+      "loss": 0.257,
+      "step": 4778
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.306838524883638e-05,
+      "loss": 0.394,
+      "step": 4779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3064804869316155e-05,
+      "loss": 0.2403,
+      "step": 4780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.306122448979592e-05,
+      "loss": 0.4604,
+      "step": 4781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3057644110275694e-05,
+      "loss": 0.3308,
+      "step": 4782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.305406373075546e-05,
+      "loss": 0.4137,
+      "step": 4783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3050483351235234e-05,
+      "loss": 0.3441,
+      "step": 4784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.304690297171501e-05,
+      "loss": 0.4644,
+      "step": 4785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.304332259219477e-05,
+      "loss": 0.4611,
+      "step": 4786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3039742212674547e-05,
+      "loss": 0.5165,
+      "step": 4787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.303616183315432e-05,
+      "loss": 0.3968,
+      "step": 4788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3032581453634086e-05,
+      "loss": 0.2803,
+      "step": 4789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.302900107411386e-05,
+      "loss": 0.4503,
+      "step": 4790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3025420694593626e-05,
+      "loss": 0.3698,
+      "step": 4791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.30218403150734e-05,
+      "loss": 0.4561,
+      "step": 4792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.301825993555317e-05,
+      "loss": 0.287,
+      "step": 4793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.301467955603294e-05,
+      "loss": 0.3009,
+      "step": 4794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.301109917651271e-05,
+      "loss": 0.2775,
+      "step": 4795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.300751879699248e-05,
+      "loss": 0.4323,
+      "step": 4796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.300393841747225e-05,
+      "loss": 0.3588,
+      "step": 4797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.3000358037952025e-05,
+      "loss": 0.287,
+      "step": 4798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.299677765843179e-05,
+      "loss": 0.2857,
+      "step": 4799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2993197278911564e-05,
+      "loss": 0.3398,
+      "step": 4800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.298961689939134e-05,
+      "loss": 0.4153,
+      "step": 4801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2986036519871104e-05,
+      "loss": 0.2597,
+      "step": 4802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2982456140350884e-05,
+      "loss": 0.3917,
+      "step": 4803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297887576083065e-05,
+      "loss": 0.391,
+      "step": 4804
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.297529538131042e-05,
+      "loss": 0.3167,
+      "step": 4805
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.2971715001790196e-05,
+      "loss": 0.3252,
+      "step": 4806
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.296813462226996e-05,
+      "loss": 0.4113,
+      "step": 4807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2964554242749736e-05,
+      "loss": 0.4482,
+      "step": 4808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.29609738632295e-05,
+      "loss": 0.2379,
+      "step": 4809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2957393483709276e-05,
+      "loss": 0.3201,
+      "step": 4810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.295381310418905e-05,
+      "loss": 0.3356,
+      "step": 4811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2950232724668815e-05,
+      "loss": 0.315,
+      "step": 4812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.294665234514859e-05,
+      "loss": 0.5344,
+      "step": 4813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2943071965628355e-05,
+      "loss": 0.2908,
+      "step": 4814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.293949158610813e-05,
+      "loss": 0.3616,
+      "step": 4815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.29359112065879e-05,
+      "loss": 0.377,
+      "step": 4816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.293233082706767e-05,
+      "loss": 0.364,
+      "step": 4817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.292875044754744e-05,
+      "loss": 0.2986,
+      "step": 4818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2925170068027214e-05,
+      "loss": 0.4154,
+      "step": 4819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.292158968850698e-05,
+      "loss": 0.4982,
+      "step": 4820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2918009308986754e-05,
+      "loss": 0.4136,
+      "step": 4821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.291442892946652e-05,
+      "loss": 0.4935,
+      "step": 4822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.291084854994629e-05,
+      "loss": 0.3517,
+      "step": 4823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2907268170426066e-05,
+      "loss": 0.2175,
+      "step": 4824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.290368779090583e-05,
+      "loss": 0.5778,
+      "step": 4825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.290010741138561e-05,
+      "loss": 0.4367,
+      "step": 4826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289652703186538e-05,
+      "loss": 0.4024,
+      "step": 4827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.289294665234515e-05,
+      "loss": 0.3263,
+      "step": 4828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2889366272824926e-05,
+      "loss": 0.367,
+      "step": 4829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.288578589330469e-05,
+      "loss": 0.3221,
+      "step": 4830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2882205513784465e-05,
+      "loss": 0.4469,
+      "step": 4831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287862513426423e-05,
+      "loss": 0.4615,
+      "step": 4832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2875044754744005e-05,
+      "loss": 0.4221,
+      "step": 4833
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.287146437522378e-05,
+      "loss": 0.3579,
+      "step": 4834
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.2867883995703544e-05,
+      "loss": 0.4093,
+      "step": 4835
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.286430361618332e-05,
+      "loss": 0.3175,
+      "step": 4836
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.286072323666309e-05,
+      "loss": 0.4429,
+      "step": 4837
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.285714285714286e-05,
+      "loss": 0.3777,
+      "step": 4838
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.285356247762263e-05,
+      "loss": 0.4364,
+      "step": 4839
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.28499820981024e-05,
+      "loss": 0.3691,
+      "step": 4840
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.284640171858217e-05,
+      "loss": 0.3053,
+      "step": 4841
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.284282133906194e-05,
+      "loss": 0.3758,
+      "step": 4842
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.283924095954171e-05,
+      "loss": 0.2804,
+      "step": 4843
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.283566058002148e-05,
+      "loss": 0.3366,
+      "step": 4844
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2832080200501256e-05,
+      "loss": 0.2592,
+      "step": 4845
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.282849982098102e-05,
+      "loss": 0.3889,
+      "step": 4846
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2824919441460796e-05,
+      "loss": 0.3839,
+      "step": 4847
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.282133906194056e-05,
+      "loss": 0.2552,
+      "step": 4848
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2817758682420335e-05,
+      "loss": 0.4252,
+      "step": 4849
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.281417830290011e-05,
+      "loss": 0.4185,
+      "step": 4850
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.281059792337988e-05,
+      "loss": 0.3042,
+      "step": 4851
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2807017543859655e-05,
+      "loss": 0.34,
+      "step": 4852
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.280343716433942e-05,
+      "loss": 0.5027,
+      "step": 4853
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2799856784819194e-05,
+      "loss": 0.2464,
+      "step": 4854
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.279627640529897e-05,
+      "loss": 0.3314,
+      "step": 4855
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2792696025778734e-05,
+      "loss": 0.4406,
+      "step": 4856
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278911564625851e-05,
+      "loss": 0.28,
+      "step": 4857
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2785535266738273e-05,
+      "loss": 0.2709,
+      "step": 4858
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.278195488721805e-05,
+      "loss": 0.3692,
+      "step": 4859
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277837450769782e-05,
+      "loss": 0.3683,
+      "step": 4860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.2774794128177586e-05,
+      "loss": 0.4529,
+      "step": 4861
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.277121374865736e-05,
+      "loss": 0.5235,
+      "step": 4862
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.276763336913713e-05,
+      "loss": 0.4328,
+      "step": 4863
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.27640529896169e-05,
+      "loss": 0.3309,
+      "step": 4864
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.276047261009667e-05,
+      "loss": 0.4386,
+      "step": 4865
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.275689223057644e-05,
+      "loss": 0.3392,
+      "step": 4866
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.275331185105621e-05,
+      "loss": 0.2341,
+      "step": 4867
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2749731471535985e-05,
+      "loss": 0.2408,
+      "step": 4868
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.274615109201575e-05,
+      "loss": 0.3151,
+      "step": 4869
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2742570712495525e-05,
+      "loss": 0.375,
+      "step": 4870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.27389903329753e-05,
+      "loss": 0.449,
+      "step": 4871
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2735409953455064e-05,
+      "loss": 0.3088,
+      "step": 4872
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2731829573934844e-05,
+      "loss": 0.2779,
+      "step": 4873
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.272824919441461e-05,
+      "loss": 0.3206,
+      "step": 4874
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2724668814894384e-05,
+      "loss": 0.2134,
+      "step": 4875
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.272108843537415e-05,
+      "loss": 0.2575,
+      "step": 4876
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2717508055853923e-05,
+      "loss": 0.3885,
+      "step": 4877
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2713927676333697e-05,
+      "loss": 0.4924,
+      "step": 4878
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.271034729681346e-05,
+      "loss": 0.4374,
+      "step": 4879
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2706766917293236e-05,
+      "loss": 0.3906,
+      "step": 4880
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.270318653777301e-05,
+      "loss": 0.5142,
+      "step": 4881
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2699606158252776e-05,
+      "loss": 0.4977,
+      "step": 4882
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.269602577873255e-05,
+      "loss": 0.4789,
+      "step": 4883
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2692445399212315e-05,
+      "loss": 0.4445,
+      "step": 4884
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268886501969209e-05,
+      "loss": 0.3425,
+      "step": 4885
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268528464017186e-05,
+      "loss": 0.5187,
+      "step": 4886
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.268170426065163e-05,
+      "loss": 0.3275,
+      "step": 4887
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.26781238811314e-05,
+      "loss": 0.3247,
+      "step": 4888
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2674543501611175e-05,
+      "loss": 0.4356,
+      "step": 4889
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.267096312209094e-05,
+      "loss": 0.4338,
+      "step": 4890
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.2667382742570714e-05,
+      "loss": 0.3711,
+      "step": 4891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.266380236305048e-05,
+      "loss": 0.3949,
+      "step": 4892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2660221983530254e-05,
+      "loss": 0.3998,
+      "step": 4893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.265664160401003e-05,
+      "loss": 0.226,
+      "step": 4894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.265306122448979e-05,
+      "loss": 0.2569,
+      "step": 4895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.264948084496957e-05,
+      "loss": 0.3755,
+      "step": 4896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.264590046544934e-05,
+      "loss": 0.3968,
+      "step": 4897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.264232008592911e-05,
+      "loss": 0.289,
+      "step": 4898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2638739706408886e-05,
+      "loss": 0.3166,
+      "step": 4899
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.263515932688865e-05,
+      "loss": 0.4681,
+      "step": 4900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2631578947368426e-05,
+      "loss": 0.3738,
+      "step": 4901
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.262799856784819e-05,
+      "loss": 0.3459,
+      "step": 4902
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2624418188327965e-05,
+      "loss": 0.3202,
+      "step": 4903
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.262083780880774e-05,
+      "loss": 0.3138,
+      "step": 4904
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2617257429287505e-05,
+      "loss": 0.3736,
+      "step": 4905
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.261367704976728e-05,
+      "loss": 0.5796,
+      "step": 4906
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2610096670247044e-05,
+      "loss": 0.4491,
+      "step": 4907
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.260651629072682e-05,
+      "loss": 0.2881,
+      "step": 4908
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.260293591120659e-05,
+      "loss": 0.4406,
+      "step": 4909
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.259935553168636e-05,
+      "loss": 0.2813,
+      "step": 4910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.259577515216613e-05,
+      "loss": 0.2851,
+      "step": 4911
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2592194772645904e-05,
+      "loss": 0.3383,
+      "step": 4912
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258861439312567e-05,
+      "loss": 0.3356,
+      "step": 4913
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258503401360544e-05,
+      "loss": 0.3705,
+      "step": 4914
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.258145363408521e-05,
+      "loss": 0.4009,
+      "step": 4915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257787325456498e-05,
+      "loss": 0.2894,
+      "step": 4916
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2574292875044756e-05,
+      "loss": 0.4404,
+      "step": 4917
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.257071249552452e-05,
+      "loss": 0.3854,
+      "step": 4918
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.2567132116004296e-05,
+      "loss": 0.4415,
+      "step": 4919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.256355173648407e-05,
+      "loss": 0.3353,
+      "step": 4920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.255997135696384e-05,
+      "loss": 0.3573,
+      "step": 4921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2556390977443615e-05,
+      "loss": 0.3618,
+      "step": 4922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.255281059792338e-05,
+      "loss": 0.3733,
+      "step": 4923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2549230218403155e-05,
+      "loss": 0.4132,
+      "step": 4924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.254564983888292e-05,
+      "loss": 0.4247,
+      "step": 4925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2542069459362694e-05,
+      "loss": 0.2232,
+      "step": 4926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.253848907984247e-05,
+      "loss": 0.3227,
+      "step": 4927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2534908700322234e-05,
+      "loss": 0.3258,
+      "step": 4928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.253132832080201e-05,
+      "loss": 0.3208,
+      "step": 4929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.252774794128178e-05,
+      "loss": 0.2339,
+      "step": 4930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.252416756176155e-05,
+      "loss": 0.277,
+      "step": 4931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.252058718224132e-05,
+      "loss": 0.2971,
+      "step": 4932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2517006802721086e-05,
+      "loss": 0.3826,
+      "step": 4933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.251342642320086e-05,
+      "loss": 0.3355,
+      "step": 4934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.250984604368063e-05,
+      "loss": 0.3885,
+      "step": 4935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.25062656641604e-05,
+      "loss": 0.3869,
+      "step": 4936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.250268528464017e-05,
+      "loss": 0.218,
+      "step": 4937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2499104905119946e-05,
+      "loss": 0.4225,
+      "step": 4938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.249552452559971e-05,
+      "loss": 0.5699,
+      "step": 4939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2491944146079485e-05,
+      "loss": 0.2559,
+      "step": 4940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.248836376655925e-05,
+      "loss": 0.3266,
+      "step": 4941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2484783387039025e-05,
+      "loss": 0.3182,
+      "step": 4942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.24812030075188e-05,
+      "loss": 0.4138,
+      "step": 4943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247762262799857e-05,
+      "loss": 0.3796,
+      "step": 4944
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2474042248478344e-05,
+      "loss": 0.4512,
+      "step": 4945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.247046186895811e-05,
+      "loss": 0.2935,
+      "step": 4946
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.2466881489437884e-05,
+      "loss": 0.3688,
+      "step": 4947
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.246330110991766e-05,
+      "loss": 0.2329,
+      "step": 4948
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2459720730397423e-05,
+      "loss": 0.414,
+      "step": 4949
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.24561403508772e-05,
+      "loss": 0.282,
+      "step": 4950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.245255997135696e-05,
+      "loss": 0.3347,
+      "step": 4951
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2448979591836736e-05,
+      "loss": 0.402,
+      "step": 4952
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.244539921231651e-05,
+      "loss": 0.4968,
+      "step": 4953
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2441818832796276e-05,
+      "loss": 0.4,
+      "step": 4954
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.243823845327605e-05,
+      "loss": 0.4225,
+      "step": 4955
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.243465807375582e-05,
+      "loss": 0.4128,
+      "step": 4956
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.243107769423559e-05,
+      "loss": 0.4394,
+      "step": 4957
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.242749731471536e-05,
+      "loss": 0.4079,
+      "step": 4958
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.242391693519513e-05,
+      "loss": 0.3798,
+      "step": 4959
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.24203365556749e-05,
+      "loss": 0.359,
+      "step": 4960
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2416756176154675e-05,
+      "loss": 0.3845,
+      "step": 4961
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.241317579663444e-05,
+      "loss": 0.3526,
+      "step": 4962
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2409595417114214e-05,
+      "loss": 0.3622,
+      "step": 4963
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.240601503759399e-05,
+      "loss": 0.5079,
+      "step": 4964
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2402434658073754e-05,
+      "loss": 0.4772,
+      "step": 4965
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2398854278553534e-05,
+      "loss": 0.2947,
+      "step": 4966
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.23952738990333e-05,
+      "loss": 0.3228,
+      "step": 4967
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2391693519513073e-05,
+      "loss": 0.4461,
+      "step": 4968
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238811313999284e-05,
+      "loss": 0.4188,
+      "step": 4969
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.238453276047261e-05,
+      "loss": 0.3408,
+      "step": 4970
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2380952380952386e-05,
+      "loss": 0.3408,
+      "step": 4971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.237737200143215e-05,
+      "loss": 0.4851,
+      "step": 4972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2373791621911926e-05,
+      "loss": 0.3255,
+      "step": 4973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.23702112423917e-05,
+      "loss": 0.515,
+      "step": 4974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.2366630862871465e-05,
+      "loss": 0.342,
+      "step": 4975
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.236305048335124e-05,
+      "loss": 0.228,
+      "step": 4976
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2359470103831005e-05,
+      "loss": 0.2796,
+      "step": 4977
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.235588972431078e-05,
+      "loss": 0.4372,
+      "step": 4978
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.235230934479055e-05,
+      "loss": 0.3294,
+      "step": 4979
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.234872896527032e-05,
+      "loss": 0.2824,
+      "step": 4980
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.234514858575009e-05,
+      "loss": 0.2678,
+      "step": 4981
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2341568206229864e-05,
+      "loss": 0.4991,
+      "step": 4982
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.233798782670963e-05,
+      "loss": 0.3068,
+      "step": 4983
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2334407447189404e-05,
+      "loss": 0.3498,
+      "step": 4984
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.233082706766917e-05,
+      "loss": 0.2329,
+      "step": 4985
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.232724668814894e-05,
+      "loss": 0.351,
+      "step": 4986
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2323666308628717e-05,
+      "loss": 0.3351,
+      "step": 4987
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.232008592910848e-05,
+      "loss": 0.3718,
+      "step": 4988
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2316505549588256e-05,
+      "loss": 0.3942,
+      "step": 4989
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.231292517006803e-05,
+      "loss": 0.3828,
+      "step": 4990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.23093447905478e-05,
+      "loss": 0.3414,
+      "step": 4991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2305764411027576e-05,
+      "loss": 0.2667,
+      "step": 4992
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.230218403150734e-05,
+      "loss": 0.37,
+      "step": 4993
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2298603651987115e-05,
+      "loss": 0.397,
+      "step": 4994
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.229502327246688e-05,
+      "loss": 0.4774,
+      "step": 4995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2291442892946655e-05,
+      "loss": 0.2872,
+      "step": 4996
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228786251342643e-05,
+      "loss": 0.4364,
+      "step": 4997
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2284282133906194e-05,
+      "loss": 0.3233,
+      "step": 4998
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.228070175438597e-05,
+      "loss": 0.4854,
+      "step": 4999
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227712137486574e-05,
+      "loss": 0.3761,
+      "step": 5000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.227354099534551e-05,
+      "loss": 0.2977,
+      "step": 5001
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.226996061582528e-05,
+      "loss": 0.4156,
+      "step": 5002
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.226638023630505e-05,
+      "loss": 0.3732,
+      "step": 5003
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.226279985678482e-05,
+      "loss": 0.4318,
+      "step": 5004
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.225921947726459e-05,
+      "loss": 0.232,
+      "step": 5005
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.225563909774436e-05,
+      "loss": 0.3557,
+      "step": 5006
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.225205871822413e-05,
+      "loss": 0.3671,
+      "step": 5007
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.22484783387039e-05,
+      "loss": 0.319,
+      "step": 5008
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.224489795918367e-05,
+      "loss": 0.4201,
+      "step": 5009
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2241317579663446e-05,
+      "loss": 0.3068,
+      "step": 5010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.223773720014321e-05,
+      "loss": 0.3478,
+      "step": 5011
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2234156820622985e-05,
+      "loss": 0.3173,
+      "step": 5012
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.223057644110276e-05,
+      "loss": 0.5327,
+      "step": 5013
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.222699606158253e-05,
+      "loss": 0.4249,
+      "step": 5014
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2223415682062305e-05,
+      "loss": 0.2903,
+      "step": 5015
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.221983530254207e-05,
+      "loss": 0.2853,
+      "step": 5016
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2216254923021844e-05,
+      "loss": 0.4092,
+      "step": 5017
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.221267454350161e-05,
+      "loss": 0.2773,
+      "step": 5018
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2209094163981384e-05,
+      "loss": 0.2165,
+      "step": 5019
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.220551378446116e-05,
+      "loss": 0.2823,
+      "step": 5020
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2201933404940924e-05,
+      "loss": 0.3993,
+      "step": 5021
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.21983530254207e-05,
+      "loss": 0.4144,
+      "step": 5022
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.219477264590047e-05,
+      "loss": 0.5655,
+      "step": 5023
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2191192266380236e-05,
+      "loss": 0.2952,
+      "step": 5024
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218761188686001e-05,
+      "loss": 0.2958,
+      "step": 5025
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2184031507339776e-05,
+      "loss": 0.2706,
+      "step": 5026
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.218045112781955e-05,
+      "loss": 0.4432,
+      "step": 5027
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.217687074829932e-05,
+      "loss": 0.2884,
+      "step": 5028
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.217329036877909e-05,
+      "loss": 0.4009,
+      "step": 5029
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.216970998925886e-05,
+      "loss": 0.3224,
+      "step": 5030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2166129609738635e-05,
+      "loss": 0.2896,
+      "step": 5031
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.21625492302184e-05,
+      "loss": 0.2778,
+      "step": 5032
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2158968850698175e-05,
+      "loss": 0.4491,
+      "step": 5033
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.215538847117794e-05,
+      "loss": 0.3412,
+      "step": 5034
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2151808091657714e-05,
+      "loss": 0.2796,
+      "step": 5035
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.214822771213749e-05,
+      "loss": 0.4036,
+      "step": 5036
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.214464733261726e-05,
+      "loss": 0.4095,
+      "step": 5037
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2141066953097034e-05,
+      "loss": 0.3511,
+      "step": 5038
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.21374865735768e-05,
+      "loss": 0.3734,
+      "step": 5039
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2133906194056573e-05,
+      "loss": 0.345,
+      "step": 5040
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.213032581453635e-05,
+      "loss": 0.3793,
+      "step": 5041
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.212674543501611e-05,
+      "loss": 0.2758,
+      "step": 5042
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2123165055495886e-05,
+      "loss": 0.401,
+      "step": 5043
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.211958467597565e-05,
+      "loss": 0.2996,
+      "step": 5044
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2116004296455426e-05,
+      "loss": 0.3291,
+      "step": 5045
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.21124239169352e-05,
+      "loss": 0.2817,
+      "step": 5046
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2108843537414965e-05,
+      "loss": 0.3109,
+      "step": 5047
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.210526315789474e-05,
+      "loss": 0.4289,
+      "step": 5048
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.210168277837451e-05,
+      "loss": 0.3393,
+      "step": 5049
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209810239885428e-05,
+      "loss": 0.2937,
+      "step": 5050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209452201933405e-05,
+      "loss": 0.3488,
+      "step": 5051
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.209094163981382e-05,
+      "loss": 0.3042,
+      "step": 5052
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208736126029359e-05,
+      "loss": 0.416,
+      "step": 5053
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2083780880773364e-05,
+      "loss": 0.5053,
+      "step": 5054
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.208020050125313e-05,
+      "loss": 0.3342,
+      "step": 5055
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2076620121732904e-05,
+      "loss": 0.3813,
+      "step": 5056
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.207303974221268e-05,
+      "loss": 0.2728,
+      "step": 5057
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.2069459362692443e-05,
+      "loss": 0.4149,
+      "step": 5058
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.206587898317222e-05,
+      "loss": 0.2374,
+      "step": 5059
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.206229860365199e-05,
+      "loss": 0.382,
+      "step": 5060
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.205871822413176e-05,
+      "loss": 0.344,
+      "step": 5061
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.205513784461153e-05,
+      "loss": 0.4714,
+      "step": 5062
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.20515574650913e-05,
+      "loss": 0.3934,
+      "step": 5063
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2047977085571076e-05,
+      "loss": 0.2991,
+      "step": 5064
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.204439670605084e-05,
+      "loss": 0.3973,
+      "step": 5065
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2040816326530615e-05,
+      "loss": 0.5251,
+      "step": 5066
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.203723594701039e-05,
+      "loss": 0.3838,
+      "step": 5067
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2033655567490155e-05,
+      "loss": 0.2584,
+      "step": 5068
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.203007518796993e-05,
+      "loss": 0.2654,
+      "step": 5069
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2026494808449695e-05,
+      "loss": 0.4568,
+      "step": 5070
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.202291442892947e-05,
+      "loss": 0.4177,
+      "step": 5071
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.201933404940924e-05,
+      "loss": 0.2901,
+      "step": 5072
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.201575366988901e-05,
+      "loss": 0.3639,
+      "step": 5073
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.201217329036878e-05,
+      "loss": 0.3279,
+      "step": 5074
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.2008592910848554e-05,
+      "loss": 0.4373,
+      "step": 5075
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.200501253132832e-05,
+      "loss": 0.2536,
+      "step": 5076
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.200143215180809e-05,
+      "loss": 0.2994,
+      "step": 5077
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199785177228786e-05,
+      "loss": 0.2948,
+      "step": 5078
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.199427139276763e-05,
+      "loss": 0.4091,
+      "step": 5079
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1990691013247406e-05,
+      "loss": 0.1917,
+      "step": 5080
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.198711063372717e-05,
+      "loss": 0.3157,
+      "step": 5081
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1983530254206946e-05,
+      "loss": 0.305,
+      "step": 5082
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197994987468672e-05,
+      "loss": 0.4053,
+      "step": 5083
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.197636949516649e-05,
+      "loss": 0.3565,
+      "step": 5084
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1972789115646265e-05,
+      "loss": 0.4953,
+      "step": 5085
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.196920873612603e-05,
+      "loss": 0.3258,
+      "step": 5086
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.1965628356605805e-05,
+      "loss": 0.4044,
+      "step": 5087
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.196204797708557e-05,
+      "loss": 0.3116,
+      "step": 5088
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1958467597565344e-05,
+      "loss": 0.3428,
+      "step": 5089
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.195488721804512e-05,
+      "loss": 0.2301,
+      "step": 5090
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1951306838524884e-05,
+      "loss": 0.3953,
+      "step": 5091
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.194772645900466e-05,
+      "loss": 0.235,
+      "step": 5092
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.194414607948443e-05,
+      "loss": 0.3314,
+      "step": 5093
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.19405656999642e-05,
+      "loss": 0.3309,
+      "step": 5094
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.193698532044397e-05,
+      "loss": 0.2957,
+      "step": 5095
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1933404940923736e-05,
+      "loss": 0.2912,
+      "step": 5096
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.192982456140351e-05,
+      "loss": 0.3321,
+      "step": 5097
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.192624418188328e-05,
+      "loss": 0.2598,
+      "step": 5098
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.192266380236305e-05,
+      "loss": 0.4019,
+      "step": 5099
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.191908342284282e-05,
+      "loss": 0.4322,
+      "step": 5100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.191550304332259e-05,
+      "loss": 0.1993,
+      "step": 5101
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.191192266380236e-05,
+      "loss": 0.4819,
+      "step": 5102
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1908342284282135e-05,
+      "loss": 0.5189,
+      "step": 5103
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.19047619047619e-05,
+      "loss": 0.5695,
+      "step": 5104
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1901181525241675e-05,
+      "loss": 0.2909,
+      "step": 5105
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189760114572145e-05,
+      "loss": 0.4668,
+      "step": 5106
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.189402076620122e-05,
+      "loss": 0.3724,
+      "step": 5107
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1890440386680994e-05,
+      "loss": 0.3007,
+      "step": 5108
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.188686000716076e-05,
+      "loss": 0.3128,
+      "step": 5109
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1883279627640534e-05,
+      "loss": 0.2751,
+      "step": 5110
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.187969924812031e-05,
+      "loss": 0.3916,
+      "step": 5111
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1876118868600074e-05,
+      "loss": 0.3937,
+      "step": 5112
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.187253848907985e-05,
+      "loss": 0.262,
+      "step": 5113
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.186895810955961e-05,
+      "loss": 0.3086,
+      "step": 5114
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.1865377730039386e-05,
+      "loss": 0.3238,
+      "step": 5115
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.186179735051916e-05,
+      "loss": 0.2702,
+      "step": 5116
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1858216970998926e-05,
+      "loss": 0.2997,
+      "step": 5117
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.18546365914787e-05,
+      "loss": 0.4226,
+      "step": 5118
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1851056211958466e-05,
+      "loss": 0.2848,
+      "step": 5119
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.184747583243824e-05,
+      "loss": 0.2787,
+      "step": 5120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.184389545291801e-05,
+      "loss": 0.3532,
+      "step": 5121
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.184031507339778e-05,
+      "loss": 0.3313,
+      "step": 5122
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.183673469387755e-05,
+      "loss": 0.2961,
+      "step": 5123
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1833154314357325e-05,
+      "loss": 0.4872,
+      "step": 5124
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.182957393483709e-05,
+      "loss": 0.4253,
+      "step": 5125
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1825993555316864e-05,
+      "loss": 0.319,
+      "step": 5126
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.182241317579663e-05,
+      "loss": 0.3165,
+      "step": 5127
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1818832796276404e-05,
+      "loss": 0.3581,
+      "step": 5128
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.181525241675618e-05,
+      "loss": 0.4157,
+      "step": 5129
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.181167203723595e-05,
+      "loss": 0.3031,
+      "step": 5130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1808091657715723e-05,
+      "loss": 0.384,
+      "step": 5131
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.180451127819549e-05,
+      "loss": 0.3108,
+      "step": 5132
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.180093089867526e-05,
+      "loss": 0.2922,
+      "step": 5133
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1797350519155036e-05,
+      "loss": 0.348,
+      "step": 5134
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.17937701396348e-05,
+      "loss": 0.3337,
+      "step": 5135
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1790189760114576e-05,
+      "loss": 0.38,
+      "step": 5136
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.178660938059434e-05,
+      "loss": 0.2835,
+      "step": 5137
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1783029001074115e-05,
+      "loss": 0.4625,
+      "step": 5138
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177944862155389e-05,
+      "loss": 0.2701,
+      "step": 5139
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.1775868242033655e-05,
+      "loss": 0.3588,
+      "step": 5140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.177228786251343e-05,
+      "loss": 0.3027,
+      "step": 5141
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.17687074829932e-05,
+      "loss": 0.3227,
+      "step": 5142
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.176512710347297e-05,
+      "loss": 0.4676,
+      "step": 5143
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.176154672395274e-05,
+      "loss": 0.4003,
+      "step": 5144
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.175796634443251e-05,
+      "loss": 0.3944,
+      "step": 5145
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.175438596491228e-05,
+      "loss": 0.2802,
+      "step": 5146
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1750805585392054e-05,
+      "loss": 0.2902,
+      "step": 5147
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.174722520587182e-05,
+      "loss": 0.5097,
+      "step": 5148
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1743644826351593e-05,
+      "loss": 0.2958,
+      "step": 5149
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.174006444683137e-05,
+      "loss": 0.3557,
+      "step": 5150
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.173648406731113e-05,
+      "loss": 0.3064,
+      "step": 5151
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1732903687790906e-05,
+      "loss": 0.2102,
+      "step": 5152
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.172932330827068e-05,
+      "loss": 0.2492,
+      "step": 5153
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.172574292875045e-05,
+      "loss": 0.3135,
+      "step": 5154
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.172216254923022e-05,
+      "loss": 0.3588,
+      "step": 5155
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.171858216970999e-05,
+      "loss": 0.3705,
+      "step": 5156
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1715001790189765e-05,
+      "loss": 0.4674,
+      "step": 5157
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.171142141066953e-05,
+      "loss": 0.3445,
+      "step": 5158
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1707841031149305e-05,
+      "loss": 0.3018,
+      "step": 5159
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.170426065162908e-05,
+      "loss": 0.3003,
+      "step": 5160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1700680272108845e-05,
+      "loss": 0.3267,
+      "step": 5161
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.169709989258862e-05,
+      "loss": 0.4664,
+      "step": 5162
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.1693519513068384e-05,
+      "loss": 0.2245,
+      "step": 5163
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.168993913354816e-05,
+      "loss": 0.3111,
+      "step": 5164
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.168635875402793e-05,
+      "loss": 0.4881,
+      "step": 5165
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.16827783745077e-05,
+      "loss": 0.4196,
+      "step": 5166
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167919799498747e-05,
+      "loss": 0.2656,
+      "step": 5167
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167561761546724e-05,
+      "loss": 0.3833,
+      "step": 5168
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.167203723594701e-05,
+      "loss": 0.2922,
+      "step": 5169
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166845685642678e-05,
+      "loss": 0.3195,
+      "step": 5170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.166487647690655e-05,
+      "loss": 0.3117,
+      "step": 5171
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.166129609738632e-05,
+      "loss": 0.3882,
+      "step": 5172
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1657715717866096e-05,
+      "loss": 0.1729,
+      "step": 5173
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.165413533834586e-05,
+      "loss": 0.3255,
+      "step": 5174
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1650554958825635e-05,
+      "loss": 0.4424,
+      "step": 5175
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.164697457930541e-05,
+      "loss": 0.4011,
+      "step": 5176
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.164339419978518e-05,
+      "loss": 0.2831,
+      "step": 5177
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1639813820264955e-05,
+      "loss": 0.3577,
+      "step": 5178
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.163623344074472e-05,
+      "loss": 0.5603,
+      "step": 5179
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1632653061224494e-05,
+      "loss": 0.4032,
+      "step": 5180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.162907268170426e-05,
+      "loss": 0.3704,
+      "step": 5181
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1625492302184034e-05,
+      "loss": 0.4802,
+      "step": 5182
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.162191192266381e-05,
+      "loss": 0.3918,
+      "step": 5183
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1618331543143574e-05,
+      "loss": 0.3464,
+      "step": 5184
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.161475116362335e-05,
+      "loss": 0.3039,
+      "step": 5185
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.161117078410312e-05,
+      "loss": 0.3172,
+      "step": 5186
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1607590404582886e-05,
+      "loss": 0.2593,
+      "step": 5187
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.160401002506266e-05,
+      "loss": 0.4179,
+      "step": 5188
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1600429645542426e-05,
+      "loss": 0.3177,
+      "step": 5189
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.15968492660222e-05,
+      "loss": 0.2338,
+      "step": 5190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.159326888650197e-05,
+      "loss": 0.3782,
+      "step": 5191
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158968850698174e-05,
+      "loss": 0.3504,
+      "step": 5192
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158610812746151e-05,
+      "loss": 0.2717,
+      "step": 5193
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.158252774794128e-05,
+      "loss": 0.2558,
+      "step": 5194
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.2361,
+      "step": 5195
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1575366988900825e-05,
+      "loss": 0.3005,
+      "step": 5196
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.157178660938059e-05,
+      "loss": 0.3833,
+      "step": 5197
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1568206229860364e-05,
+      "loss": 0.3908,
+      "step": 5198
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.156462585034014e-05,
+      "loss": 0.3085,
+      "step": 5199
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.156104547081991e-05,
+      "loss": 0.415,
+      "step": 5200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1557465091299684e-05,
+      "loss": 0.2882,
+      "step": 5201
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.155388471177945e-05,
+      "loss": 0.3494,
+      "step": 5202
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1550304332259224e-05,
+      "loss": 0.2961,
+      "step": 5203
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1546723952739e-05,
+      "loss": 0.2127,
+      "step": 5204
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.154314357321876e-05,
+      "loss": 0.3024,
+      "step": 5205
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1539563193698536e-05,
+      "loss": 0.3361,
+      "step": 5206
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.15359828141783e-05,
+      "loss": 0.2698,
+      "step": 5207
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1532402434658076e-05,
+      "loss": 0.272,
+      "step": 5208
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.152882205513785e-05,
+      "loss": 0.3267,
+      "step": 5209
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1525241675617616e-05,
+      "loss": 0.3482,
+      "step": 5210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.152166129609739e-05,
+      "loss": 0.1568,
+      "step": 5211
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1518080916577155e-05,
+      "loss": 0.3614,
+      "step": 5212
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.151450053705693e-05,
+      "loss": 0.341,
+      "step": 5213
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.15109201575367e-05,
+      "loss": 0.2676,
+      "step": 5214
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.150733977801647e-05,
+      "loss": 0.4167,
+      "step": 5215
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.150375939849624e-05,
+      "loss": 0.3282,
+      "step": 5216
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1500179018976014e-05,
+      "loss": 0.6178,
+      "step": 5217
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.149659863945578e-05,
+      "loss": 0.3247,
+      "step": 5218
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1493018259935554e-05,
+      "loss": 0.4302,
+      "step": 5219
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148943788041532e-05,
+      "loss": 0.1899,
+      "step": 5220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1485857500895094e-05,
+      "loss": 0.3203,
+      "step": 5221
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.148227712137487e-05,
+      "loss": 0.4368,
+      "step": 5222
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147869674185464e-05,
+      "loss": 0.4573,
+      "step": 5223
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147511636233441e-05,
+      "loss": 0.3964,
+      "step": 5224
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.147153598281418e-05,
+      "loss": 0.3175,
+      "step": 5225
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.146795560329395e-05,
+      "loss": 0.3395,
+      "step": 5226
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.1464375223773726e-05,
+      "loss": 0.3216,
+      "step": 5227
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.146079484425349e-05,
+      "loss": 0.338,
+      "step": 5228
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1457214464733265e-05,
+      "loss": 0.4173,
+      "step": 5229
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.145363408521303e-05,
+      "loss": 0.3751,
+      "step": 5230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1450053705692805e-05,
+      "loss": 0.5582,
+      "step": 5231
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.144647332617258e-05,
+      "loss": 0.2989,
+      "step": 5232
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1442892946652345e-05,
+      "loss": 0.4788,
+      "step": 5233
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.143931256713212e-05,
+      "loss": 0.3207,
+      "step": 5234
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.143573218761189e-05,
+      "loss": 0.2738,
+      "step": 5235
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.143215180809166e-05,
+      "loss": 0.3643,
+      "step": 5236
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.142857142857143e-05,
+      "loss": 0.3837,
+      "step": 5237
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.14249910490512e-05,
+      "loss": 0.5002,
+      "step": 5238
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.142141066953097e-05,
+      "loss": 0.3346,
+      "step": 5239
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1417830290010743e-05,
+      "loss": 0.2501,
+      "step": 5240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.141424991049051e-05,
+      "loss": 0.3608,
+      "step": 5241
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.141066953097028e-05,
+      "loss": 0.3848,
+      "step": 5242
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1407089151450056e-05,
+      "loss": 0.3331,
+      "step": 5243
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.140350877192982e-05,
+      "loss": 0.457,
+      "step": 5244
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1399928392409596e-05,
+      "loss": 0.29,
+      "step": 5245
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139634801288936e-05,
+      "loss": 0.3325,
+      "step": 5246
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.139276763336914e-05,
+      "loss": 0.3204,
+      "step": 5247
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.138918725384891e-05,
+      "loss": 0.4542,
+      "step": 5248
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.138560687432868e-05,
+      "loss": 0.3757,
+      "step": 5249
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1382026494808455e-05,
+      "loss": 0.3774,
+      "step": 5250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137844611528822e-05,
+      "loss": 0.3825,
+      "step": 5251
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1374865735767995e-05,
+      "loss": 0.413,
+      "step": 5252
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.137128535624777e-05,
+      "loss": 0.3692,
+      "step": 5253
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.1367704976727534e-05,
+      "loss": 0.247,
+      "step": 5254
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.136412459720731e-05,
+      "loss": 0.3702,
+      "step": 5255
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1360544217687074e-05,
+      "loss": 0.351,
+      "step": 5256
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.135696383816685e-05,
+      "loss": 0.3875,
+      "step": 5257
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.135338345864662e-05,
+      "loss": 0.3465,
+      "step": 5258
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1349803079126387e-05,
+      "loss": 0.3573,
+      "step": 5259
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.134622269960616e-05,
+      "loss": 0.3691,
+      "step": 5260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.134264232008593e-05,
+      "loss": 0.3432,
+      "step": 5261
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.13390619405657e-05,
+      "loss": 0.4341,
+      "step": 5262
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.133548156104547e-05,
+      "loss": 0.3588,
+      "step": 5263
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.133190118152524e-05,
+      "loss": 0.3222,
+      "step": 5264
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.132832080200501e-05,
+      "loss": 0.2935,
+      "step": 5265
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1324740422484785e-05,
+      "loss": 0.2885,
+      "step": 5266
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.132116004296455e-05,
+      "loss": 0.4519,
+      "step": 5267
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1317579663444325e-05,
+      "loss": 0.4215,
+      "step": 5268
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.13139992839241e-05,
+      "loss": 0.391,
+      "step": 5269
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.131041890440387e-05,
+      "loss": 0.3219,
+      "step": 5270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1306838524883644e-05,
+      "loss": 0.3229,
+      "step": 5271
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.130325814536341e-05,
+      "loss": 0.3695,
+      "step": 5272
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1299677765843184e-05,
+      "loss": 0.2729,
+      "step": 5273
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.129609738632295e-05,
+      "loss": 0.3528,
+      "step": 5274
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1292517006802724e-05,
+      "loss": 0.4285,
+      "step": 5275
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.12889366272825e-05,
+      "loss": 0.3795,
+      "step": 5276
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.128535624776226e-05,
+      "loss": 0.3393,
+      "step": 5277
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1281775868242036e-05,
+      "loss": 0.3695,
+      "step": 5278
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.127819548872181e-05,
+      "loss": 0.3322,
+      "step": 5279
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1274615109201576e-05,
+      "loss": 0.4181,
+      "step": 5280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.127103472968135e-05,
+      "loss": 0.4613,
+      "step": 5281
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.1267454350161116e-05,
+      "loss": 0.3909,
+      "step": 5282
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.126387397064089e-05,
+      "loss": 0.37,
+      "step": 5283
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.126029359112066e-05,
+      "loss": 0.4186,
+      "step": 5284
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.125671321160043e-05,
+      "loss": 0.3066,
+      "step": 5285
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.12531328320802e-05,
+      "loss": 0.3514,
+      "step": 5286
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.124955245255997e-05,
+      "loss": 0.3169,
+      "step": 5287
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.124597207303974e-05,
+      "loss": 0.3125,
+      "step": 5288
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1242391693519514e-05,
+      "loss": 0.3486,
+      "step": 5289
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.123881131399928e-05,
+      "loss": 0.3332,
+      "step": 5290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1235230934479054e-05,
+      "loss": 0.2644,
+      "step": 5291
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.123165055495883e-05,
+      "loss": 0.4352,
+      "step": 5292
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.12280701754386e-05,
+      "loss": 0.5168,
+      "step": 5293
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1224489795918374e-05,
+      "loss": 0.3114,
+      "step": 5294
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.122090941639814e-05,
+      "loss": 0.4789,
+      "step": 5295
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.121732903687791e-05,
+      "loss": 0.4035,
+      "step": 5296
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1213748657357686e-05,
+      "loss": 0.3974,
+      "step": 5297
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.121016827783745e-05,
+      "loss": 0.3727,
+      "step": 5298
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1206587898317226e-05,
+      "loss": 0.3971,
+      "step": 5299
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.120300751879699e-05,
+      "loss": 0.28,
+      "step": 5300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1199427139276766e-05,
+      "loss": 0.2963,
+      "step": 5301
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.119584675975654e-05,
+      "loss": 0.2275,
+      "step": 5302
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1192266380236305e-05,
+      "loss": 0.3166,
+      "step": 5303
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118868600071608e-05,
+      "loss": 0.3136,
+      "step": 5304
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1185105621195845e-05,
+      "loss": 0.3422,
+      "step": 5305
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.118152524167562e-05,
+      "loss": 0.3773,
+      "step": 5306
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117794486215539e-05,
+      "loss": 0.2812,
+      "step": 5307
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117436448263516e-05,
+      "loss": 0.3134,
+      "step": 5308
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.117078410311493e-05,
+      "loss": 0.2757,
+      "step": 5309
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.1167203723594704e-05,
+      "loss": 0.2524,
+      "step": 5310
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.116362334407447e-05,
+      "loss": 0.2726,
+      "step": 5311
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1160042964554244e-05,
+      "loss": 0.3,
+      "step": 5312
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.115646258503401e-05,
+      "loss": 0.2712,
+      "step": 5313
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.115288220551378e-05,
+      "loss": 0.3386,
+      "step": 5314
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1149301825993556e-05,
+      "loss": 0.3478,
+      "step": 5315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.114572144647332e-05,
+      "loss": 0.365,
+      "step": 5316
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.11421410669531e-05,
+      "loss": 0.2672,
+      "step": 5317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.113856068743287e-05,
+      "loss": 0.214,
+      "step": 5318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.113498030791264e-05,
+      "loss": 0.3631,
+      "step": 5319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1131399928392415e-05,
+      "loss": 0.4289,
+      "step": 5320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.112781954887218e-05,
+      "loss": 0.35,
+      "step": 5321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1124239169351955e-05,
+      "loss": 0.4012,
+      "step": 5322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.112065878983172e-05,
+      "loss": 0.2941,
+      "step": 5323
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1117078410311495e-05,
+      "loss": 0.3466,
+      "step": 5324
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.111349803079127e-05,
+      "loss": 0.4955,
+      "step": 5325
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1109917651271034e-05,
+      "loss": 0.4024,
+      "step": 5326
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.110633727175081e-05,
+      "loss": 0.3656,
+      "step": 5327
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.110275689223058e-05,
+      "loss": 0.3803,
+      "step": 5328
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109917651271035e-05,
+      "loss": 0.2341,
+      "step": 5329
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109559613319012e-05,
+      "loss": 0.3952,
+      "step": 5330
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.109201575366989e-05,
+      "loss": 0.3206,
+      "step": 5331
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108843537414966e-05,
+      "loss": 0.4878,
+      "step": 5332
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.108485499462943e-05,
+      "loss": 0.2784,
+      "step": 5333
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.10812746151092e-05,
+      "loss": 0.2786,
+      "step": 5334
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.107769423558897e-05,
+      "loss": 0.3874,
+      "step": 5335
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1074113856068746e-05,
+      "loss": 0.4291,
+      "step": 5336
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.107053347654851e-05,
+      "loss": 0.3031,
+      "step": 5337
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.1066953097028285e-05,
+      "loss": 0.449,
+      "step": 5338
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.106337271750805e-05,
+      "loss": 0.306,
+      "step": 5339
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.105979233798783e-05,
+      "loss": 0.3592,
+      "step": 5340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.10562119584676e-05,
+      "loss": 0.2573,
+      "step": 5341
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.105263157894737e-05,
+      "loss": 0.2886,
+      "step": 5342
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1049051199427145e-05,
+      "loss": 0.4877,
+      "step": 5343
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.104547081990691e-05,
+      "loss": 0.4096,
+      "step": 5344
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1041890440386684e-05,
+      "loss": 0.3343,
+      "step": 5345
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.103831006086646e-05,
+      "loss": 0.3281,
+      "step": 5346
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1034729681346224e-05,
+      "loss": 0.4621,
+      "step": 5347
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1031149301826e-05,
+      "loss": 0.4478,
+      "step": 5348
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1027568922305763e-05,
+      "loss": 0.4475,
+      "step": 5349
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1023988542785537e-05,
+      "loss": 0.3176,
+      "step": 5350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.102040816326531e-05,
+      "loss": 0.2977,
+      "step": 5351
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.1016827783745076e-05,
+      "loss": 0.3558,
+      "step": 5352
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.101324740422485e-05,
+      "loss": 0.2495,
+      "step": 5353
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100966702470462e-05,
+      "loss": 0.3852,
+      "step": 5354
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100608664518439e-05,
+      "loss": 0.3762,
+      "step": 5355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.100250626566416e-05,
+      "loss": 0.3308,
+      "step": 5356
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.099892588614393e-05,
+      "loss": 0.4422,
+      "step": 5357
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09953455066237e-05,
+      "loss": 0.2858,
+      "step": 5358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0991765127103475e-05,
+      "loss": 0.4504,
+      "step": 5359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098818474758324e-05,
+      "loss": 0.5507,
+      "step": 5360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0984604368063015e-05,
+      "loss": 0.3198,
+      "step": 5361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.098102398854279e-05,
+      "loss": 0.2922,
+      "step": 5362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.097744360902256e-05,
+      "loss": 0.3886,
+      "step": 5363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0973863229502334e-05,
+      "loss": 0.3672,
+      "step": 5364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.09702828499821e-05,
+      "loss": 0.3141,
+      "step": 5365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.0966702470461874e-05,
+      "loss": 0.3374,
+      "step": 5366
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.096312209094164e-05,
+      "loss": 0.3711,
+      "step": 5367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.095954171142141e-05,
+      "loss": 0.4767,
+      "step": 5368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0955961331901186e-05,
+      "loss": 0.3603,
+      "step": 5369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.3717,
+      "step": 5370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0948800572860726e-05,
+      "loss": 0.2854,
+      "step": 5371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.09452201933405e-05,
+      "loss": 0.3255,
+      "step": 5372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0941639813820266e-05,
+      "loss": 0.342,
+      "step": 5373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.093805943430004e-05,
+      "loss": 0.4159,
+      "step": 5374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0934479054779805e-05,
+      "loss": 0.3806,
+      "step": 5375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.093089867525958e-05,
+      "loss": 0.3542,
+      "step": 5376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.092731829573935e-05,
+      "loss": 0.2795,
+      "step": 5377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.092373791621912e-05,
+      "loss": 0.2636,
+      "step": 5378
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.092015753669889e-05,
+      "loss": 0.2301,
+      "step": 5379
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0916577157178664e-05,
+      "loss": 0.4435,
+      "step": 5380
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.091299677765843e-05,
+      "loss": 0.3242,
+      "step": 5381
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0909416398138204e-05,
+      "loss": 0.3956,
+      "step": 5382
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.090583601861797e-05,
+      "loss": 0.2876,
+      "step": 5383
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0902255639097744e-05,
+      "loss": 0.3224,
+      "step": 5384
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.089867525957752e-05,
+      "loss": 0.4336,
+      "step": 5385
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.089509488005728e-05,
+      "loss": 0.3491,
+      "step": 5386
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.089151450053706e-05,
+      "loss": 0.3777,
+      "step": 5387
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.088793412101683e-05,
+      "loss": 0.2596,
+      "step": 5388
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.08843537414966e-05,
+      "loss": 0.3809,
+      "step": 5389
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0880773361976376e-05,
+      "loss": 0.3878,
+      "step": 5390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.087719298245614e-05,
+      "loss": 0.238,
+      "step": 5391
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0873612602935916e-05,
+      "loss": 0.3394,
+      "step": 5392
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.087003222341568e-05,
+      "loss": 0.306,
+      "step": 5393
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.0866451843895455e-05,
+      "loss": 0.3197,
+      "step": 5394
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.086287146437523e-05,
+      "loss": 0.46,
+      "step": 5395
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0859291084854995e-05,
+      "loss": 0.5159,
+      "step": 5396
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.085571070533477e-05,
+      "loss": 0.4036,
+      "step": 5397
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0852130325814534e-05,
+      "loss": 0.3342,
+      "step": 5398
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.084854994629431e-05,
+      "loss": 0.4672,
+      "step": 5399
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.084496956677408e-05,
+      "loss": 0.4613,
+      "step": 5400
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.084138918725385e-05,
+      "loss": 0.2556,
+      "step": 5401
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.083780880773362e-05,
+      "loss": 0.3211,
+      "step": 5402
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0834228428213394e-05,
+      "loss": 0.3245,
+      "step": 5403
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.083064804869316e-05,
+      "loss": 0.3431,
+      "step": 5404
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.082706766917293e-05,
+      "loss": 0.3223,
+      "step": 5405
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.08234872896527e-05,
+      "loss": 0.3494,
+      "step": 5406
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.081990691013247e-05,
+      "loss": 0.3008,
+      "step": 5407
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0816326530612246e-05,
+      "loss": 0.3466,
+      "step": 5408
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.081274615109201e-05,
+      "loss": 0.3586,
+      "step": 5409
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.080916577157179e-05,
+      "loss": 0.3387,
+      "step": 5410
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.080558539205156e-05,
+      "loss": 0.3725,
+      "step": 5411
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.080200501253133e-05,
+      "loss": 0.3515,
+      "step": 5412
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0798424633011105e-05,
+      "loss": 0.2968,
+      "step": 5413
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.079484425349087e-05,
+      "loss": 0.4339,
+      "step": 5414
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0791263873970645e-05,
+      "loss": 0.2382,
+      "step": 5415
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078768349445041e-05,
+      "loss": 0.3015,
+      "step": 5416
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0784103114930184e-05,
+      "loss": 0.3845,
+      "step": 5417
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.078052273540996e-05,
+      "loss": 0.2558,
+      "step": 5418
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.0776942355889724e-05,
+      "loss": 0.3262,
+      "step": 5419
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.07733619763695e-05,
+      "loss": 0.31,
+      "step": 5420
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076978159684927e-05,
+      "loss": 0.3405,
+      "step": 5421
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076620121732904e-05,
+      "loss": 0.5675,
+      "step": 5422
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.076262083780881e-05,
+      "loss": 0.4074,
+      "step": 5423
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0759040458288576e-05,
+      "loss": 0.316,
+      "step": 5424
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.075546007876835e-05,
+      "loss": 0.359,
+      "step": 5425
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.075187969924812e-05,
+      "loss": 0.4492,
+      "step": 5426
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.074829931972789e-05,
+      "loss": 0.2685,
+      "step": 5427
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.074471894020766e-05,
+      "loss": 0.3422,
+      "step": 5428
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0741138560687435e-05,
+      "loss": 0.4364,
+      "step": 5429
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.07375581811672e-05,
+      "loss": 0.2661,
+      "step": 5430
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0733977801646975e-05,
+      "loss": 0.4386,
+      "step": 5431
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.073039742212674e-05,
+      "loss": 0.278,
+      "step": 5432
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.072681704260652e-05,
+      "loss": 0.2647,
+      "step": 5433
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.072323666308629e-05,
+      "loss": 0.2929,
+      "step": 5434
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.071965628356606e-05,
+      "loss": 0.2571,
+      "step": 5435
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0716075904045834e-05,
+      "loss": 0.4049,
+      "step": 5436
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.07124955245256e-05,
+      "loss": 0.2803,
+      "step": 5437
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0708915145005374e-05,
+      "loss": 0.3689,
+      "step": 5438
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.070533476548515e-05,
+      "loss": 0.3153,
+      "step": 5439
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0701754385964913e-05,
+      "loss": 0.2978,
+      "step": 5440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0698174006444687e-05,
+      "loss": 0.3838,
+      "step": 5441
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.069459362692445e-05,
+      "loss": 0.2954,
+      "step": 5442
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0691013247404226e-05,
+      "loss": 0.3929,
+      "step": 5443
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0687432867884e-05,
+      "loss": 0.4087,
+      "step": 5444
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.0683852488363766e-05,
+      "loss": 0.3366,
+      "step": 5445
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.068027210884354e-05,
+      "loss": 0.3647,
+      "step": 5446
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.067669172932331e-05,
+      "loss": 0.3549,
+      "step": 5447
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.067311134980308e-05,
+      "loss": 0.2568,
+      "step": 5448
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066953097028285e-05,
+      "loss": 0.3397,
+      "step": 5449
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066595059076262e-05,
+      "loss": 0.3185,
+      "step": 5450
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.066237021124239e-05,
+      "loss": 0.2721,
+      "step": 5451
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0658789831722165e-05,
+      "loss": 0.3685,
+      "step": 5452
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.065520945220193e-05,
+      "loss": 0.2977,
+      "step": 5453
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0651629072681704e-05,
+      "loss": 0.4788,
+      "step": 5454
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.064804869316148e-05,
+      "loss": 0.4022,
+      "step": 5455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0644468313641244e-05,
+      "loss": 0.2765,
+      "step": 5456
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0640887934121024e-05,
+      "loss": 0.3821,
+      "step": 5457
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.063730755460079e-05,
+      "loss": 0.3643,
+      "step": 5458
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.063372717508056e-05,
+      "loss": 0.2679,
+      "step": 5459
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.063014679556033e-05,
+      "loss": 0.2596,
+      "step": 5460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.06265664160401e-05,
+      "loss": 0.3481,
+      "step": 5461
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0622986036519876e-05,
+      "loss": 0.3506,
+      "step": 5462
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.061940565699964e-05,
+      "loss": 0.3053,
+      "step": 5463
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0615825277479416e-05,
+      "loss": 0.4175,
+      "step": 5464
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.061224489795919e-05,
+      "loss": 0.2769,
+      "step": 5465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0608664518438955e-05,
+      "loss": 0.3944,
+      "step": 5466
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.060508413891873e-05,
+      "loss": 0.2452,
+      "step": 5467
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0601503759398495e-05,
+      "loss": 0.2999,
+      "step": 5468
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059792337987827e-05,
+      "loss": 0.3089,
+      "step": 5469
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059434300035804e-05,
+      "loss": 0.3981,
+      "step": 5470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.059076262083781e-05,
+      "loss": 0.3345,
+      "step": 5471
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.058718224131758e-05,
+      "loss": 0.3368,
+      "step": 5472
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0583601861797354e-05,
+      "loss": 0.4547,
+      "step": 5473
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.058002148227712e-05,
+      "loss": 0.2803,
+      "step": 5474
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0576441102756894e-05,
+      "loss": 0.2763,
+      "step": 5475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.057286072323666e-05,
+      "loss": 0.3609,
+      "step": 5476
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.056928034371643e-05,
+      "loss": 0.3447,
+      "step": 5477
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.0565699964196206e-05,
+      "loss": 0.3788,
+      "step": 5478
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.056211958467597e-05,
+      "loss": 0.3027,
+      "step": 5479
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.055853920515575e-05,
+      "loss": 0.4397,
+      "step": 5480
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.055495882563552e-05,
+      "loss": 0.3705,
+      "step": 5481
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.055137844611529e-05,
+      "loss": 0.2341,
+      "step": 5482
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0547798066595066e-05,
+      "loss": 0.385,
+      "step": 5483
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.054421768707483e-05,
+      "loss": 0.3446,
+      "step": 5484
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0540637307554605e-05,
+      "loss": 0.4041,
+      "step": 5485
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.053705692803437e-05,
+      "loss": 0.4355,
+      "step": 5486
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0533476548514145e-05,
+      "loss": 0.3752,
+      "step": 5487
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.052989616899392e-05,
+      "loss": 0.2786,
+      "step": 5488
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0526315789473684e-05,
+      "loss": 0.3896,
+      "step": 5489
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.052273540995346e-05,
+      "loss": 0.2806,
+      "step": 5490
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.051915503043323e-05,
+      "loss": 0.2661,
+      "step": 5491
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0515574650912997e-05,
+      "loss": 0.2905,
+      "step": 5492
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.051199427139277e-05,
+      "loss": 0.2624,
+      "step": 5493
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.050841389187254e-05,
+      "loss": 0.3478,
+      "step": 5494
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.050483351235231e-05,
+      "loss": 0.3759,
+      "step": 5495
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.050125313283208e-05,
+      "loss": 0.3242,
+      "step": 5496
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.049767275331185e-05,
+      "loss": 0.3776,
+      "step": 5497
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0494092373791623e-05,
+      "loss": 0.3732,
+      "step": 5498
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0490511994271393e-05,
+      "loss": 0.4789,
+      "step": 5499
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0486931614751162e-05,
+      "loss": 0.3398,
+      "step": 5500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0483351235230932e-05,
+      "loss": 0.2025,
+      "step": 5501
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0479770855710705e-05,
+      "loss": 0.3734,
+      "step": 5502
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0476190476190482e-05,
+      "loss": 0.3057,
+      "step": 5503
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0472610096670252e-05,
+      "loss": 0.368,
+      "step": 5504
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.046902971715002e-05,
+      "loss": 0.3601,
+      "step": 5505
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.046544933762979e-05,
+      "loss": 0.354,
+      "step": 5506
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.0461868958109564e-05,
+      "loss": 0.3879,
+      "step": 5507
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0458288578589334e-05,
+      "loss": 0.3647,
+      "step": 5508
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0454708199069104e-05,
+      "loss": 0.3954,
+      "step": 5509
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0451127819548874e-05,
+      "loss": 0.2677,
+      "step": 5510
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0447547440028644e-05,
+      "loss": 0.4498,
+      "step": 5511
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0443967060508417e-05,
+      "loss": 0.3458,
+      "step": 5512
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0440386680988187e-05,
+      "loss": 0.3274,
+      "step": 5513
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0436806301467956e-05,
+      "loss": 0.3851,
+      "step": 5514
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0433225921947726e-05,
+      "loss": 0.3629,
+      "step": 5515
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.04296455424275e-05,
+      "loss": 0.5282,
+      "step": 5516
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.042606516290727e-05,
+      "loss": 0.2983,
+      "step": 5517
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.042248478338704e-05,
+      "loss": 0.4052,
+      "step": 5518
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.041890440386681e-05,
+      "loss": 0.2924,
+      "step": 5519
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0415324024346582e-05,
+      "loss": 0.3355,
+      "step": 5520
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0411743644826352e-05,
+      "loss": 0.4233,
+      "step": 5521
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.040816326530612e-05,
+      "loss": 0.3222,
+      "step": 5522
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.040458288578589e-05,
+      "loss": 0.3638,
+      "step": 5523
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0401002506265665e-05,
+      "loss": 0.3512,
+      "step": 5524
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0397422126745434e-05,
+      "loss": 0.365,
+      "step": 5525
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0393841747225204e-05,
+      "loss": 0.2858,
+      "step": 5526
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.039026136770498e-05,
+      "loss": 0.4213,
+      "step": 5527
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.038668098818475e-05,
+      "loss": 0.323,
+      "step": 5528
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.038310060866452e-05,
+      "loss": 0.4855,
+      "step": 5529
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0379520229144294e-05,
+      "loss": 0.2882,
+      "step": 5530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0375939849624063e-05,
+      "loss": 0.2548,
+      "step": 5531
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0372359470103833e-05,
+      "loss": 0.2426,
+      "step": 5532
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0368779090583603e-05,
+      "loss": 0.306,
+      "step": 5533
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0365198711063376e-05,
+      "loss": 0.2655,
+      "step": 5534
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.0361618331543146e-05,
+      "loss": 0.3671,
+      "step": 5535
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0358037952022916e-05,
+      "loss": 0.2554,
+      "step": 5536
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0354457572502686e-05,
+      "loss": 0.3728,
+      "step": 5537
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.035087719298246e-05,
+      "loss": 0.2995,
+      "step": 5538
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.034729681346223e-05,
+      "loss": 0.5309,
+      "step": 5539
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0343716433942e-05,
+      "loss": 0.3194,
+      "step": 5540
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0340136054421768e-05,
+      "loss": 0.3662,
+      "step": 5541
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.033655567490154e-05,
+      "loss": 0.422,
+      "step": 5542
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.033297529538131e-05,
+      "loss": 0.2552,
+      "step": 5543
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.032939491586108e-05,
+      "loss": 0.3662,
+      "step": 5544
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.032581453634085e-05,
+      "loss": 0.2091,
+      "step": 5545
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0322234156820624e-05,
+      "loss": 0.3549,
+      "step": 5546
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0318653777300394e-05,
+      "loss": 0.2498,
+      "step": 5547
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0315073397780164e-05,
+      "loss": 0.5077,
+      "step": 5548
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0311493018259933e-05,
+      "loss": 0.3394,
+      "step": 5549
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.030791263873971e-05,
+      "loss": 0.2099,
+      "step": 5550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.030433225921948e-05,
+      "loss": 0.3233,
+      "step": 5551
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0300751879699253e-05,
+      "loss": 0.2191,
+      "step": 5552
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0297171500179023e-05,
+      "loss": 0.2937,
+      "step": 5553
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0293591120658793e-05,
+      "loss": 0.344,
+      "step": 5554
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0290010741138562e-05,
+      "loss": 0.3006,
+      "step": 5555
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0286430361618335e-05,
+      "loss": 0.3091,
+      "step": 5556
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0282849982098105e-05,
+      "loss": 0.4087,
+      "step": 5557
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0279269602577875e-05,
+      "loss": 0.255,
+      "step": 5558
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0275689223057645e-05,
+      "loss": 0.2462,
+      "step": 5559
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0272108843537418e-05,
+      "loss": 0.3111,
+      "step": 5560
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0268528464017188e-05,
+      "loss": 0.5309,
+      "step": 5561
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0264948084496958e-05,
+      "loss": 0.4016,
+      "step": 5562
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.0261367704976727e-05,
+      "loss": 0.3887,
+      "step": 5563
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.02577873254565e-05,
+      "loss": 0.3188,
+      "step": 5564
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.025420694593627e-05,
+      "loss": 0.3232,
+      "step": 5565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.025062656641604e-05,
+      "loss": 0.4383,
+      "step": 5566
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.024704618689581e-05,
+      "loss": 0.3398,
+      "step": 5567
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0243465807375583e-05,
+      "loss": 0.4004,
+      "step": 5568
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0239885427855353e-05,
+      "loss": 0.2951,
+      "step": 5569
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0236305048335123e-05,
+      "loss": 0.4106,
+      "step": 5570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0232724668814893e-05,
+      "loss": 0.3809,
+      "step": 5571
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0229144289294662e-05,
+      "loss": 0.3689,
+      "step": 5572
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.022556390977444e-05,
+      "loss": 0.2757,
+      "step": 5573
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0221983530254212e-05,
+      "loss": 0.2636,
+      "step": 5574
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0218403150733982e-05,
+      "loss": 0.2722,
+      "step": 5575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0214822771213752e-05,
+      "loss": 0.368,
+      "step": 5576
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.021124239169352e-05,
+      "loss": 0.2933,
+      "step": 5577
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0207662012173295e-05,
+      "loss": 0.1971,
+      "step": 5578
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0204081632653065e-05,
+      "loss": 0.3506,
+      "step": 5579
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0200501253132834e-05,
+      "loss": 0.3106,
+      "step": 5580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0196920873612604e-05,
+      "loss": 0.3547,
+      "step": 5581
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0193340494092377e-05,
+      "loss": 0.32,
+      "step": 5582
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0189760114572147e-05,
+      "loss": 0.2853,
+      "step": 5583
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0186179735051917e-05,
+      "loss": 0.3102,
+      "step": 5584
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0182599355531687e-05,
+      "loss": 0.3093,
+      "step": 5585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.017901897601146e-05,
+      "loss": 0.1802,
+      "step": 5586
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.017543859649123e-05,
+      "loss": 0.2809,
+      "step": 5587
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0171858216971e-05,
+      "loss": 0.222,
+      "step": 5588
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.016827783745077e-05,
+      "loss": 0.3968,
+      "step": 5589
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.016469745793054e-05,
+      "loss": 0.6157,
+      "step": 5590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.0161117078410312e-05,
+      "loss": 0.3128,
+      "step": 5591
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0157536698890082e-05,
+      "loss": 0.292,
+      "step": 5592
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0153956319369852e-05,
+      "loss": 0.288,
+      "step": 5593
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0150375939849622e-05,
+      "loss": 0.3372,
+      "step": 5594
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0146795560329395e-05,
+      "loss": 0.4117,
+      "step": 5595
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0143215180809165e-05,
+      "loss": 0.3905,
+      "step": 5596
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.013963480128894e-05,
+      "loss": 0.4217,
+      "step": 5597
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.013605442176871e-05,
+      "loss": 0.3209,
+      "step": 5598
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.013247404224848e-05,
+      "loss": 0.3947,
+      "step": 5599
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0128893662728254e-05,
+      "loss": 0.3155,
+      "step": 5600
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0125313283208024e-05,
+      "loss": 0.4997,
+      "step": 5601
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0121732903687794e-05,
+      "loss": 0.2837,
+      "step": 5602
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0118152524167563e-05,
+      "loss": 0.2852,
+      "step": 5603
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0114572144647337e-05,
+      "loss": 0.2407,
+      "step": 5604
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0110991765127106e-05,
+      "loss": 0.3336,
+      "step": 5605
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0107411385606876e-05,
+      "loss": 0.3049,
+      "step": 5606
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.8378860519567399,
+      "eval_loss": 0.38664689660072327,
+      "eval_runtime": 24.8968,
+      "eval_samples_per_second": 1801.238,
+      "eval_steps_per_second": 28.156,
+      "step": 5606
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0103831006086646e-05,
+      "loss": 0.1981,
+      "step": 5607
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0100250626566416e-05,
+      "loss": 0.201,
+      "step": 5608
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.009667024704619e-05,
+      "loss": 0.1561,
+      "step": 5609
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.009308986752596e-05,
+      "loss": 0.1604,
+      "step": 5610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.008950948800573e-05,
+      "loss": 0.197,
+      "step": 5611
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.00859291084855e-05,
+      "loss": 0.2458,
+      "step": 5612
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.008234872896527e-05,
+      "loss": 0.2486,
+      "step": 5613
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.007876834944504e-05,
+      "loss": 0.2621,
+      "step": 5614
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.007518796992481e-05,
+      "loss": 0.1491,
+      "step": 5615
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.007160759040458e-05,
+      "loss": 0.2635,
+      "step": 5616
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0068027210884354e-05,
+      "loss": 0.1564,
+      "step": 5617
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0064446831364124e-05,
+      "loss": 0.2269,
+      "step": 5618
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.0060866451843894e-05,
+      "loss": 0.2347,
+      "step": 5619
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.005728607232367e-05,
+      "loss": 0.321,
+      "step": 5620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.005370569280344e-05,
+      "loss": 0.2523,
+      "step": 5621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.005012531328321e-05,
+      "loss": 0.1767,
+      "step": 5622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0046544933762983e-05,
+      "loss": 0.2459,
+      "step": 5623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0042964554242753e-05,
+      "loss": 0.1455,
+      "step": 5624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0039384174722523e-05,
+      "loss": 0.141,
+      "step": 5625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0035803795202293e-05,
+      "loss": 0.2508,
+      "step": 5626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0032223415682066e-05,
+      "loss": 0.1972,
+      "step": 5627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0028643036161836e-05,
+      "loss": 0.2709,
+      "step": 5628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0025062656641605e-05,
+      "loss": 0.1584,
+      "step": 5629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0021482277121375e-05,
+      "loss": 0.2688,
+      "step": 5630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.001790189760115e-05,
+      "loss": 0.3314,
+      "step": 5631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0014321518080918e-05,
+      "loss": 0.153,
+      "step": 5632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0010741138560688e-05,
+      "loss": 0.1789,
+      "step": 5633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0007160759040458e-05,
+      "loss": 0.3045,
+      "step": 5634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.000358037952023e-05,
+      "loss": 0.1657,
+      "step": 5635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3e-05,
+      "loss": 0.0841,
+      "step": 5636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.999641962047977e-05,
+      "loss": 0.231,
+      "step": 5637
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.999283924095954e-05,
+      "loss": 0.1652,
+      "step": 5638
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9989258861439314e-05,
+      "loss": 0.138,
+      "step": 5639
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9985678481919083e-05,
+      "loss": 0.217,
+      "step": 5640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9982098102398853e-05,
+      "loss": 0.2806,
+      "step": 5641
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9978517722878623e-05,
+      "loss": 0.2862,
+      "step": 5642
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9974937343358396e-05,
+      "loss": 0.2161,
+      "step": 5643
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.997135696383817e-05,
+      "loss": 0.1854,
+      "step": 5644
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9967776584317943e-05,
+      "loss": 0.213,
+      "step": 5645
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9964196204797712e-05,
+      "loss": 0.1511,
+      "step": 5646
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9960615825277482e-05,
+      "loss": 0.2688,
+      "step": 5647
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.9957035445757252e-05,
+      "loss": 0.1874,
+      "step": 5648
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9953455066237025e-05,
+      "loss": 0.1844,
+      "step": 5649
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9949874686716795e-05,
+      "loss": 0.1535,
+      "step": 5650
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9946294307196565e-05,
+      "loss": 0.3122,
+      "step": 5651
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9942713927676334e-05,
+      "loss": 0.3143,
+      "step": 5652
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9939133548156108e-05,
+      "loss": 0.1839,
+      "step": 5653
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9935553168635877e-05,
+      "loss": 0.1171,
+      "step": 5654
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9931972789115647e-05,
+      "loss": 0.3043,
+      "step": 5655
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9928392409595417e-05,
+      "loss": 0.2299,
+      "step": 5656
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.992481203007519e-05,
+      "loss": 0.2453,
+      "step": 5657
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.992123165055496e-05,
+      "loss": 0.143,
+      "step": 5658
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.991765127103473e-05,
+      "loss": 0.2522,
+      "step": 5659
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.99140708915145e-05,
+      "loss": 0.2193,
+      "step": 5660
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9910490511994273e-05,
+      "loss": 0.2466,
+      "step": 5661
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9906910132474043e-05,
+      "loss": 0.229,
+      "step": 5662
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9903329752953812e-05,
+      "loss": 0.2152,
+      "step": 5663
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9899749373433582e-05,
+      "loss": 0.1733,
+      "step": 5664
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9896168993913355e-05,
+      "loss": 0.2063,
+      "step": 5665
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9892588614393125e-05,
+      "loss": 0.3048,
+      "step": 5666
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9889008234872902e-05,
+      "loss": 0.2819,
+      "step": 5667
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.988542785535267e-05,
+      "loss": 0.2001,
+      "step": 5668
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.988184747583244e-05,
+      "loss": 0.1586,
+      "step": 5669
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.987826709631221e-05,
+      "loss": 0.2183,
+      "step": 5670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9874686716791984e-05,
+      "loss": 0.1584,
+      "step": 5671
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9871106337271754e-05,
+      "loss": 0.1722,
+      "step": 5672
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9867525957751524e-05,
+      "loss": 0.1866,
+      "step": 5673
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9863945578231294e-05,
+      "loss": 0.2985,
+      "step": 5674
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9860365198711067e-05,
+      "loss": 0.1871,
+      "step": 5675
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.9856784819190837e-05,
+      "loss": 0.2847,
+      "step": 5676
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9853204439670607e-05,
+      "loss": 0.1758,
+      "step": 5677
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9849624060150376e-05,
+      "loss": 0.139,
+      "step": 5678
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.984604368063015e-05,
+      "loss": 0.2551,
+      "step": 5679
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.984246330110992e-05,
+      "loss": 0.2881,
+      "step": 5680
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.983888292158969e-05,
+      "loss": 0.4382,
+      "step": 5681
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.983530254206946e-05,
+      "loss": 0.2339,
+      "step": 5682
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9831722162549232e-05,
+      "loss": 0.1527,
+      "step": 5683
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9828141783029002e-05,
+      "loss": 0.3221,
+      "step": 5684
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9824561403508772e-05,
+      "loss": 0.3258,
+      "step": 5685
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.982098102398854e-05,
+      "loss": 0.2442,
+      "step": 5686
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.981740064446831e-05,
+      "loss": 0.244,
+      "step": 5687
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9813820264948085e-05,
+      "loss": 0.3008,
+      "step": 5688
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9810239885427854e-05,
+      "loss": 0.1611,
+      "step": 5689
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.980665950590763e-05,
+      "loss": 0.1771,
+      "step": 5690
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.98030791263874e-05,
+      "loss": 0.2975,
+      "step": 5691
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.979949874686717e-05,
+      "loss": 0.2501,
+      "step": 5692
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9795918367346944e-05,
+      "loss": 0.4263,
+      "step": 5693
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9792337987826714e-05,
+      "loss": 0.2562,
+      "step": 5694
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9788757608306483e-05,
+      "loss": 0.2267,
+      "step": 5695
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9785177228786253e-05,
+      "loss": 0.2026,
+      "step": 5696
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9781596849266026e-05,
+      "loss": 0.2945,
+      "step": 5697
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9778016469745796e-05,
+      "loss": 0.2817,
+      "step": 5698
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9774436090225566e-05,
+      "loss": 0.2995,
+      "step": 5699
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9770855710705336e-05,
+      "loss": 0.2746,
+      "step": 5700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9767275331185105e-05,
+      "loss": 0.2127,
+      "step": 5701
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.976369495166488e-05,
+      "loss": 0.1685,
+      "step": 5702
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.976011457214465e-05,
+      "loss": 0.2424,
+      "step": 5703
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9756534192624418e-05,
+      "loss": 0.1619,
+      "step": 5704
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9752953813104188e-05,
+      "loss": 0.2654,
+      "step": 5705
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.974937343358396e-05,
+      "loss": 0.2281,
+      "step": 5706
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.974579305406373e-05,
+      "loss": 0.2413,
+      "step": 5707
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.97422126745435e-05,
+      "loss": 0.1986,
+      "step": 5708
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.973863229502327e-05,
+      "loss": 0.1984,
+      "step": 5709
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9735051915503044e-05,
+      "loss": 0.284,
+      "step": 5710
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9731471535982814e-05,
+      "loss": 0.2847,
+      "step": 5711
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9727891156462583e-05,
+      "loss": 0.1855,
+      "step": 5712
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9724310776942353e-05,
+      "loss": 0.184,
+      "step": 5713
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.972073039742213e-05,
+      "loss": 0.2302,
+      "step": 5714
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9717150017901903e-05,
+      "loss": 0.3156,
+      "step": 5715
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9713569638381673e-05,
+      "loss": 0.2263,
+      "step": 5716
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9709989258861443e-05,
+      "loss": 0.2971,
+      "step": 5717
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9706408879341212e-05,
+      "loss": 0.1709,
+      "step": 5718
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9702828499820982e-05,
+      "loss": 0.1697,
+      "step": 5719
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9699248120300755e-05,
+      "loss": 0.3032,
+      "step": 5720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9695667740780525e-05,
+      "loss": 0.2715,
+      "step": 5721
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9692087361260295e-05,
+      "loss": 0.2962,
+      "step": 5722
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9688506981740065e-05,
+      "loss": 0.262,
+      "step": 5723
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9684926602219838e-05,
+      "loss": 0.2298,
+      "step": 5724
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9681346222699608e-05,
+      "loss": 0.1992,
+      "step": 5725
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9677765843179378e-05,
+      "loss": 0.2604,
+      "step": 5726
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9674185463659147e-05,
+      "loss": 0.1788,
+      "step": 5727
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.967060508413892e-05,
+      "loss": 0.1544,
+      "step": 5728
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.966702470461869e-05,
+      "loss": 0.2198,
+      "step": 5729
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.966344432509846e-05,
+      "loss": 0.2699,
+      "step": 5730
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.965986394557823e-05,
+      "loss": 0.2644,
+      "step": 5731
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9656283566058003e-05,
+      "loss": 0.3264,
+      "step": 5732
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9652703186537773e-05,
+      "loss": 0.1282,
+      "step": 5733
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9649122807017543e-05,
+      "loss": 0.1945,
+      "step": 5734
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9645542427497313e-05,
+      "loss": 0.2194,
+      "step": 5735
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9641962047977086e-05,
+      "loss": 0.3418,
+      "step": 5736
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.963838166845686e-05,
+      "loss": 0.1274,
+      "step": 5737
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9634801288936632e-05,
+      "loss": 0.3074,
+      "step": 5738
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9631220909416402e-05,
+      "loss": 0.219,
+      "step": 5739
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9627640529896172e-05,
+      "loss": 0.1797,
+      "step": 5740
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.962406015037594e-05,
+      "loss": 0.2438,
+      "step": 5741
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9620479770855715e-05,
+      "loss": 0.2218,
+      "step": 5742
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9616899391335484e-05,
+      "loss": 0.206,
+      "step": 5743
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9613319011815254e-05,
+      "loss": 0.3294,
+      "step": 5744
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9609738632295024e-05,
+      "loss": 0.2076,
+      "step": 5745
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9606158252774797e-05,
+      "loss": 0.1721,
+      "step": 5746
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9602577873254567e-05,
+      "loss": 0.0985,
+      "step": 5747
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9598997493734337e-05,
+      "loss": 0.3356,
+      "step": 5748
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9595417114214107e-05,
+      "loss": 0.3041,
+      "step": 5749
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.959183673469388e-05,
+      "loss": 0.1736,
+      "step": 5750
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.958825635517365e-05,
+      "loss": 0.2553,
+      "step": 5751
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.958467597565342e-05,
+      "loss": 0.3293,
+      "step": 5752
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.958109559613319e-05,
+      "loss": 0.1727,
+      "step": 5753
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9577515216612962e-05,
+      "loss": 0.124,
+      "step": 5754
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9573934837092732e-05,
+      "loss": 0.2078,
+      "step": 5755
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9570354457572502e-05,
+      "loss": 0.3492,
+      "step": 5756
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9566774078052272e-05,
+      "loss": 0.0911,
+      "step": 5757
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9563193698532045e-05,
+      "loss": 0.2942,
+      "step": 5758
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9559613319011815e-05,
+      "loss": 0.1869,
+      "step": 5759
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.955603293949159e-05,
+      "loss": 0.2278,
+      "step": 5760
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.955245255997136e-05,
+      "loss": 0.2763,
+      "step": 5761
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.954887218045113e-05,
+      "loss": 0.388,
+      "step": 5762
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.95452918009309e-05,
+      "loss": 0.2367,
+      "step": 5763
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9541711421410674e-05,
+      "loss": 0.2333,
+      "step": 5764
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9538131041890444e-05,
+      "loss": 0.2828,
+      "step": 5765
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9534550662370214e-05,
+      "loss": 0.1945,
+      "step": 5766
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9530970282849983e-05,
+      "loss": 0.2528,
+      "step": 5767
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9527389903329757e-05,
+      "loss": 0.1509,
+      "step": 5768
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9523809523809526e-05,
+      "loss": 0.2913,
+      "step": 5769
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9520229144289296e-05,
+      "loss": 0.2368,
+      "step": 5770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9516648764769066e-05,
+      "loss": 0.2025,
+      "step": 5771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.951306838524884e-05,
+      "loss": 0.2737,
+      "step": 5772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.950948800572861e-05,
+      "loss": 0.197,
+      "step": 5773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.950590762620838e-05,
+      "loss": 0.2046,
+      "step": 5774
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.950232724668815e-05,
+      "loss": 0.269,
+      "step": 5775
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9498746867167922e-05,
+      "loss": 0.2325,
+      "step": 5776
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.949516648764769e-05,
+      "loss": 0.18,
+      "step": 5777
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.949158610812746e-05,
+      "loss": 0.2805,
+      "step": 5778
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.948800572860723e-05,
+      "loss": 0.1863,
+      "step": 5779
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9484425349087e-05,
+      "loss": 0.2502,
+      "step": 5780
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9480844969566774e-05,
+      "loss": 0.1996,
+      "step": 5781
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9477264590046544e-05,
+      "loss": 0.2522,
+      "step": 5782
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9473684210526314e-05,
+      "loss": 0.3008,
+      "step": 5783
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.947010383100609e-05,
+      "loss": 0.1302,
+      "step": 5784
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.946652345148586e-05,
+      "loss": 0.2552,
+      "step": 5785
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9462943071965633e-05,
+      "loss": 0.1707,
+      "step": 5786
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9459362692445403e-05,
+      "loss": 0.1978,
+      "step": 5787
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.9455782312925173e-05,
+      "loss": 0.413,
+      "step": 5788
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9452201933404943e-05,
+      "loss": 0.1772,
+      "step": 5789
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9448621553884716e-05,
+      "loss": 0.2314,
+      "step": 5790
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9445041174364486e-05,
+      "loss": 0.2552,
+      "step": 5791
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9441460794844255e-05,
+      "loss": 0.167,
+      "step": 5792
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9437880415324025e-05,
+      "loss": 0.2654,
+      "step": 5793
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.94343000358038e-05,
+      "loss": 0.2583,
+      "step": 5794
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9430719656283568e-05,
+      "loss": 0.0753,
+      "step": 5795
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9427139276763338e-05,
+      "loss": 0.1182,
+      "step": 5796
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9423558897243108e-05,
+      "loss": 0.36,
+      "step": 5797
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9419978517722878e-05,
+      "loss": 0.1898,
+      "step": 5798
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.941639813820265e-05,
+      "loss": 0.2932,
+      "step": 5799
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.941281775868242e-05,
+      "loss": 0.2304,
+      "step": 5800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.940923737916219e-05,
+      "loss": 0.241,
+      "step": 5801
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.940565699964196e-05,
+      "loss": 0.2773,
+      "step": 5802
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9402076620121733e-05,
+      "loss": 0.2306,
+      "step": 5803
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9398496240601503e-05,
+      "loss": 0.2151,
+      "step": 5804
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9394915861081273e-05,
+      "loss": 0.1409,
+      "step": 5805
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9391335481561043e-05,
+      "loss": 0.2381,
+      "step": 5806
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.938775510204082e-05,
+      "loss": 0.1796,
+      "step": 5807
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9384174722520593e-05,
+      "loss": 0.2558,
+      "step": 5808
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9380594343000362e-05,
+      "loss": 0.2313,
+      "step": 5809
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9377013963480132e-05,
+      "loss": 0.1651,
+      "step": 5810
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9373433583959902e-05,
+      "loss": 0.1817,
+      "step": 5811
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9369853204439672e-05,
+      "loss": 0.3767,
+      "step": 5812
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9366272824919445e-05,
+      "loss": 0.2123,
+      "step": 5813
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9362692445399215e-05,
+      "loss": 0.2058,
+      "step": 5814
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9359112065878985e-05,
+      "loss": 0.2032,
+      "step": 5815
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.9355531686358754e-05,
+      "loss": 0.1634,
+      "step": 5816
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9351951306838528e-05,
+      "loss": 0.2452,
+      "step": 5817
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9348370927318297e-05,
+      "loss": 0.1587,
+      "step": 5818
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9344790547798067e-05,
+      "loss": 0.1763,
+      "step": 5819
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9341210168277837e-05,
+      "loss": 0.3323,
+      "step": 5820
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.933762978875761e-05,
+      "loss": 0.1136,
+      "step": 5821
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.933404940923738e-05,
+      "loss": 0.2912,
+      "step": 5822
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.933046902971715e-05,
+      "loss": 0.1434,
+      "step": 5823
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.932688865019692e-05,
+      "loss": 0.1648,
+      "step": 5824
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9323308270676693e-05,
+      "loss": 0.3636,
+      "step": 5825
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9319727891156463e-05,
+      "loss": 0.1691,
+      "step": 5826
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9316147511636232e-05,
+      "loss": 0.2339,
+      "step": 5827
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9312567132116002e-05,
+      "loss": 0.3402,
+      "step": 5828
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9308986752595775e-05,
+      "loss": 0.3244,
+      "step": 5829
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.930540637307555e-05,
+      "loss": 0.1782,
+      "step": 5830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9301825993555322e-05,
+      "loss": 0.2096,
+      "step": 5831
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.929824561403509e-05,
+      "loss": 0.2213,
+      "step": 5832
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.929466523451486e-05,
+      "loss": 0.2846,
+      "step": 5833
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.929108485499463e-05,
+      "loss": 0.1281,
+      "step": 5834
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9287504475474404e-05,
+      "loss": 0.242,
+      "step": 5835
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9283924095954174e-05,
+      "loss": 0.1965,
+      "step": 5836
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9280343716433944e-05,
+      "loss": 0.4108,
+      "step": 5837
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9276763336913714e-05,
+      "loss": 0.2032,
+      "step": 5838
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9273182957393487e-05,
+      "loss": 0.1577,
+      "step": 5839
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9269602577873257e-05,
+      "loss": 0.1425,
+      "step": 5840
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9266022198353026e-05,
+      "loss": 0.1193,
+      "step": 5841
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.9262441818832796e-05,
+      "loss": 0.1851,
+      "step": 5842
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.925886143931257e-05,
+      "loss": 0.3082,
+      "step": 5843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.925528105979234e-05,
+      "loss": 0.1769,
+      "step": 5844
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.925170068027211e-05,
+      "loss": 0.1582,
+      "step": 5845
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.924812030075188e-05,
+      "loss": 0.3003,
+      "step": 5846
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9244539921231652e-05,
+      "loss": 0.2682,
+      "step": 5847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9240959541711422e-05,
+      "loss": 0.1501,
+      "step": 5848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.923737916219119e-05,
+      "loss": 0.1224,
+      "step": 5849
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.923379878267096e-05,
+      "loss": 0.2353,
+      "step": 5850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9230218403150735e-05,
+      "loss": 0.2104,
+      "step": 5851
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9226638023630504e-05,
+      "loss": 0.3256,
+      "step": 5852
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9223057644110274e-05,
+      "loss": 0.1804,
+      "step": 5853
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.921947726459005e-05,
+      "loss": 0.2129,
+      "step": 5854
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.921589688506982e-05,
+      "loss": 0.1913,
+      "step": 5855
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.921231650554959e-05,
+      "loss": 0.1961,
+      "step": 5856
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9208736126029364e-05,
+      "loss": 0.308,
+      "step": 5857
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9205155746509133e-05,
+      "loss": 0.2505,
+      "step": 5858
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9201575366988903e-05,
+      "loss": 0.3081,
+      "step": 5859
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9197994987468673e-05,
+      "loss": 0.2202,
+      "step": 5860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9194414607948446e-05,
+      "loss": 0.1746,
+      "step": 5861
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9190834228428216e-05,
+      "loss": 0.2251,
+      "step": 5862
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9187253848907986e-05,
+      "loss": 0.1128,
+      "step": 5863
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9183673469387756e-05,
+      "loss": 0.2609,
+      "step": 5864
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.918009308986753e-05,
+      "loss": 0.2105,
+      "step": 5865
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.91765127103473e-05,
+      "loss": 0.1794,
+      "step": 5866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.917293233082707e-05,
+      "loss": 0.2292,
+      "step": 5867
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.9169351951306838e-05,
+      "loss": 0.095,
+      "step": 5868
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.916577157178661e-05,
+      "loss": 0.2003,
+      "step": 5869
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.916219119226638e-05,
+      "loss": 0.1513,
+      "step": 5870
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.915861081274615e-05,
+      "loss": 0.3384,
+      "step": 5871
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.915503043322592e-05,
+      "loss": 0.2503,
+      "step": 5872
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9151450053705694e-05,
+      "loss": 0.2417,
+      "step": 5873
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9147869674185464e-05,
+      "loss": 0.1578,
+      "step": 5874
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9144289294665234e-05,
+      "loss": 0.1686,
+      "step": 5875
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9140708915145003e-05,
+      "loss": 0.1401,
+      "step": 5876
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.913712853562478e-05,
+      "loss": 0.2795,
+      "step": 5877
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.913354815610455e-05,
+      "loss": 0.2417,
+      "step": 5878
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9129967776584323e-05,
+      "loss": 0.3033,
+      "step": 5879
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9126387397064093e-05,
+      "loss": 0.2212,
+      "step": 5880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9122807017543863e-05,
+      "loss": 0.2566,
+      "step": 5881
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9119226638023632e-05,
+      "loss": 0.1833,
+      "step": 5882
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9115646258503405e-05,
+      "loss": 0.1889,
+      "step": 5883
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9112065878983175e-05,
+      "loss": 0.1097,
+      "step": 5884
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9108485499462945e-05,
+      "loss": 0.1392,
+      "step": 5885
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9104905119942715e-05,
+      "loss": 0.2385,
+      "step": 5886
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9101324740422488e-05,
+      "loss": 0.1592,
+      "step": 5887
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9097744360902258e-05,
+      "loss": 0.2066,
+      "step": 5888
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9094163981382028e-05,
+      "loss": 0.2273,
+      "step": 5889
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9090583601861797e-05,
+      "loss": 0.124,
+      "step": 5890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9087003222341567e-05,
+      "loss": 0.2377,
+      "step": 5891
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.908342284282134e-05,
+      "loss": 0.1199,
+      "step": 5892
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.907984246330111e-05,
+      "loss": 0.309,
+      "step": 5893
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.907626208378088e-05,
+      "loss": 0.4053,
+      "step": 5894
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.907268170426065e-05,
+      "loss": 0.1542,
+      "step": 5895
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9069101324740423e-05,
+      "loss": 0.1229,
+      "step": 5896
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9065520945220193e-05,
+      "loss": 0.2821,
+      "step": 5897
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9061940565699963e-05,
+      "loss": 0.3029,
+      "step": 5898
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9058360186179732e-05,
+      "loss": 0.2117,
+      "step": 5899
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.905477980665951e-05,
+      "loss": 0.246,
+      "step": 5900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9051199427139282e-05,
+      "loss": 0.1708,
+      "step": 5901
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9047619047619052e-05,
+      "loss": 0.3551,
+      "step": 5902
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9044038668098822e-05,
+      "loss": 0.1167,
+      "step": 5903
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.904045828857859e-05,
+      "loss": 0.2489,
+      "step": 5904
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9036877909058365e-05,
+      "loss": 0.1341,
+      "step": 5905
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9033297529538135e-05,
+      "loss": 0.1634,
+      "step": 5906
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9029717150017904e-05,
+      "loss": 0.1363,
+      "step": 5907
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9026136770497674e-05,
+      "loss": 0.2947,
+      "step": 5908
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9022556390977444e-05,
+      "loss": 0.1385,
+      "step": 5909
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9018976011457217e-05,
+      "loss": 0.2217,
+      "step": 5910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9015395631936987e-05,
+      "loss": 0.1698,
+      "step": 5911
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9011815252416757e-05,
+      "loss": 0.2406,
+      "step": 5912
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9008234872896527e-05,
+      "loss": 0.3133,
+      "step": 5913
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.90046544933763e-05,
+      "loss": 0.2753,
+      "step": 5914
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.900107411385607e-05,
+      "loss": 0.2544,
+      "step": 5915
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.899749373433584e-05,
+      "loss": 0.1893,
+      "step": 5916
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.899391335481561e-05,
+      "loss": 0.2192,
+      "step": 5917
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8990332975295382e-05,
+      "loss": 0.2332,
+      "step": 5918
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8986752595775152e-05,
+      "loss": 0.1783,
+      "step": 5919
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8983172216254922e-05,
+      "loss": 0.2641,
+      "step": 5920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8979591836734692e-05,
+      "loss": 0.1221,
+      "step": 5921
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8976011457214465e-05,
+      "loss": 0.2129,
+      "step": 5922
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8972431077694235e-05,
+      "loss": 0.3713,
+      "step": 5923
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.896885069817401e-05,
+      "loss": 0.3279,
+      "step": 5924
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.896527031865378e-05,
+      "loss": 0.3499,
+      "step": 5925
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.896168993913355e-05,
+      "loss": 0.2061,
+      "step": 5926
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.895810955961332e-05,
+      "loss": 0.2356,
+      "step": 5927
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.8954529180093094e-05,
+      "loss": 0.3253,
+      "step": 5928
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8950948800572864e-05,
+      "loss": 0.2628,
+      "step": 5929
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8947368421052634e-05,
+      "loss": 0.3504,
+      "step": 5930
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8943788041532403e-05,
+      "loss": 0.3829,
+      "step": 5931
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8940207662012176e-05,
+      "loss": 0.2054,
+      "step": 5932
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8936627282491946e-05,
+      "loss": 0.208,
+      "step": 5933
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8933046902971716e-05,
+      "loss": 0.4022,
+      "step": 5934
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8929466523451486e-05,
+      "loss": 0.1423,
+      "step": 5935
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.892588614393126e-05,
+      "loss": 0.1623,
+      "step": 5936
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.892230576441103e-05,
+      "loss": 0.1801,
+      "step": 5937
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.89187253848908e-05,
+      "loss": 0.2806,
+      "step": 5938
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.891514500537057e-05,
+      "loss": 0.288,
+      "step": 5939
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.891156462585034e-05,
+      "loss": 0.186,
+      "step": 5940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.890798424633011e-05,
+      "loss": 0.1312,
+      "step": 5941
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.890440386680988e-05,
+      "loss": 0.1518,
+      "step": 5942
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.890082348728965e-05,
+      "loss": 0.1837,
+      "step": 5943
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8897243107769424e-05,
+      "loss": 0.2434,
+      "step": 5944
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8893662728249194e-05,
+      "loss": 0.1829,
+      "step": 5945
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8890082348728964e-05,
+      "loss": 0.255,
+      "step": 5946
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.888650196920874e-05,
+      "loss": 0.2965,
+      "step": 5947
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.888292158968851e-05,
+      "loss": 0.1949,
+      "step": 5948
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.887934121016828e-05,
+      "loss": 0.2146,
+      "step": 5949
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8875760830648053e-05,
+      "loss": 0.1561,
+      "step": 5950
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8872180451127823e-05,
+      "loss": 0.2787,
+      "step": 5951
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8868600071607593e-05,
+      "loss": 0.2289,
+      "step": 5952
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8865019692087363e-05,
+      "loss": 0.1864,
+      "step": 5953
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8861439312567136e-05,
+      "loss": 0.1503,
+      "step": 5954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8857858933046906e-05,
+      "loss": 0.2752,
+      "step": 5955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.8854278553526675e-05,
+      "loss": 0.1559,
+      "step": 5956
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8850698174006445e-05,
+      "loss": 0.4187,
+      "step": 5957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.884711779448622e-05,
+      "loss": 0.1953,
+      "step": 5958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8843537414965988e-05,
+      "loss": 0.2426,
+      "step": 5959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8839957035445758e-05,
+      "loss": 0.1651,
+      "step": 5960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8836376655925528e-05,
+      "loss": 0.1839,
+      "step": 5961
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.88327962764053e-05,
+      "loss": 0.2168,
+      "step": 5962
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.882921589688507e-05,
+      "loss": 0.267,
+      "step": 5963
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.882563551736484e-05,
+      "loss": 0.3932,
+      "step": 5964
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.882205513784461e-05,
+      "loss": 0.316,
+      "step": 5965
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8818474758324384e-05,
+      "loss": 0.2641,
+      "step": 5966
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8814894378804153e-05,
+      "loss": 0.281,
+      "step": 5967
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8811313999283923e-05,
+      "loss": 0.1934,
+      "step": 5968
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8807733619763693e-05,
+      "loss": 0.309,
+      "step": 5969
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.880415324024347e-05,
+      "loss": 0.2629,
+      "step": 5970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.880057286072324e-05,
+      "loss": 0.2398,
+      "step": 5971
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8796992481203013e-05,
+      "loss": 0.1853,
+      "step": 5972
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8793412101682782e-05,
+      "loss": 0.1842,
+      "step": 5973
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8789831722162552e-05,
+      "loss": 0.2597,
+      "step": 5974
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8786251342642322e-05,
+      "loss": 0.2587,
+      "step": 5975
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8782670963122095e-05,
+      "loss": 0.3636,
+      "step": 5976
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8779090583601865e-05,
+      "loss": 0.1635,
+      "step": 5977
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8775510204081635e-05,
+      "loss": 0.3296,
+      "step": 5978
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8771929824561404e-05,
+      "loss": 0.2526,
+      "step": 5979
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8768349445041178e-05,
+      "loss": 0.1797,
+      "step": 5980
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8764769065520947e-05,
+      "loss": 0.0995,
+      "step": 5981
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8761188686000717e-05,
+      "loss": 0.2694,
+      "step": 5982
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8757608306480487e-05,
+      "loss": 0.4171,
+      "step": 5983
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.875402792696026e-05,
+      "loss": 0.1592,
+      "step": 5984
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.875044754744003e-05,
+      "loss": 0.3225,
+      "step": 5985
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.87468671679198e-05,
+      "loss": 0.2761,
+      "step": 5986
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.874328678839957e-05,
+      "loss": 0.3578,
+      "step": 5987
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.873970640887934e-05,
+      "loss": 0.1989,
+      "step": 5988
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8736126029359113e-05,
+      "loss": 0.1464,
+      "step": 5989
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8732545649838882e-05,
+      "loss": 0.1563,
+      "step": 5990
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8728965270318652e-05,
+      "loss": 0.1551,
+      "step": 5991
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8725384890798422e-05,
+      "loss": 0.1971,
+      "step": 5992
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8721804511278195e-05,
+      "loss": 0.2647,
+      "step": 5993
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8718224131757972e-05,
+      "loss": 0.2782,
+      "step": 5994
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.871464375223774e-05,
+      "loss": 0.2187,
+      "step": 5995
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.871106337271751e-05,
+      "loss": 0.3142,
+      "step": 5996
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.870748299319728e-05,
+      "loss": 0.2126,
+      "step": 5997
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8703902613677054e-05,
+      "loss": 0.3069,
+      "step": 5998
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8700322234156824e-05,
+      "loss": 0.2115,
+      "step": 5999
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8696741854636594e-05,
+      "loss": 0.182,
+      "step": 6000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8693161475116364e-05,
+      "loss": 0.318,
+      "step": 6001
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8689581095596134e-05,
+      "loss": 0.2514,
+      "step": 6002
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8686000716075907e-05,
+      "loss": 0.1662,
+      "step": 6003
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8682420336555677e-05,
+      "loss": 0.3251,
+      "step": 6004
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8678839957035446e-05,
+      "loss": 0.2368,
+      "step": 6005
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8675259577515216e-05,
+      "loss": 0.1986,
+      "step": 6006
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.867167919799499e-05,
+      "loss": 0.3173,
+      "step": 6007
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.866809881847476e-05,
+      "loss": 0.3817,
+      "step": 6008
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.866451843895453e-05,
+      "loss": 0.3041,
+      "step": 6009
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.86609380594343e-05,
+      "loss": 0.236,
+      "step": 6010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8657357679914072e-05,
+      "loss": 0.2712,
+      "step": 6011
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8653777300393842e-05,
+      "loss": 0.2214,
+      "step": 6012
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.865019692087361e-05,
+      "loss": 0.1711,
+      "step": 6013
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.864661654135338e-05,
+      "loss": 0.2124,
+      "step": 6014
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8643036161833155e-05,
+      "loss": 0.3178,
+      "step": 6015
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8639455782312924e-05,
+      "loss": 0.1225,
+      "step": 6016
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.86358754027927e-05,
+      "loss": 0.273,
+      "step": 6017
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.863229502327247e-05,
+      "loss": 0.1811,
+      "step": 6018
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.862871464375224e-05,
+      "loss": 0.1703,
+      "step": 6019
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.862513426423201e-05,
+      "loss": 0.2417,
+      "step": 6020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8621553884711784e-05,
+      "loss": 0.4159,
+      "step": 6021
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8617973505191553e-05,
+      "loss": 0.1941,
+      "step": 6022
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8614393125671323e-05,
+      "loss": 0.3948,
+      "step": 6023
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8610812746151093e-05,
+      "loss": 0.2026,
+      "step": 6024
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8607232366630866e-05,
+      "loss": 0.2417,
+      "step": 6025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8603651987110636e-05,
+      "loss": 0.2726,
+      "step": 6026
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8600071607590406e-05,
+      "loss": 0.1924,
+      "step": 6027
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8596491228070175e-05,
+      "loss": 0.3196,
+      "step": 6028
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.859291084854995e-05,
+      "loss": 0.3281,
+      "step": 6029
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.858933046902972e-05,
+      "loss": 0.2842,
+      "step": 6030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8585750089509488e-05,
+      "loss": 0.2108,
+      "step": 6031
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8582169709989258e-05,
+      "loss": 0.3214,
+      "step": 6032
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.857858933046903e-05,
+      "loss": 0.1516,
+      "step": 6033
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.85750089509488e-05,
+      "loss": 0.2804,
+      "step": 6034
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.2859,
+      "step": 6035
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.856784819190834e-05,
+      "loss": 0.2491,
+      "step": 6036
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8564267812388114e-05,
+      "loss": 0.2851,
+      "step": 6037
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8560687432867884e-05,
+      "loss": 0.21,
+      "step": 6038
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8557107053347653e-05,
+      "loss": 0.1867,
+      "step": 6039
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8553526673827423e-05,
+      "loss": 0.1666,
+      "step": 6040
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.85499462943072e-05,
+      "loss": 0.2587,
+      "step": 6041
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.854636591478697e-05,
+      "loss": 0.3664,
+      "step": 6042
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8542785535266743e-05,
+      "loss": 0.2862,
+      "step": 6043
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8539205155746513e-05,
+      "loss": 0.2308,
+      "step": 6044
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8535624776226282e-05,
+      "loss": 0.1123,
+      "step": 6045
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8532044396706052e-05,
+      "loss": 0.1731,
+      "step": 6046
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8528464017185825e-05,
+      "loss": 0.3318,
+      "step": 6047
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8524883637665595e-05,
+      "loss": 0.1881,
+      "step": 6048
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8521303258145365e-05,
+      "loss": 0.2787,
+      "step": 6049
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8517722878625135e-05,
+      "loss": 0.2187,
+      "step": 6050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8514142499104908e-05,
+      "loss": 0.1882,
+      "step": 6051
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8510562119584678e-05,
+      "loss": 0.2414,
+      "step": 6052
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8506981740064448e-05,
+      "loss": 0.2441,
+      "step": 6053
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8503401360544217e-05,
+      "loss": 0.2406,
+      "step": 6054
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.849982098102399e-05,
+      "loss": 0.1618,
+      "step": 6055
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.849624060150376e-05,
+      "loss": 0.1544,
+      "step": 6056
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.849266022198353e-05,
+      "loss": 0.2737,
+      "step": 6057
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.84890798424633e-05,
+      "loss": 0.261,
+      "step": 6058
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8485499462943073e-05,
+      "loss": 0.2646,
+      "step": 6059
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8481919083422843e-05,
+      "loss": 0.2417,
+      "step": 6060
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8478338703902613e-05,
+      "loss": 0.1761,
+      "step": 6061
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8474758324382383e-05,
+      "loss": 0.1673,
+      "step": 6062
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8471177944862156e-05,
+      "loss": 0.2038,
+      "step": 6063
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.846759756534193e-05,
+      "loss": 0.2248,
+      "step": 6064
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8464017185821702e-05,
+      "loss": 0.2106,
+      "step": 6065
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8460436806301472e-05,
+      "loss": 0.184,
+      "step": 6066
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8456856426781242e-05,
+      "loss": 0.2286,
+      "step": 6067
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.845327604726101e-05,
+      "loss": 0.2272,
+      "step": 6068
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8449695667740785e-05,
+      "loss": 0.3888,
+      "step": 6069
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8446115288220554e-05,
+      "loss": 0.2385,
+      "step": 6070
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8442534908700324e-05,
+      "loss": 0.209,
+      "step": 6071
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8438954529180094e-05,
+      "loss": 0.309,
+      "step": 6072
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8435374149659867e-05,
+      "loss": 0.3154,
+      "step": 6073
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8431793770139637e-05,
+      "loss": 0.2097,
+      "step": 6074
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8428213390619407e-05,
+      "loss": 0.2019,
+      "step": 6075
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8424633011099177e-05,
+      "loss": 0.3374,
+      "step": 6076
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.842105263157895e-05,
+      "loss": 0.1461,
+      "step": 6077
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.841747225205872e-05,
+      "loss": 0.3183,
+      "step": 6078
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.841389187253849e-05,
+      "loss": 0.2161,
+      "step": 6079
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.841031149301826e-05,
+      "loss": 0.112,
+      "step": 6080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.840673111349803e-05,
+      "loss": 0.1491,
+      "step": 6081
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8403150733977802e-05,
+      "loss": 0.2336,
+      "step": 6082
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8399570354457572e-05,
+      "loss": 0.3433,
+      "step": 6083
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8395989974937342e-05,
+      "loss": 0.1435,
+      "step": 6084
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.839240959541711e-05,
+      "loss": 0.1265,
+      "step": 6085
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8388829215896885e-05,
+      "loss": 0.1976,
+      "step": 6086
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.838524883637666e-05,
+      "loss": 0.2955,
+      "step": 6087
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.838166845685643e-05,
+      "loss": 0.1875,
+      "step": 6088
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.83780880773362e-05,
+      "loss": 0.1803,
+      "step": 6089
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.837450769781597e-05,
+      "loss": 0.1551,
+      "step": 6090
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8370927318295744e-05,
+      "loss": 0.3395,
+      "step": 6091
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8367346938775514e-05,
+      "loss": 0.3555,
+      "step": 6092
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8363766559255284e-05,
+      "loss": 0.2852,
+      "step": 6093
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8360186179735053e-05,
+      "loss": 0.2382,
+      "step": 6094
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8356605800214827e-05,
+      "loss": 0.2985,
+      "step": 6095
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.8353025420694596e-05,
+      "loss": 0.1322,
+      "step": 6096
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8349445041174366e-05,
+      "loss": 0.3121,
+      "step": 6097
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8345864661654136e-05,
+      "loss": 0.1931,
+      "step": 6098
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8342284282133906e-05,
+      "loss": 0.2154,
+      "step": 6099
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.833870390261368e-05,
+      "loss": 0.3014,
+      "step": 6100
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.833512352309345e-05,
+      "loss": 0.3097,
+      "step": 6101
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.833154314357322e-05,
+      "loss": 0.3906,
+      "step": 6102
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.832796276405299e-05,
+      "loss": 0.1363,
+      "step": 6103
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.832438238453276e-05,
+      "loss": 0.2972,
+      "step": 6104
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.832080200501253e-05,
+      "loss": 0.187,
+      "step": 6105
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.83172216254923e-05,
+      "loss": 0.3488,
+      "step": 6106
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.831364124597207e-05,
+      "loss": 0.1178,
+      "step": 6107
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8310060866451844e-05,
+      "loss": 0.1782,
+      "step": 6108
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8306480486931614e-05,
+      "loss": 0.2592,
+      "step": 6109
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8302900107411384e-05,
+      "loss": 0.2542,
+      "step": 6110
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.829931972789116e-05,
+      "loss": 0.3987,
+      "step": 6111
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.829573934837093e-05,
+      "loss": 0.1818,
+      "step": 6112
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.82921589688507e-05,
+      "loss": 0.168,
+      "step": 6113
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8288578589330473e-05,
+      "loss": 0.2085,
+      "step": 6114
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8284998209810243e-05,
+      "loss": 0.2755,
+      "step": 6115
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8281417830290013e-05,
+      "loss": 0.1897,
+      "step": 6116
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8277837450769783e-05,
+      "loss": 0.2307,
+      "step": 6117
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8274257071249556e-05,
+      "loss": 0.2421,
+      "step": 6118
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8270676691729325e-05,
+      "loss": 0.1191,
+      "step": 6119
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8267096312209095e-05,
+      "loss": 0.1887,
+      "step": 6120
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8263515932688865e-05,
+      "loss": 0.2063,
+      "step": 6121
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8259935553168638e-05,
+      "loss": 0.2876,
+      "step": 6122
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8256355173648408e-05,
+      "loss": 0.1759,
+      "step": 6123
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.8252774794128178e-05,
+      "loss": 0.1032,
+      "step": 6124
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8249194414607948e-05,
+      "loss": 0.315,
+      "step": 6125
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.824561403508772e-05,
+      "loss": 0.1979,
+      "step": 6126
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.824203365556749e-05,
+      "loss": 0.2188,
+      "step": 6127
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.823845327604726e-05,
+      "loss": 0.327,
+      "step": 6128
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.823487289652703e-05,
+      "loss": 0.21,
+      "step": 6129
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8231292517006803e-05,
+      "loss": 0.2813,
+      "step": 6130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8227712137486573e-05,
+      "loss": 0.1896,
+      "step": 6131
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8224131757966343e-05,
+      "loss": 0.2102,
+      "step": 6132
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8220551378446113e-05,
+      "loss": 0.1948,
+      "step": 6133
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.821697099892589e-05,
+      "loss": 0.2729,
+      "step": 6134
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.821339061940566e-05,
+      "loss": 0.2672,
+      "step": 6135
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8209810239885432e-05,
+      "loss": 0.1777,
+      "step": 6136
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8206229860365202e-05,
+      "loss": 0.3381,
+      "step": 6137
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8202649480844972e-05,
+      "loss": 0.2598,
+      "step": 6138
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8199069101324742e-05,
+      "loss": 0.1638,
+      "step": 6139
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8195488721804515e-05,
+      "loss": 0.1967,
+      "step": 6140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8191908342284285e-05,
+      "loss": 0.1877,
+      "step": 6141
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8188327962764055e-05,
+      "loss": 0.2068,
+      "step": 6142
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8184747583243824e-05,
+      "loss": 0.2637,
+      "step": 6143
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8181167203723598e-05,
+      "loss": 0.1574,
+      "step": 6144
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8177586824203367e-05,
+      "loss": 0.2577,
+      "step": 6145
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8174006444683137e-05,
+      "loss": 0.2383,
+      "step": 6146
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8170426065162907e-05,
+      "loss": 0.2303,
+      "step": 6147
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.816684568564268e-05,
+      "loss": 0.3364,
+      "step": 6148
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.816326530612245e-05,
+      "loss": 0.2072,
+      "step": 6149
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.815968492660222e-05,
+      "loss": 0.3366,
+      "step": 6150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.815610454708199e-05,
+      "loss": 0.2521,
+      "step": 6151
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.8152524167561763e-05,
+      "loss": 0.1657,
+      "step": 6152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8148943788041533e-05,
+      "loss": 0.4032,
+      "step": 6153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8145363408521302e-05,
+      "loss": 0.1728,
+      "step": 6154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8141783029001072e-05,
+      "loss": 0.3699,
+      "step": 6155
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8138202649480845e-05,
+      "loss": 0.3615,
+      "step": 6156
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.813462226996062e-05,
+      "loss": 0.1768,
+      "step": 6157
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8131041890440392e-05,
+      "loss": 0.2794,
+      "step": 6158
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.812746151092016e-05,
+      "loss": 0.2079,
+      "step": 6159
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.812388113139993e-05,
+      "loss": 0.2278,
+      "step": 6160
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.81203007518797e-05,
+      "loss": 0.2357,
+      "step": 6161
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8116720372359474e-05,
+      "loss": 0.1891,
+      "step": 6162
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8113139992839244e-05,
+      "loss": 0.2469,
+      "step": 6163
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8109559613319014e-05,
+      "loss": 0.3087,
+      "step": 6164
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8105979233798784e-05,
+      "loss": 0.1755,
+      "step": 6165
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8102398854278557e-05,
+      "loss": 0.3708,
+      "step": 6166
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8098818474758327e-05,
+      "loss": 0.2756,
+      "step": 6167
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8095238095238096e-05,
+      "loss": 0.1293,
+      "step": 6168
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8091657715717866e-05,
+      "loss": 0.2723,
+      "step": 6169
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.808807733619764e-05,
+      "loss": 0.3084,
+      "step": 6170
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.808449695667741e-05,
+      "loss": 0.2517,
+      "step": 6171
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.808091657715718e-05,
+      "loss": 0.1546,
+      "step": 6172
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.807733619763695e-05,
+      "loss": 0.1646,
+      "step": 6173
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8073755818116722e-05,
+      "loss": 0.2372,
+      "step": 6174
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.2571,
+      "step": 6175
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.806659505907626e-05,
+      "loss": 0.2371,
+      "step": 6176
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.806301467955603e-05,
+      "loss": 0.3506,
+      "step": 6177
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.80594343000358e-05,
+      "loss": 0.1359,
+      "step": 6178
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8055853920515574e-05,
+      "loss": 0.3341,
+      "step": 6179
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.8052273540995344e-05,
+      "loss": 0.2138,
+      "step": 6180
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.804869316147512e-05,
+      "loss": 0.2864,
+      "step": 6181
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.804511278195489e-05,
+      "loss": 0.3068,
+      "step": 6182
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.804153240243466e-05,
+      "loss": 0.2852,
+      "step": 6183
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8037952022914434e-05,
+      "loss": 0.2384,
+      "step": 6184
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8034371643394203e-05,
+      "loss": 0.1728,
+      "step": 6185
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8030791263873973e-05,
+      "loss": 0.5276,
+      "step": 6186
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8027210884353743e-05,
+      "loss": 0.2169,
+      "step": 6187
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8023630504833516e-05,
+      "loss": 0.1637,
+      "step": 6188
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8020050125313286e-05,
+      "loss": 0.1887,
+      "step": 6189
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8016469745793056e-05,
+      "loss": 0.1554,
+      "step": 6190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8012889366272826e-05,
+      "loss": 0.1541,
+      "step": 6191
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.8009308986752595e-05,
+      "loss": 0.0999,
+      "step": 6192
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.800572860723237e-05,
+      "loss": 0.3266,
+      "step": 6193
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.800214822771214e-05,
+      "loss": 0.1877,
+      "step": 6194
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7998567848191908e-05,
+      "loss": 0.3492,
+      "step": 6195
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7994987468671678e-05,
+      "loss": 0.2226,
+      "step": 6196
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.799140708915145e-05,
+      "loss": 0.2833,
+      "step": 6197
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.798782670963122e-05,
+      "loss": 0.2967,
+      "step": 6198
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.798424633011099e-05,
+      "loss": 0.2306,
+      "step": 6199
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.798066595059076e-05,
+      "loss": 0.3281,
+      "step": 6200
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7977085571070534e-05,
+      "loss": 0.2224,
+      "step": 6201
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7973505191550304e-05,
+      "loss": 0.1835,
+      "step": 6202
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7969924812030073e-05,
+      "loss": 0.2352,
+      "step": 6203
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.796634443250985e-05,
+      "loss": 0.2208,
+      "step": 6204
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.796276405298962e-05,
+      "loss": 0.3252,
+      "step": 6205
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7959183673469393e-05,
+      "loss": 0.2147,
+      "step": 6206
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7955603293949163e-05,
+      "loss": 0.1731,
+      "step": 6207
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.7952022914428933e-05,
+      "loss": 0.2394,
+      "step": 6208
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7948442534908702e-05,
+      "loss": 0.2571,
+      "step": 6209
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7944862155388472e-05,
+      "loss": 0.2344,
+      "step": 6210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7941281775868245e-05,
+      "loss": 0.3169,
+      "step": 6211
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7937701396348015e-05,
+      "loss": 0.1852,
+      "step": 6212
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7934121016827785e-05,
+      "loss": 0.2421,
+      "step": 6213
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7930540637307555e-05,
+      "loss": 0.2415,
+      "step": 6214
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7926960257787328e-05,
+      "loss": 0.3826,
+      "step": 6215
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7923379878267098e-05,
+      "loss": 0.1753,
+      "step": 6216
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7919799498746867e-05,
+      "loss": 0.2775,
+      "step": 6217
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7916219119226637e-05,
+      "loss": 0.1045,
+      "step": 6218
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.791263873970641e-05,
+      "loss": 0.1927,
+      "step": 6219
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.790905836018618e-05,
+      "loss": 0.1911,
+      "step": 6220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.790547798066595e-05,
+      "loss": 0.2397,
+      "step": 6221
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.790189760114572e-05,
+      "loss": 0.2179,
+      "step": 6222
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7898317221625493e-05,
+      "loss": 0.2375,
+      "step": 6223
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7894736842105263e-05,
+      "loss": 0.2154,
+      "step": 6224
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7891156462585033e-05,
+      "loss": 0.163,
+      "step": 6225
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7887576083064802e-05,
+      "loss": 0.2989,
+      "step": 6226
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.788399570354458e-05,
+      "loss": 0.3193,
+      "step": 6227
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.788041532402435e-05,
+      "loss": 0.2566,
+      "step": 6228
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7876834944504122e-05,
+      "loss": 0.158,
+      "step": 6229
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7873254564983892e-05,
+      "loss": 0.2572,
+      "step": 6230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.786967418546366e-05,
+      "loss": 0.2677,
+      "step": 6231
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.786609380594343e-05,
+      "loss": 0.2544,
+      "step": 6232
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7862513426423205e-05,
+      "loss": 0.2619,
+      "step": 6233
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7858933046902974e-05,
+      "loss": 0.3167,
+      "step": 6234
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7855352667382744e-05,
+      "loss": 0.28,
+      "step": 6235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.7851772287862514e-05,
+      "loss": 0.1955,
+      "step": 6236
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7848191908342287e-05,
+      "loss": 0.1803,
+      "step": 6237
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7844611528822057e-05,
+      "loss": 0.2214,
+      "step": 6238
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7841031149301827e-05,
+      "loss": 0.245,
+      "step": 6239
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7837450769781597e-05,
+      "loss": 0.2118,
+      "step": 6240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.783387039026137e-05,
+      "loss": 0.3379,
+      "step": 6241
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.783029001074114e-05,
+      "loss": 0.2618,
+      "step": 6242
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.782670963122091e-05,
+      "loss": 0.2905,
+      "step": 6243
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.782312925170068e-05,
+      "loss": 0.2495,
+      "step": 6244
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7819548872180452e-05,
+      "loss": 0.3338,
+      "step": 6245
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7815968492660222e-05,
+      "loss": 0.1929,
+      "step": 6246
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7812388113139992e-05,
+      "loss": 0.2725,
+      "step": 6247
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7808807733619762e-05,
+      "loss": 0.2351,
+      "step": 6248
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7805227354099535e-05,
+      "loss": 0.2923,
+      "step": 6249
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7801646974579305e-05,
+      "loss": 0.2006,
+      "step": 6250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.779806659505908e-05,
+      "loss": 0.1306,
+      "step": 6251
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.779448621553885e-05,
+      "loss": 0.1812,
+      "step": 6252
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.779090583601862e-05,
+      "loss": 0.2071,
+      "step": 6253
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.778732545649839e-05,
+      "loss": 0.2676,
+      "step": 6254
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7783745076978164e-05,
+      "loss": 0.3263,
+      "step": 6255
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7780164697457934e-05,
+      "loss": 0.1478,
+      "step": 6256
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7776584317937704e-05,
+      "loss": 0.3123,
+      "step": 6257
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7773003938417473e-05,
+      "loss": 0.1519,
+      "step": 6258
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7769423558897246e-05,
+      "loss": 0.2819,
+      "step": 6259
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7765843179377016e-05,
+      "loss": 0.3201,
+      "step": 6260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7762262799856786e-05,
+      "loss": 0.3489,
+      "step": 6261
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.7758682420336556e-05,
+      "loss": 0.2642,
+      "step": 6262
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.775510204081633e-05,
+      "loss": 0.1902,
+      "step": 6263
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.77515216612961e-05,
+      "loss": 0.1646,
+      "step": 6264
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.774794128177587e-05,
+      "loss": 0.288,
+      "step": 6265
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.774436090225564e-05,
+      "loss": 0.1842,
+      "step": 6266
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.774078052273541e-05,
+      "loss": 0.1598,
+      "step": 6267
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.773720014321518e-05,
+      "loss": 0.1721,
+      "step": 6268
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.773361976369495e-05,
+      "loss": 0.2355,
+      "step": 6269
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.773003938417472e-05,
+      "loss": 0.2365,
+      "step": 6270
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.772645900465449e-05,
+      "loss": 0.3611,
+      "step": 6271
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7722878625134264e-05,
+      "loss": 0.3887,
+      "step": 6272
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7719298245614034e-05,
+      "loss": 0.273,
+      "step": 6273
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.771571786609381e-05,
+      "loss": 0.2352,
+      "step": 6274
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.771213748657358e-05,
+      "loss": 0.3373,
+      "step": 6275
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.770855710705335e-05,
+      "loss": 0.2066,
+      "step": 6276
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7704976727533123e-05,
+      "loss": 0.3488,
+      "step": 6277
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7701396348012893e-05,
+      "loss": 0.3545,
+      "step": 6278
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7697815968492663e-05,
+      "loss": 0.2164,
+      "step": 6279
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7694235588972433e-05,
+      "loss": 0.1564,
+      "step": 6280
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7690655209452206e-05,
+      "loss": 0.1999,
+      "step": 6281
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7687074829931976e-05,
+      "loss": 0.1805,
+      "step": 6282
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7683494450411745e-05,
+      "loss": 0.2252,
+      "step": 6283
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7679914070891515e-05,
+      "loss": 0.1748,
+      "step": 6284
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.767633369137129e-05,
+      "loss": 0.1959,
+      "step": 6285
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7672753311851058e-05,
+      "loss": 0.2118,
+      "step": 6286
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7669172932330828e-05,
+      "loss": 0.2195,
+      "step": 6287
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7665592552810598e-05,
+      "loss": 0.2277,
+      "step": 6288
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.7662012173290368e-05,
+      "loss": 0.2823,
+      "step": 6289
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.765843179377014e-05,
+      "loss": 0.2421,
+      "step": 6290
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.765485141424991e-05,
+      "loss": 0.2482,
+      "step": 6291
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.765127103472968e-05,
+      "loss": 0.2444,
+      "step": 6292
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.764769065520945e-05,
+      "loss": 0.2285,
+      "step": 6293
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7644110275689223e-05,
+      "loss": 0.1906,
+      "step": 6294
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7640529896168993e-05,
+      "loss": 0.2037,
+      "step": 6295
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7636949516648763e-05,
+      "loss": 0.1867,
+      "step": 6296
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.763336913712854e-05,
+      "loss": 0.2461,
+      "step": 6297
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.762978875760831e-05,
+      "loss": 0.2079,
+      "step": 6298
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7626208378088083e-05,
+      "loss": 0.2083,
+      "step": 6299
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7622627998567852e-05,
+      "loss": 0.3619,
+      "step": 6300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7619047619047622e-05,
+      "loss": 0.2049,
+      "step": 6301
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7615467239527392e-05,
+      "loss": 0.2223,
+      "step": 6302
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7611886860007162e-05,
+      "loss": 0.2126,
+      "step": 6303
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7608306480486935e-05,
+      "loss": 0.2992,
+      "step": 6304
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7604726100966705e-05,
+      "loss": 0.4927,
+      "step": 6305
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7601145721446474e-05,
+      "loss": 0.2648,
+      "step": 6306
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7597565341926244e-05,
+      "loss": 0.1576,
+      "step": 6307
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7593984962406017e-05,
+      "loss": 0.3247,
+      "step": 6308
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7590404582885787e-05,
+      "loss": 0.2109,
+      "step": 6309
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7586824203365557e-05,
+      "loss": 0.1974,
+      "step": 6310
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7583243823845327e-05,
+      "loss": 0.1479,
+      "step": 6311
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.75796634443251e-05,
+      "loss": 0.1934,
+      "step": 6312
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.757608306480487e-05,
+      "loss": 0.1425,
+      "step": 6313
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.757250268528464e-05,
+      "loss": 0.1957,
+      "step": 6314
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.756892230576441e-05,
+      "loss": 0.1303,
+      "step": 6315
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7565341926244183e-05,
+      "loss": 0.1177,
+      "step": 6316
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7561761546723952e-05,
+      "loss": 0.2144,
+      "step": 6317
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7558181167203722e-05,
+      "loss": 0.1818,
+      "step": 6318
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7554600787683492e-05,
+      "loss": 0.1771,
+      "step": 6319
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.7551020408163265e-05,
+      "loss": 0.1817,
+      "step": 6320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.754744002864304e-05,
+      "loss": 0.1897,
+      "step": 6321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.754385964912281e-05,
+      "loss": 0.2388,
+      "step": 6322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.754027926960258e-05,
+      "loss": 0.1934,
+      "step": 6323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.753669889008235e-05,
+      "loss": 0.2357,
+      "step": 6324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.753311851056212e-05,
+      "loss": 0.2951,
+      "step": 6325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7529538131041894e-05,
+      "loss": 0.1829,
+      "step": 6326
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7525957751521664e-05,
+      "loss": 0.1314,
+      "step": 6327
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7522377372001434e-05,
+      "loss": 0.2323,
+      "step": 6328
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7518796992481204e-05,
+      "loss": 0.371,
+      "step": 6329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7515216612960977e-05,
+      "loss": 0.1988,
+      "step": 6330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7511636233440747e-05,
+      "loss": 0.2506,
+      "step": 6331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7508055853920516e-05,
+      "loss": 0.1101,
+      "step": 6332
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7504475474400286e-05,
+      "loss": 0.1211,
+      "step": 6333
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.750089509488006e-05,
+      "loss": 0.1641,
+      "step": 6334
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.749731471535983e-05,
+      "loss": 0.2899,
+      "step": 6335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.74937343358396e-05,
+      "loss": 0.3253,
+      "step": 6336
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.749015395631937e-05,
+      "loss": 0.3333,
+      "step": 6337
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7486573576799142e-05,
+      "loss": 0.4125,
+      "step": 6338
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7482993197278912e-05,
+      "loss": 0.1351,
+      "step": 6339
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.747941281775868e-05,
+      "loss": 0.2149,
+      "step": 6340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.747583243823845e-05,
+      "loss": 0.1674,
+      "step": 6341
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7472252058718225e-05,
+      "loss": 0.2223,
+      "step": 6342
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7468671679197994e-05,
+      "loss": 0.2027,
+      "step": 6343
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.746509129967777e-05,
+      "loss": 0.2688,
+      "step": 6344
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.746151092015754e-05,
+      "loss": 0.1715,
+      "step": 6345
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.745793054063731e-05,
+      "loss": 0.2127,
+      "step": 6346
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.745435016111708e-05,
+      "loss": 0.2988,
+      "step": 6347
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.7450769781596854e-05,
+      "loss": 0.31,
+      "step": 6348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7447189402076623e-05,
+      "loss": 0.2452,
+      "step": 6349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7443609022556393e-05,
+      "loss": 0.1128,
+      "step": 6350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7440028643036163e-05,
+      "loss": 0.1121,
+      "step": 6351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7436448263515936e-05,
+      "loss": 0.1552,
+      "step": 6352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7432867883995706e-05,
+      "loss": 0.1711,
+      "step": 6353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7429287504475476e-05,
+      "loss": 0.1532,
+      "step": 6354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7425707124955245e-05,
+      "loss": 0.2031,
+      "step": 6355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.742212674543502e-05,
+      "loss": 0.219,
+      "step": 6356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.741854636591479e-05,
+      "loss": 0.3256,
+      "step": 6357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7414965986394558e-05,
+      "loss": 0.1686,
+      "step": 6358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7411385606874328e-05,
+      "loss": 0.2977,
+      "step": 6359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.74078052273541e-05,
+      "loss": 0.2683,
+      "step": 6360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.740422484783387e-05,
+      "loss": 0.2336,
+      "step": 6361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.740064446831364e-05,
+      "loss": 0.1674,
+      "step": 6362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.739706408879341e-05,
+      "loss": 0.1767,
+      "step": 6363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7393483709273184e-05,
+      "loss": 0.2167,
+      "step": 6364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7389903329752954e-05,
+      "loss": 0.1941,
+      "step": 6365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7386322950232723e-05,
+      "loss": 0.1589,
+      "step": 6366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.73827425707125e-05,
+      "loss": 0.3289,
+      "step": 6367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.737916219119227e-05,
+      "loss": 0.2559,
+      "step": 6368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.737558181167204e-05,
+      "loss": 0.157,
+      "step": 6369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7372001432151813e-05,
+      "loss": 0.346,
+      "step": 6370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7368421052631583e-05,
+      "loss": 0.1919,
+      "step": 6371
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7364840673111352e-05,
+      "loss": 0.1602,
+      "step": 6372
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7361260293591122e-05,
+      "loss": 0.1607,
+      "step": 6373
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7357679914070895e-05,
+      "loss": 0.1219,
+      "step": 6374
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7354099534550665e-05,
+      "loss": 0.2272,
+      "step": 6375
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7350519155030435e-05,
+      "loss": 0.1635,
+      "step": 6376
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7346938775510205e-05,
+      "loss": 0.298,
+      "step": 6377
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7343358395989978e-05,
+      "loss": 0.2804,
+      "step": 6378
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7339778016469748e-05,
+      "loss": 0.1582,
+      "step": 6379
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7336197636949518e-05,
+      "loss": 0.2211,
+      "step": 6380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7332617257429287e-05,
+      "loss": 0.2846,
+      "step": 6381
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7329036877909057e-05,
+      "loss": 0.2099,
+      "step": 6382
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.732545649838883e-05,
+      "loss": 0.2777,
+      "step": 6383
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.73218761188686e-05,
+      "loss": 0.1545,
+      "step": 6384
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.731829573934837e-05,
+      "loss": 0.2145,
+      "step": 6385
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.731471535982814e-05,
+      "loss": 0.3576,
+      "step": 6386
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7311134980307913e-05,
+      "loss": 0.1583,
+      "step": 6387
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7307554600787683e-05,
+      "loss": 0.2064,
+      "step": 6388
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7303974221267453e-05,
+      "loss": 0.2825,
+      "step": 6389
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7300393841747222e-05,
+      "loss": 0.0934,
+      "step": 6390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7296813462227e-05,
+      "loss": 0.3295,
+      "step": 6391
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7293233082706772e-05,
+      "loss": 0.2739,
+      "step": 6392
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7289652703186542e-05,
+      "loss": 0.283,
+      "step": 6393
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7286072323666312e-05,
+      "loss": 0.325,
+      "step": 6394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.728249194414608e-05,
+      "loss": 0.2225,
+      "step": 6395
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7278911564625855e-05,
+      "loss": 0.1702,
+      "step": 6396
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7275331185105625e-05,
+      "loss": 0.3223,
+      "step": 6397
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7271750805585394e-05,
+      "loss": 0.2725,
+      "step": 6398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7268170426065164e-05,
+      "loss": 0.2124,
+      "step": 6399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7264590046544934e-05,
+      "loss": 0.2414,
+      "step": 6400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7261009667024707e-05,
+      "loss": 0.2445,
+      "step": 6401
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7257429287504477e-05,
+      "loss": 0.2101,
+      "step": 6402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7253848907984247e-05,
+      "loss": 0.1528,
+      "step": 6403
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7250268528464016e-05,
+      "loss": 0.2515,
+      "step": 6404
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.724668814894379e-05,
+      "loss": 0.3168,
+      "step": 6405
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.724310776942356e-05,
+      "loss": 0.1962,
+      "step": 6406
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.723952738990333e-05,
+      "loss": 0.2533,
+      "step": 6407
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.72359470103831e-05,
+      "loss": 0.1428,
+      "step": 6408
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7232366630862872e-05,
+      "loss": 0.2743,
+      "step": 6409
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7228786251342642e-05,
+      "loss": 0.2447,
+      "step": 6410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7225205871822412e-05,
+      "loss": 0.1506,
+      "step": 6411
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.722162549230218e-05,
+      "loss": 0.3029,
+      "step": 6412
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7218045112781955e-05,
+      "loss": 0.2128,
+      "step": 6413
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.721446473326173e-05,
+      "loss": 0.273,
+      "step": 6414
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.72108843537415e-05,
+      "loss": 0.1368,
+      "step": 6415
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.720730397422127e-05,
+      "loss": 0.1697,
+      "step": 6416
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.720372359470104e-05,
+      "loss": 0.2722,
+      "step": 6417
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.720014321518081e-05,
+      "loss": 0.3735,
+      "step": 6418
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7196562835660584e-05,
+      "loss": 0.1812,
+      "step": 6419
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7192982456140354e-05,
+      "loss": 0.2464,
+      "step": 6420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7189402076620123e-05,
+      "loss": 0.1872,
+      "step": 6421
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7185821697099893e-05,
+      "loss": 0.2611,
+      "step": 6422
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7182241317579666e-05,
+      "loss": 0.3311,
+      "step": 6423
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7178660938059436e-05,
+      "loss": 0.2679,
+      "step": 6424
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7175080558539206e-05,
+      "loss": 0.2061,
+      "step": 6425
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7171500179018976e-05,
+      "loss": 0.2484,
+      "step": 6426
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.716791979949875e-05,
+      "loss": 0.1705,
+      "step": 6427
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.716433941997852e-05,
+      "loss": 0.2023,
+      "step": 6428
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.716075904045829e-05,
+      "loss": 0.17,
+      "step": 6429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.715717866093806e-05,
+      "loss": 0.292,
+      "step": 6430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.715359828141783e-05,
+      "loss": 0.2602,
+      "step": 6431
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.71500179018976e-05,
+      "loss": 0.3682,
+      "step": 6432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.714643752237737e-05,
+      "loss": 0.3355,
+      "step": 6433
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.714285714285714e-05,
+      "loss": 0.2365,
+      "step": 6434
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7139276763336914e-05,
+      "loss": 0.1547,
+      "step": 6435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7135696383816684e-05,
+      "loss": 0.1833,
+      "step": 6436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7132116004296454e-05,
+      "loss": 0.3095,
+      "step": 6437
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712853562477623e-05,
+      "loss": 0.2434,
+      "step": 6438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7124955245256e-05,
+      "loss": 0.1372,
+      "step": 6439
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712137486573577e-05,
+      "loss": 0.1674,
+      "step": 6440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7117794486215543e-05,
+      "loss": 0.1846,
+      "step": 6441
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7114214106695313e-05,
+      "loss": 0.236,
+      "step": 6442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7110633727175083e-05,
+      "loss": 0.2788,
+      "step": 6443
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7107053347654853e-05,
+      "loss": 0.2046,
+      "step": 6444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7103472968134626e-05,
+      "loss": 0.2305,
+      "step": 6445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7099892588614395e-05,
+      "loss": 0.2714,
+      "step": 6446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7096312209094165e-05,
+      "loss": 0.2558,
+      "step": 6447
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7092731829573935e-05,
+      "loss": 0.1246,
+      "step": 6448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7089151450053708e-05,
+      "loss": 0.2379,
+      "step": 6449
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7085571070533478e-05,
+      "loss": 0.3614,
+      "step": 6450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7081990691013248e-05,
+      "loss": 0.2561,
+      "step": 6451
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7078410311493018e-05,
+      "loss": 0.286,
+      "step": 6452
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.707482993197279e-05,
+      "loss": 0.2143,
+      "step": 6453
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.707124955245256e-05,
+      "loss": 0.1903,
+      "step": 6454
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.706766917293233e-05,
+      "loss": 0.1845,
+      "step": 6455
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.70640887934121e-05,
+      "loss": 0.3552,
+      "step": 6456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7060508413891873e-05,
+      "loss": 0.2268,
+      "step": 6457
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7056928034371643e-05,
+      "loss": 0.19,
+      "step": 6458
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7053347654851413e-05,
+      "loss": 0.2451,
+      "step": 6459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7049767275331183e-05,
+      "loss": 0.2019,
+      "step": 6460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.704618689581096e-05,
+      "loss": 0.1904,
+      "step": 6461
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.704260651629073e-05,
+      "loss": 0.23,
+      "step": 6462
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7039026136770502e-05,
+      "loss": 0.2314,
+      "step": 6463
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7035445757250272e-05,
+      "loss": 0.2356,
+      "step": 6464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7031865377730042e-05,
+      "loss": 0.3098,
+      "step": 6465
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7028284998209812e-05,
+      "loss": 0.1863,
+      "step": 6466
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7024704618689585e-05,
+      "loss": 0.1804,
+      "step": 6467
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7021124239169355e-05,
+      "loss": 0.1165,
+      "step": 6468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7017543859649125e-05,
+      "loss": 0.2233,
+      "step": 6469
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7013963480128894e-05,
+      "loss": 0.4796,
+      "step": 6470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7010383100608668e-05,
+      "loss": 0.3219,
+      "step": 6471
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7006802721088437e-05,
+      "loss": 0.2264,
+      "step": 6472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.7003222341568207e-05,
+      "loss": 0.2285,
+      "step": 6473
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6999641962047977e-05,
+      "loss": 0.2182,
+      "step": 6474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.699606158252775e-05,
+      "loss": 0.1879,
+      "step": 6475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.699248120300752e-05,
+      "loss": 0.1946,
+      "step": 6476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.698890082348729e-05,
+      "loss": 0.29,
+      "step": 6477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.698532044396706e-05,
+      "loss": 0.2056,
+      "step": 6478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.698174006444683e-05,
+      "loss": 0.356,
+      "step": 6479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6978159684926603e-05,
+      "loss": 0.1958,
+      "step": 6480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6974579305406372e-05,
+      "loss": 0.3719,
+      "step": 6481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6970998925886142e-05,
+      "loss": 0.2627,
+      "step": 6482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6967418546365912e-05,
+      "loss": 0.218,
+      "step": 6483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.696383816684569e-05,
+      "loss": 0.1955,
+      "step": 6484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6960257787325462e-05,
+      "loss": 0.198,
+      "step": 6485
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.695667740780523e-05,
+      "loss": 0.2798,
+      "step": 6486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6953097028285e-05,
+      "loss": 0.2625,
+      "step": 6487
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.694951664876477e-05,
+      "loss": 0.3111,
+      "step": 6488
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6945936269244544e-05,
+      "loss": 0.2606,
+      "step": 6489
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6942355889724314e-05,
+      "loss": 0.1329,
+      "step": 6490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6938775510204084e-05,
+      "loss": 0.1681,
+      "step": 6491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6935195130683854e-05,
+      "loss": 0.1779,
+      "step": 6492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6931614751163627e-05,
+      "loss": 0.2853,
+      "step": 6493
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6928034371643397e-05,
+      "loss": 0.157,
+      "step": 6494
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6924453992123166e-05,
+      "loss": 0.1553,
+      "step": 6495
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6920873612602936e-05,
+      "loss": 0.2826,
+      "step": 6496
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6917293233082706e-05,
+      "loss": 0.2711,
+      "step": 6497
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.691371285356248e-05,
+      "loss": 0.2761,
+      "step": 6498
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.691013247404225e-05,
+      "loss": 0.1976,
+      "step": 6499
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.690655209452202e-05,
+      "loss": 0.2564,
+      "step": 6500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.690297171500179e-05,
+      "loss": 0.4854,
+      "step": 6501
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6899391335481562e-05,
+      "loss": 0.1687,
+      "step": 6502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.689581095596133e-05,
+      "loss": 0.1585,
+      "step": 6503
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.68922305764411e-05,
+      "loss": 0.2875,
+      "step": 6504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.688865019692087e-05,
+      "loss": 0.3192,
+      "step": 6505
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6885069817400644e-05,
+      "loss": 0.1785,
+      "step": 6506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6881489437880414e-05,
+      "loss": 0.2415,
+      "step": 6507
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.687790905836019e-05,
+      "loss": 0.2683,
+      "step": 6508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.687432867883996e-05,
+      "loss": 0.2026,
+      "step": 6509
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.687074829931973e-05,
+      "loss": 0.2255,
+      "step": 6510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.68671679197995e-05,
+      "loss": 0.2531,
+      "step": 6511
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6863587540279273e-05,
+      "loss": 0.2294,
+      "step": 6512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6860007160759043e-05,
+      "loss": 0.3548,
+      "step": 6513
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6856426781238813e-05,
+      "loss": 0.2534,
+      "step": 6514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6852846401718583e-05,
+      "loss": 0.3592,
+      "step": 6515
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6849266022198356e-05,
+      "loss": 0.2632,
+      "step": 6516
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6845685642678126e-05,
+      "loss": 0.158,
+      "step": 6517
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6842105263157896e-05,
+      "loss": 0.4631,
+      "step": 6518
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6838524883637665e-05,
+      "loss": 0.1804,
+      "step": 6519
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.683494450411744e-05,
+      "loss": 0.244,
+      "step": 6520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.683136412459721e-05,
+      "loss": 0.1371,
+      "step": 6521
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6827783745076978e-05,
+      "loss": 0.2255,
+      "step": 6522
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6824203365556748e-05,
+      "loss": 0.2003,
+      "step": 6523
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.682062298603652e-05,
+      "loss": 0.2838,
+      "step": 6524
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.681704260651629e-05,
+      "loss": 0.2257,
+      "step": 6525
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.681346222699606e-05,
+      "loss": 0.2738,
+      "step": 6526
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.680988184747583e-05,
+      "loss": 0.163,
+      "step": 6527
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6806301467955604e-05,
+      "loss": 0.148,
+      "step": 6528
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6802721088435374e-05,
+      "loss": 0.2298,
+      "step": 6529
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6799140708915143e-05,
+      "loss": 0.3181,
+      "step": 6530
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.679556032939492e-05,
+      "loss": 0.2098,
+      "step": 6531
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.679197994987469e-05,
+      "loss": 0.0993,
+      "step": 6532
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.678839957035446e-05,
+      "loss": 0.1576,
+      "step": 6533
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6784819190834233e-05,
+      "loss": 0.2199,
+      "step": 6534
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6781238811314003e-05,
+      "loss": 0.204,
+      "step": 6535
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6777658431793772e-05,
+      "loss": 0.1073,
+      "step": 6536
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6774078052273542e-05,
+      "loss": 0.209,
+      "step": 6537
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6770497672753315e-05,
+      "loss": 0.4356,
+      "step": 6538
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6766917293233085e-05,
+      "loss": 0.1917,
+      "step": 6539
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6763336913712855e-05,
+      "loss": 0.3294,
+      "step": 6540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6759756534192625e-05,
+      "loss": 0.2122,
+      "step": 6541
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6756176154672398e-05,
+      "loss": 0.3024,
+      "step": 6542
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6752595775152168e-05,
+      "loss": 0.2234,
+      "step": 6543
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6749015395631937e-05,
+      "loss": 0.2635,
+      "step": 6544
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.6745435016111707e-05,
+      "loss": 0.2067,
+      "step": 6545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.674185463659148e-05,
+      "loss": 0.2524,
+      "step": 6546
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.673827425707125e-05,
+      "loss": 0.287,
+      "step": 6547
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.673469387755102e-05,
+      "loss": 0.246,
+      "step": 6548
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.673111349803079e-05,
+      "loss": 0.2437,
+      "step": 6549
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6727533118510563e-05,
+      "loss": 0.2408,
+      "step": 6550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6723952738990333e-05,
+      "loss": 0.179,
+      "step": 6551
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6720372359470103e-05,
+      "loss": 0.2282,
+      "step": 6552
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6716791979949872e-05,
+      "loss": 0.3211,
+      "step": 6553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.671321160042965e-05,
+      "loss": 0.1115,
+      "step": 6554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.670963122090942e-05,
+      "loss": 0.1139,
+      "step": 6555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6706050841389192e-05,
+      "loss": 0.2627,
+      "step": 6556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6702470461868962e-05,
+      "loss": 0.1722,
+      "step": 6557
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.669889008234873e-05,
+      "loss": 0.144,
+      "step": 6558
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.66953097028285e-05,
+      "loss": 0.224,
+      "step": 6559
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6691729323308275e-05,
+      "loss": 0.222,
+      "step": 6560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6688148943788044e-05,
+      "loss": 0.1933,
+      "step": 6561
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6684568564267814e-05,
+      "loss": 0.1554,
+      "step": 6562
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6680988184747584e-05,
+      "loss": 0.2702,
+      "step": 6563
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6677407805227357e-05,
+      "loss": 0.2567,
+      "step": 6564
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6673827425707127e-05,
+      "loss": 0.1583,
+      "step": 6565
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6670247046186897e-05,
+      "loss": 0.1748,
+      "step": 6566
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.3838,
+      "step": 6567
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.666308628714644e-05,
+      "loss": 0.2401,
+      "step": 6568
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.665950590762621e-05,
+      "loss": 0.2667,
+      "step": 6569
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.665592552810598e-05,
+      "loss": 0.3085,
+      "step": 6570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.665234514858575e-05,
+      "loss": 0.2587,
+      "step": 6571
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.664876476906552e-05,
+      "loss": 0.1368,
+      "step": 6572
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.6645184389545292e-05,
+      "loss": 0.2321,
+      "step": 6573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6641604010025062e-05,
+      "loss": 0.1825,
+      "step": 6574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6638023630504832e-05,
+      "loss": 0.2081,
+      "step": 6575
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.66344432509846e-05,
+      "loss": 0.273,
+      "step": 6576
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6630862871464375e-05,
+      "loss": 0.1796,
+      "step": 6577
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.662728249194415e-05,
+      "loss": 0.2871,
+      "step": 6578
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.662370211242392e-05,
+      "loss": 0.1407,
+      "step": 6579
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.662012173290369e-05,
+      "loss": 0.2664,
+      "step": 6580
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.661654135338346e-05,
+      "loss": 0.1665,
+      "step": 6581
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6612960973863234e-05,
+      "loss": 0.1807,
+      "step": 6582
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6609380594343004e-05,
+      "loss": 0.1232,
+      "step": 6583
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6605800214822774e-05,
+      "loss": 0.3501,
+      "step": 6584
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6602219835302543e-05,
+      "loss": 0.2632,
+      "step": 6585
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6598639455782316e-05,
+      "loss": 0.2708,
+      "step": 6586
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6595059076262086e-05,
+      "loss": 0.1413,
+      "step": 6587
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6591478696741856e-05,
+      "loss": 0.2983,
+      "step": 6588
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6587898317221626e-05,
+      "loss": 0.1604,
+      "step": 6589
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6584317937701396e-05,
+      "loss": 0.1678,
+      "step": 6590
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.658073755818117e-05,
+      "loss": 0.2078,
+      "step": 6591
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.657715717866094e-05,
+      "loss": 0.1969,
+      "step": 6592
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.657357679914071e-05,
+      "loss": 0.1847,
+      "step": 6593
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6569996419620478e-05,
+      "loss": 0.1966,
+      "step": 6594
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.656641604010025e-05,
+      "loss": 0.2671,
+      "step": 6595
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.656283566058002e-05,
+      "loss": 0.2646,
+      "step": 6596
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.655925528105979e-05,
+      "loss": 0.3341,
+      "step": 6597
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.655567490153956e-05,
+      "loss": 0.224,
+      "step": 6598
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6552094522019334e-05,
+      "loss": 0.2286,
+      "step": 6599
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6548514142499104e-05,
+      "loss": 0.1938,
+      "step": 6600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.654493376297888e-05,
+      "loss": 0.2369,
+      "step": 6601
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.654135338345865e-05,
+      "loss": 0.1784,
+      "step": 6602
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.653777300393842e-05,
+      "loss": 0.1292,
+      "step": 6603
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6534192624418193e-05,
+      "loss": 0.2533,
+      "step": 6604
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6530612244897963e-05,
+      "loss": 0.3243,
+      "step": 6605
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6527031865377733e-05,
+      "loss": 0.2697,
+      "step": 6606
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6523451485857503e-05,
+      "loss": 0.1581,
+      "step": 6607
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6519871106337272e-05,
+      "loss": 0.1792,
+      "step": 6608
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6516290726817046e-05,
+      "loss": 0.1771,
+      "step": 6609
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6512710347296815e-05,
+      "loss": 0.2281,
+      "step": 6610
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6509129967776585e-05,
+      "loss": 0.2116,
+      "step": 6611
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6505549588256355e-05,
+      "loss": 0.1709,
+      "step": 6612
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6501969208736128e-05,
+      "loss": 0.1973,
+      "step": 6613
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6498388829215898e-05,
+      "loss": 0.3662,
+      "step": 6614
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6494808449695668e-05,
+      "loss": 0.1799,
+      "step": 6615
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6491228070175438e-05,
+      "loss": 0.1668,
+      "step": 6616
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.648764769065521e-05,
+      "loss": 0.281,
+      "step": 6617
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.648406731113498e-05,
+      "loss": 0.3233,
+      "step": 6618
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.648048693161475e-05,
+      "loss": 0.1829,
+      "step": 6619
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.647690655209452e-05,
+      "loss": 0.2077,
+      "step": 6620
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6473326172574293e-05,
+      "loss": 0.2955,
+      "step": 6621
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6469745793054063e-05,
+      "loss": 0.2179,
+      "step": 6622
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6466165413533833e-05,
+      "loss": 0.2245,
+      "step": 6623
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.646258503401361e-05,
+      "loss": 0.3174,
+      "step": 6624
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.645900465449338e-05,
+      "loss": 0.2961,
+      "step": 6625
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.645542427497315e-05,
+      "loss": 0.3134,
+      "step": 6626
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6451843895452922e-05,
+      "loss": 0.1724,
+      "step": 6627
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6448263515932692e-05,
+      "loss": 0.385,
+      "step": 6628
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6444683136412462e-05,
+      "loss": 0.2801,
+      "step": 6629
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6441102756892232e-05,
+      "loss": 0.2878,
+      "step": 6630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6437522377372005e-05,
+      "loss": 0.2513,
+      "step": 6631
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6433941997851775e-05,
+      "loss": 0.224,
+      "step": 6632
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6430361618331545e-05,
+      "loss": 0.3039,
+      "step": 6633
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6426781238811314e-05,
+      "loss": 0.2476,
+      "step": 6634
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6423200859291087e-05,
+      "loss": 0.2334,
+      "step": 6635
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6419620479770857e-05,
+      "loss": 0.2228,
+      "step": 6636
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6416040100250627e-05,
+      "loss": 0.187,
+      "step": 6637
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6412459720730397e-05,
+      "loss": 0.2677,
+      "step": 6638
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.640887934121017e-05,
+      "loss": 0.2217,
+      "step": 6639
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.640529896168994e-05,
+      "loss": 0.2916,
+      "step": 6640
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.640171858216971e-05,
+      "loss": 0.2156,
+      "step": 6641
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.639813820264948e-05,
+      "loss": 0.1908,
+      "step": 6642
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6394557823129253e-05,
+      "loss": 0.3119,
+      "step": 6643
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6390977443609022e-05,
+      "loss": 0.255,
+      "step": 6644
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6387397064088792e-05,
+      "loss": 0.2717,
+      "step": 6645
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6383816684568562e-05,
+      "loss": 0.2837,
+      "step": 6646
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6380236305048335e-05,
+      "loss": 0.3229,
+      "step": 6647
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.637665592552811e-05,
+      "loss": 0.2437,
+      "step": 6648
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.637307554600788e-05,
+      "loss": 0.2992,
+      "step": 6649
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.636949516648765e-05,
+      "loss": 0.2078,
+      "step": 6650
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.636591478696742e-05,
+      "loss": 0.1607,
+      "step": 6651
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.636233440744719e-05,
+      "loss": 0.151,
+      "step": 6652
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6358754027926964e-05,
+      "loss": 0.2819,
+      "step": 6653
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6355173648406734e-05,
+      "loss": 0.2969,
+      "step": 6654
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6351593268886504e-05,
+      "loss": 0.1802,
+      "step": 6655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6348012889366274e-05,
+      "loss": 0.2911,
+      "step": 6656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.6344432509846047e-05,
+      "loss": 0.1063,
+      "step": 6657
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6340852130325817e-05,
+      "loss": 0.1846,
+      "step": 6658
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6337271750805586e-05,
+      "loss": 0.2223,
+      "step": 6659
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6333691371285356e-05,
+      "loss": 0.314,
+      "step": 6660
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.633011099176513e-05,
+      "loss": 0.4379,
+      "step": 6661
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.63265306122449e-05,
+      "loss": 0.3538,
+      "step": 6662
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.632295023272467e-05,
+      "loss": 0.1784,
+      "step": 6663
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.631936985320444e-05,
+      "loss": 0.1871,
+      "step": 6664
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.1613,
+      "step": 6665
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6312209094163982e-05,
+      "loss": 0.2532,
+      "step": 6666
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.630862871464375e-05,
+      "loss": 0.2872,
+      "step": 6667
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.630504833512352e-05,
+      "loss": 0.136,
+      "step": 6668
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.630146795560329e-05,
+      "loss": 0.1709,
+      "step": 6669
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6297887576083064e-05,
+      "loss": 0.2436,
+      "step": 6670
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.629430719656284e-05,
+      "loss": 0.2125,
+      "step": 6671
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.629072681704261e-05,
+      "loss": 0.2645,
+      "step": 6672
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.628714643752238e-05,
+      "loss": 0.1907,
+      "step": 6673
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.628356605800215e-05,
+      "loss": 0.1467,
+      "step": 6674
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6279985678481924e-05,
+      "loss": 0.3336,
+      "step": 6675
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6276405298961693e-05,
+      "loss": 0.1996,
+      "step": 6676
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6272824919441463e-05,
+      "loss": 0.1981,
+      "step": 6677
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6269244539921233e-05,
+      "loss": 0.3046,
+      "step": 6678
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6265664160401006e-05,
+      "loss": 0.3097,
+      "step": 6679
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6262083780880776e-05,
+      "loss": 0.2979,
+      "step": 6680
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6258503401360546e-05,
+      "loss": 0.2945,
+      "step": 6681
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6254923021840315e-05,
+      "loss": 0.3317,
+      "step": 6682
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.625134264232009e-05,
+      "loss": 0.2562,
+      "step": 6683
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.624776226279986e-05,
+      "loss": 0.2389,
+      "step": 6684
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.6244181883279628e-05,
+      "loss": 0.2007,
+      "step": 6685
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6240601503759398e-05,
+      "loss": 0.2898,
+      "step": 6686
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6237021124239168e-05,
+      "loss": 0.1352,
+      "step": 6687
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.623344074471894e-05,
+      "loss": 0.1927,
+      "step": 6688
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.622986036519871e-05,
+      "loss": 0.1361,
+      "step": 6689
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.622627998567848e-05,
+      "loss": 0.2836,
+      "step": 6690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.622269960615825e-05,
+      "loss": 0.2283,
+      "step": 6691
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6219119226638024e-05,
+      "loss": 0.1434,
+      "step": 6692
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6215538847117793e-05,
+      "loss": 0.2022,
+      "step": 6693
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.621195846759757e-05,
+      "loss": 0.1884,
+      "step": 6694
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.620837808807734e-05,
+      "loss": 0.3671,
+      "step": 6695
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.620479770855711e-05,
+      "loss": 0.2276,
+      "step": 6696
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6201217329036883e-05,
+      "loss": 0.251,
+      "step": 6697
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6197636949516653e-05,
+      "loss": 0.2272,
+      "step": 6698
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6194056569996422e-05,
+      "loss": 0.1394,
+      "step": 6699
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.2027,
+      "step": 6700
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6186895810955962e-05,
+      "loss": 0.2255,
+      "step": 6701
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6183315431435735e-05,
+      "loss": 0.2836,
+      "step": 6702
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6179735051915505e-05,
+      "loss": 0.2201,
+      "step": 6703
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6176154672395275e-05,
+      "loss": 0.2359,
+      "step": 6704
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6172574292875045e-05,
+      "loss": 0.2286,
+      "step": 6705
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6168993913354818e-05,
+      "loss": 0.2833,
+      "step": 6706
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6165413533834588e-05,
+      "loss": 0.1306,
+      "step": 6707
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6161833154314357e-05,
+      "loss": 0.1597,
+      "step": 6708
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.6158252774794127e-05,
+      "loss": 0.1977,
+      "step": 6709
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.61546723952739e-05,
+      "loss": 0.1802,
+      "step": 6710
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.615109201575367e-05,
+      "loss": 0.221,
+      "step": 6711
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.614751163623344e-05,
+      "loss": 0.2361,
+      "step": 6712
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.614393125671321e-05,
+      "loss": 0.337,
+      "step": 6713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6140350877192983e-05,
+      "loss": 0.1926,
+      "step": 6714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6136770497672753e-05,
+      "loss": 0.1181,
+      "step": 6715
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6133190118152523e-05,
+      "loss": 0.0867,
+      "step": 6716
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6129609738632292e-05,
+      "loss": 0.2575,
+      "step": 6717
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.612602935911207e-05,
+      "loss": 0.2166,
+      "step": 6718
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.612244897959184e-05,
+      "loss": 0.2022,
+      "step": 6719
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6118868600071612e-05,
+      "loss": 0.3882,
+      "step": 6720
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6115288220551382e-05,
+      "loss": 0.2786,
+      "step": 6721
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.611170784103115e-05,
+      "loss": 0.2521,
+      "step": 6722
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.610812746151092e-05,
+      "loss": 0.1987,
+      "step": 6723
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6104547081990695e-05,
+      "loss": 0.1833,
+      "step": 6724
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6100966702470464e-05,
+      "loss": 0.195,
+      "step": 6725
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6097386322950234e-05,
+      "loss": 0.1889,
+      "step": 6726
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6093805943430004e-05,
+      "loss": 0.197,
+      "step": 6727
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6090225563909777e-05,
+      "loss": 0.1716,
+      "step": 6728
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6086645184389547e-05,
+      "loss": 0.123,
+      "step": 6729
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6083064804869317e-05,
+      "loss": 0.2136,
+      "step": 6730
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6079484425349086e-05,
+      "loss": 0.1444,
+      "step": 6731
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.607590404582886e-05,
+      "loss": 0.2068,
+      "step": 6732
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.607232366630863e-05,
+      "loss": 0.2343,
+      "step": 6733
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.60687432867884e-05,
+      "loss": 0.2998,
+      "step": 6734
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.606516290726817e-05,
+      "loss": 0.2198,
+      "step": 6735
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6061582527747942e-05,
+      "loss": 0.2527,
+      "step": 6736
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6058002148227712e-05,
+      "loss": 0.2558,
+      "step": 6737
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6054421768707482e-05,
+      "loss": 0.1672,
+      "step": 6738
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.605084138918725e-05,
+      "loss": 0.1733,
+      "step": 6739
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6047261009667025e-05,
+      "loss": 0.3747,
+      "step": 6740
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.6043680630146798e-05,
+      "loss": 0.237,
+      "step": 6741
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.604010025062657e-05,
+      "loss": 0.3993,
+      "step": 6742
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.603651987110634e-05,
+      "loss": 0.2103,
+      "step": 6743
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.603293949158611e-05,
+      "loss": 0.1937,
+      "step": 6744
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.602935911206588e-05,
+      "loss": 0.178,
+      "step": 6745
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6025778732545654e-05,
+      "loss": 0.2028,
+      "step": 6746
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6022198353025424e-05,
+      "loss": 0.2867,
+      "step": 6747
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6018617973505193e-05,
+      "loss": 0.1997,
+      "step": 6748
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6015037593984963e-05,
+      "loss": 0.2444,
+      "step": 6749
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6011457214464736e-05,
+      "loss": 0.3448,
+      "step": 6750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6007876834944506e-05,
+      "loss": 0.2171,
+      "step": 6751
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6004296455424276e-05,
+      "loss": 0.2906,
+      "step": 6752
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.6000716075904046e-05,
+      "loss": 0.2784,
+      "step": 6753
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.599713569638382e-05,
+      "loss": 0.1731,
+      "step": 6754
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.599355531686359e-05,
+      "loss": 0.3061,
+      "step": 6755
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.598997493734336e-05,
+      "loss": 0.3289,
+      "step": 6756
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.598639455782313e-05,
+      "loss": 0.2276,
+      "step": 6757
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.59828141783029e-05,
+      "loss": 0.1805,
+      "step": 6758
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.597923379878267e-05,
+      "loss": 0.2499,
+      "step": 6759
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.597565341926244e-05,
+      "loss": 0.3334,
+      "step": 6760
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.597207303974221e-05,
+      "loss": 0.2123,
+      "step": 6761
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5968492660221984e-05,
+      "loss": 0.1535,
+      "step": 6762
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5964912280701754e-05,
+      "loss": 0.2381,
+      "step": 6763
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.596133190118153e-05,
+      "loss": 0.1886,
+      "step": 6764
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.59577515216613e-05,
+      "loss": 0.1177,
+      "step": 6765
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.595417114214107e-05,
+      "loss": 0.2364,
+      "step": 6766
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.595059076262084e-05,
+      "loss": 0.2135,
+      "step": 6767
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5947010383100613e-05,
+      "loss": 0.173,
+      "step": 6768
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.5943430003580383e-05,
+      "loss": 0.1832,
+      "step": 6769
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5939849624060153e-05,
+      "loss": 0.2303,
+      "step": 6770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5936269244539923e-05,
+      "loss": 0.1979,
+      "step": 6771
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5932688865019696e-05,
+      "loss": 0.1735,
+      "step": 6772
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5929108485499466e-05,
+      "loss": 0.2417,
+      "step": 6773
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5925528105979235e-05,
+      "loss": 0.2502,
+      "step": 6774
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5921947726459005e-05,
+      "loss": 0.186,
+      "step": 6775
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5918367346938778e-05,
+      "loss": 0.2503,
+      "step": 6776
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5914786967418548e-05,
+      "loss": 0.1508,
+      "step": 6777
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5911206587898318e-05,
+      "loss": 0.1808,
+      "step": 6778
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5907626208378088e-05,
+      "loss": 0.2395,
+      "step": 6779
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5904045828857857e-05,
+      "loss": 0.2215,
+      "step": 6780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.590046544933763e-05,
+      "loss": 0.3044,
+      "step": 6781
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.58968850698174e-05,
+      "loss": 0.2009,
+      "step": 6782
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.589330469029717e-05,
+      "loss": 0.2587,
+      "step": 6783
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.588972431077694e-05,
+      "loss": 0.2906,
+      "step": 6784
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5886143931256713e-05,
+      "loss": 0.2079,
+      "step": 6785
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5882563551736483e-05,
+      "loss": 0.2363,
+      "step": 6786
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5878983172216253e-05,
+      "loss": 0.1941,
+      "step": 6787
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.587540279269603e-05,
+      "loss": 0.2055,
+      "step": 6788
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.58718224131758e-05,
+      "loss": 0.2057,
+      "step": 6789
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5868242033655572e-05,
+      "loss": 0.1477,
+      "step": 6790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5864661654135342e-05,
+      "loss": 0.3353,
+      "step": 6791
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5861081274615112e-05,
+      "loss": 0.1505,
+      "step": 6792
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5857500895094882e-05,
+      "loss": 0.2124,
+      "step": 6793
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5853920515574655e-05,
+      "loss": 0.2218,
+      "step": 6794
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5850340136054425e-05,
+      "loss": 0.2987,
+      "step": 6795
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5846759756534195e-05,
+      "loss": 0.1694,
+      "step": 6796
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.5843179377013964e-05,
+      "loss": 0.1902,
+      "step": 6797
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5839598997493734e-05,
+      "loss": 0.2854,
+      "step": 6798
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5836018617973507e-05,
+      "loss": 0.4344,
+      "step": 6799
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5832438238453277e-05,
+      "loss": 0.1849,
+      "step": 6800
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5828857858933047e-05,
+      "loss": 0.3613,
+      "step": 6801
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5825277479412817e-05,
+      "loss": 0.3202,
+      "step": 6802
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.582169709989259e-05,
+      "loss": 0.1482,
+      "step": 6803
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.581811672037236e-05,
+      "loss": 0.348,
+      "step": 6804
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.581453634085213e-05,
+      "loss": 0.1213,
+      "step": 6805
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.58109559613319e-05,
+      "loss": 0.2914,
+      "step": 6806
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5807375581811673e-05,
+      "loss": 0.186,
+      "step": 6807
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5803795202291442e-05,
+      "loss": 0.1155,
+      "step": 6808
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5800214822771212e-05,
+      "loss": 0.3042,
+      "step": 6809
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5796634443250982e-05,
+      "loss": 0.142,
+      "step": 6810
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.579305406373076e-05,
+      "loss": 0.1312,
+      "step": 6811
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.578947368421053e-05,
+      "loss": 0.2029,
+      "step": 6812
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.57858933046903e-05,
+      "loss": 0.1129,
+      "step": 6813
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.578231292517007e-05,
+      "loss": 0.1024,
+      "step": 6814
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.577873254564984e-05,
+      "loss": 0.1992,
+      "step": 6815
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.577515216612961e-05,
+      "loss": 0.1425,
+      "step": 6816
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5771571786609384e-05,
+      "loss": 0.1095,
+      "step": 6817
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5767991407089154e-05,
+      "loss": 0.1889,
+      "step": 6818
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5764411027568924e-05,
+      "loss": 0.1613,
+      "step": 6819
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5760830648048694e-05,
+      "loss": 0.2507,
+      "step": 6820
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5757250268528467e-05,
+      "loss": 0.315,
+      "step": 6821
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5753669889008236e-05,
+      "loss": 0.2826,
+      "step": 6822
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5750089509488006e-05,
+      "loss": 0.3312,
+      "step": 6823
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.5746509129967776e-05,
+      "loss": 0.1775,
+      "step": 6824
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.574292875044755e-05,
+      "loss": 0.1952,
+      "step": 6825
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.573934837092732e-05,
+      "loss": 0.2953,
+      "step": 6826
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.573576799140709e-05,
+      "loss": 0.2489,
+      "step": 6827
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.573218761188686e-05,
+      "loss": 0.2182,
+      "step": 6828
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5728607232366632e-05,
+      "loss": 0.2454,
+      "step": 6829
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.57250268528464e-05,
+      "loss": 0.154,
+      "step": 6830
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.572144647332617e-05,
+      "loss": 0.2921,
+      "step": 6831
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.571786609380594e-05,
+      "loss": 0.1777,
+      "step": 6832
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5714285714285714e-05,
+      "loss": 0.2782,
+      "step": 6833
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5710705334765484e-05,
+      "loss": 0.4175,
+      "step": 6834
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.570712495524526e-05,
+      "loss": 0.2496,
+      "step": 6835
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.570354457572503e-05,
+      "loss": 0.2452,
+      "step": 6836
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.56999641962048e-05,
+      "loss": 0.1207,
+      "step": 6837
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.569638381668457e-05,
+      "loss": 0.2576,
+      "step": 6838
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5692803437164343e-05,
+      "loss": 0.1728,
+      "step": 6839
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5689223057644113e-05,
+      "loss": 0.0933,
+      "step": 6840
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5685642678123883e-05,
+      "loss": 0.2078,
+      "step": 6841
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5682062298603653e-05,
+      "loss": 0.2111,
+      "step": 6842
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5678481919083426e-05,
+      "loss": 0.2559,
+      "step": 6843
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5674901539563196e-05,
+      "loss": 0.1661,
+      "step": 6844
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5671321160042966e-05,
+      "loss": 0.3126,
+      "step": 6845
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5667740780522735e-05,
+      "loss": 0.1337,
+      "step": 6846
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.566416040100251e-05,
+      "loss": 0.19,
+      "step": 6847
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.566058002148228e-05,
+      "loss": 0.1825,
+      "step": 6848
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5656999641962048e-05,
+      "loss": 0.3018,
+      "step": 6849
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.5653419262441818e-05,
+      "loss": 0.2084,
+      "step": 6850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.564983888292159e-05,
+      "loss": 0.4849,
+      "step": 6851
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.564625850340136e-05,
+      "loss": 0.2395,
+      "step": 6852
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.564267812388113e-05,
+      "loss": 0.2135,
+      "step": 6853
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.56390977443609e-05,
+      "loss": 0.1838,
+      "step": 6854
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5635517364840674e-05,
+      "loss": 0.2527,
+      "step": 6855
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5631936985320444e-05,
+      "loss": 0.274,
+      "step": 6856
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5628356605800213e-05,
+      "loss": 0.2975,
+      "step": 6857
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.562477622627999e-05,
+      "loss": 0.2703,
+      "step": 6858
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.562119584675976e-05,
+      "loss": 0.2293,
+      "step": 6859
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.561761546723953e-05,
+      "loss": 0.0861,
+      "step": 6860
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5614035087719303e-05,
+      "loss": 0.2305,
+      "step": 6861
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5610454708199073e-05,
+      "loss": 0.2608,
+      "step": 6862
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5606874328678842e-05,
+      "loss": 0.1603,
+      "step": 6863
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5603293949158612e-05,
+      "loss": 0.1955,
+      "step": 6864
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5599713569638385e-05,
+      "loss": 0.2581,
+      "step": 6865
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5596133190118155e-05,
+      "loss": 0.1382,
+      "step": 6866
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5592552810597925e-05,
+      "loss": 0.3296,
+      "step": 6867
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5588972431077695e-05,
+      "loss": 0.1784,
+      "step": 6868
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5585392051557468e-05,
+      "loss": 0.1872,
+      "step": 6869
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5581811672037238e-05,
+      "loss": 0.1464,
+      "step": 6870
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5578231292517007e-05,
+      "loss": 0.2412,
+      "step": 6871
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5574650912996777e-05,
+      "loss": 0.3786,
+      "step": 6872
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.557107053347655e-05,
+      "loss": 0.3371,
+      "step": 6873
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.556749015395632e-05,
+      "loss": 0.2355,
+      "step": 6874
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.556390977443609e-05,
+      "loss": 0.2424,
+      "step": 6875
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.556032939491586e-05,
+      "loss": 0.2418,
+      "step": 6876
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.555674901539563e-05,
+      "loss": 0.2061,
+      "step": 6877
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5553168635875403e-05,
+      "loss": 0.3939,
+      "step": 6878
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5549588256355173e-05,
+      "loss": 0.1074,
+      "step": 6879
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.5546007876834942e-05,
+      "loss": 0.4285,
+      "step": 6880
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.554242749731472e-05,
+      "loss": 0.2658,
+      "step": 6881
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.553884711779449e-05,
+      "loss": 0.1778,
+      "step": 6882
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5535266738274262e-05,
+      "loss": 0.1388,
+      "step": 6883
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5531686358754032e-05,
+      "loss": 0.1916,
+      "step": 6884
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.55281059792338e-05,
+      "loss": 0.188,
+      "step": 6885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.552452559971357e-05,
+      "loss": 0.285,
+      "step": 6886
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5520945220193345e-05,
+      "loss": 0.3197,
+      "step": 6887
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5517364840673114e-05,
+      "loss": 0.3926,
+      "step": 6888
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5513784461152884e-05,
+      "loss": 0.2048,
+      "step": 6889
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5510204081632654e-05,
+      "loss": 0.1462,
+      "step": 6890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5506623702112424e-05,
+      "loss": 0.2465,
+      "step": 6891
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5503043322592197e-05,
+      "loss": 0.2633,
+      "step": 6892
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5499462943071967e-05,
+      "loss": 0.3564,
+      "step": 6893
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5495882563551737e-05,
+      "loss": 0.2478,
+      "step": 6894
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5492302184031506e-05,
+      "loss": 0.1441,
+      "step": 6895
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.548872180451128e-05,
+      "loss": 0.1614,
+      "step": 6896
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.548514142499105e-05,
+      "loss": 0.2885,
+      "step": 6897
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.548156104547082e-05,
+      "loss": 0.3908,
+      "step": 6898
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.547798066595059e-05,
+      "loss": 0.2362,
+      "step": 6899
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5474400286430362e-05,
+      "loss": 0.209,
+      "step": 6900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5470819906910132e-05,
+      "loss": 0.1944,
+      "step": 6901
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5467239527389902e-05,
+      "loss": 0.3045,
+      "step": 6902
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.546365914786967e-05,
+      "loss": 0.2396,
+      "step": 6903
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.5460078768349445e-05,
+      "loss": 0.2318,
+      "step": 6904
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.545649838882922e-05,
+      "loss": 0.3329,
+      "step": 6905
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.545291800930899e-05,
+      "loss": 0.2178,
+      "step": 6906
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.544933762978876e-05,
+      "loss": 0.2533,
+      "step": 6907
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.544575725026853e-05,
+      "loss": 0.2373,
+      "step": 6908
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.54421768707483e-05,
+      "loss": 0.2548,
+      "step": 6909
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5438596491228074e-05,
+      "loss": 0.2943,
+      "step": 6910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5435016111707844e-05,
+      "loss": 0.1798,
+      "step": 6911
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5431435732187613e-05,
+      "loss": 0.1555,
+      "step": 6912
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5427855352667383e-05,
+      "loss": 0.336,
+      "step": 6913
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5424274973147156e-05,
+      "loss": 0.2112,
+      "step": 6914
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5420694593626926e-05,
+      "loss": 0.2839,
+      "step": 6915
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5417114214106696e-05,
+      "loss": 0.2033,
+      "step": 6916
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5413533834586466e-05,
+      "loss": 0.2307,
+      "step": 6917
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.540995345506624e-05,
+      "loss": 0.2307,
+      "step": 6918
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.540637307554601e-05,
+      "loss": 0.2152,
+      "step": 6919
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.540279269602578e-05,
+      "loss": 0.2183,
+      "step": 6920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5399212316505548e-05,
+      "loss": 0.1724,
+      "step": 6921
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.539563193698532e-05,
+      "loss": 0.1415,
+      "step": 6922
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.539205155746509e-05,
+      "loss": 0.3353,
+      "step": 6923
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.538847117794486e-05,
+      "loss": 0.2796,
+      "step": 6924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.538489079842463e-05,
+      "loss": 0.1702,
+      "step": 6925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5381310418904404e-05,
+      "loss": 0.2529,
+      "step": 6926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5377730039384174e-05,
+      "loss": 0.2498,
+      "step": 6927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.537414965986395e-05,
+      "loss": 0.172,
+      "step": 6928
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.537056928034372e-05,
+      "loss": 0.3251,
+      "step": 6929
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.536698890082349e-05,
+      "loss": 0.199,
+      "step": 6930
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.536340852130326e-05,
+      "loss": 0.281,
+      "step": 6931
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5359828141783033e-05,
+      "loss": 0.1634,
+      "step": 6932
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5356247762262803e-05,
+      "loss": 0.1233,
+      "step": 6933
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5352667382742573e-05,
+      "loss": 0.1759,
+      "step": 6934
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5349087003222342e-05,
+      "loss": 0.2458,
+      "step": 6935
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5345506623702116e-05,
+      "loss": 0.2403,
+      "step": 6936
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.5341926244181885e-05,
+      "loss": 0.2802,
+      "step": 6937
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5338345864661655e-05,
+      "loss": 0.185,
+      "step": 6938
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5334765485141425e-05,
+      "loss": 0.1883,
+      "step": 6939
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5331185105621198e-05,
+      "loss": 0.1567,
+      "step": 6940
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5327604726100968e-05,
+      "loss": 0.3301,
+      "step": 6941
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5324024346580738e-05,
+      "loss": 0.3075,
+      "step": 6942
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5320443967060508e-05,
+      "loss": 0.1622,
+      "step": 6943
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.531686358754028e-05,
+      "loss": 0.1634,
+      "step": 6944
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.531328320802005e-05,
+      "loss": 0.1773,
+      "step": 6945
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.530970282849982e-05,
+      "loss": 0.1974,
+      "step": 6946
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.530612244897959e-05,
+      "loss": 0.346,
+      "step": 6947
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5302542069459363e-05,
+      "loss": 0.2426,
+      "step": 6948
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5298961689939133e-05,
+      "loss": 0.3063,
+      "step": 6949
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5295381310418903e-05,
+      "loss": 0.259,
+      "step": 6950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.529180093089868e-05,
+      "loss": 0.2672,
+      "step": 6951
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.528822055137845e-05,
+      "loss": 0.187,
+      "step": 6952
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.528464017185822e-05,
+      "loss": 0.2632,
+      "step": 6953
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5281059792337992e-05,
+      "loss": 0.1658,
+      "step": 6954
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5277479412817762e-05,
+      "loss": 0.2453,
+      "step": 6955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5273899033297532e-05,
+      "loss": 0.1294,
+      "step": 6956
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5270318653777302e-05,
+      "loss": 0.2816,
+      "step": 6957
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5266738274257075e-05,
+      "loss": 0.2757,
+      "step": 6958
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5263157894736845e-05,
+      "loss": 0.142,
+      "step": 6959
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5259577515216615e-05,
+      "loss": 0.1676,
+      "step": 6960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5255997135696384e-05,
+      "loss": 0.1906,
+      "step": 6961
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5252416756176157e-05,
+      "loss": 0.1859,
+      "step": 6962
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5248836376655927e-05,
+      "loss": 0.3292,
+      "step": 6963
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5245255997135697e-05,
+      "loss": 0.2247,
+      "step": 6964
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.5241675617615467e-05,
+      "loss": 0.0784,
+      "step": 6965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.523809523809524e-05,
+      "loss": 0.3138,
+      "step": 6966
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.523451485857501e-05,
+      "loss": 0.1929,
+      "step": 6967
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.523093447905478e-05,
+      "loss": 0.1357,
+      "step": 6968
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.522735409953455e-05,
+      "loss": 0.1476,
+      "step": 6969
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.522377372001432e-05,
+      "loss": 0.103,
+      "step": 6970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5220193340494092e-05,
+      "loss": 0.3691,
+      "step": 6971
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5216612960973862e-05,
+      "loss": 0.1784,
+      "step": 6972
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5213032581453632e-05,
+      "loss": 0.4072,
+      "step": 6973
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5209452201933402e-05,
+      "loss": 0.1781,
+      "step": 6974
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.520587182241318e-05,
+      "loss": 0.0883,
+      "step": 6975
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.520229144289295e-05,
+      "loss": 0.1986,
+      "step": 6976
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.519871106337272e-05,
+      "loss": 0.1942,
+      "step": 6977
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.519513068385249e-05,
+      "loss": 0.2329,
+      "step": 6978
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.519155030433226e-05,
+      "loss": 0.2464,
+      "step": 6979
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5187969924812034e-05,
+      "loss": 0.2861,
+      "step": 6980
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5184389545291804e-05,
+      "loss": 0.1772,
+      "step": 6981
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5180809165771574e-05,
+      "loss": 0.3043,
+      "step": 6982
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5177228786251344e-05,
+      "loss": 0.1552,
+      "step": 6983
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5173648406731117e-05,
+      "loss": 0.3517,
+      "step": 6984
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5170068027210887e-05,
+      "loss": 0.2347,
+      "step": 6985
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5166487647690656e-05,
+      "loss": 0.2222,
+      "step": 6986
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5162907268170426e-05,
+      "loss": 0.2276,
+      "step": 6987
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5159326888650196e-05,
+      "loss": 0.3298,
+      "step": 6988
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.515574650912997e-05,
+      "loss": 0.2389,
+      "step": 6989
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.515216612960974e-05,
+      "loss": 0.2574,
+      "step": 6990
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.514858575008951e-05,
+      "loss": 0.2148,
+      "step": 6991
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.514500537056928e-05,
+      "loss": 0.2963,
+      "step": 6992
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.5141424991049052e-05,
+      "loss": 0.2052,
+      "step": 6993
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.513784461152882e-05,
+      "loss": 0.2196,
+      "step": 6994
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.513426423200859e-05,
+      "loss": 0.3057,
+      "step": 6995
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.513068385248836e-05,
+      "loss": 0.1802,
+      "step": 6996
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5127103472968134e-05,
+      "loss": 0.1391,
+      "step": 6997
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.512352309344791e-05,
+      "loss": 0.2937,
+      "step": 6998
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.511994271392768e-05,
+      "loss": 0.3408,
+      "step": 6999
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.511636233440745e-05,
+      "loss": 0.2869,
+      "step": 7000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.511278195488722e-05,
+      "loss": 0.3308,
+      "step": 7001
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.510920157536699e-05,
+      "loss": 0.1679,
+      "step": 7002
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5105621195846763e-05,
+      "loss": 0.1613,
+      "step": 7003
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5102040816326533e-05,
+      "loss": 0.2176,
+      "step": 7004
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5098460436806303e-05,
+      "loss": 0.2416,
+      "step": 7005
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5094880057286073e-05,
+      "loss": 0.2427,
+      "step": 7006
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5091299677765846e-05,
+      "loss": 0.3336,
+      "step": 7007
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5087719298245616e-05,
+      "loss": 0.2415,
+      "step": 7008
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5084138918725386e-05,
+      "loss": 0.2471,
+      "step": 7009
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5080558539205155e-05,
+      "loss": 0.3131,
+      "step": 7010
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.507697815968493e-05,
+      "loss": 0.2595,
+      "step": 7011
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5073397780164698e-05,
+      "loss": 0.2945,
+      "step": 7012
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5069817400644468e-05,
+      "loss": 0.1835,
+      "step": 7013
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5066237021124238e-05,
+      "loss": 0.3095,
+      "step": 7014
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.506265664160401e-05,
+      "loss": 0.1232,
+      "step": 7015
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.505907626208378e-05,
+      "loss": 0.2192,
+      "step": 7016
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.505549588256355e-05,
+      "loss": 0.2318,
+      "step": 7017
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.505191550304332e-05,
+      "loss": 0.3017,
+      "step": 7018
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5048335123523094e-05,
+      "loss": 0.2256,
+      "step": 7019
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.5044754744002863e-05,
+      "loss": 0.3089,
+      "step": 7020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.504117436448264e-05,
+      "loss": 0.1518,
+      "step": 7021
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.503759398496241e-05,
+      "loss": 0.2237,
+      "step": 7022
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.503401360544218e-05,
+      "loss": 0.2779,
+      "step": 7023
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.503043322592195e-05,
+      "loss": 0.2823,
+      "step": 7024
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5026852846401723e-05,
+      "loss": 0.1927,
+      "step": 7025
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5023272466881492e-05,
+      "loss": 0.2332,
+      "step": 7026
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5019692087361262e-05,
+      "loss": 0.2822,
+      "step": 7027
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5016111707841032e-05,
+      "loss": 0.2985,
+      "step": 7028
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5012531328320805e-05,
+      "loss": 0.1677,
+      "step": 7029
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5008950948800575e-05,
+      "loss": 0.2321,
+      "step": 7030
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5005370569280345e-05,
+      "loss": 0.1705,
+      "step": 7031
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5001790189760115e-05,
+      "loss": 0.2589,
+      "step": 7032
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4998209810239888e-05,
+      "loss": 0.3111,
+      "step": 7033
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4994629430719658e-05,
+      "loss": 0.1941,
+      "step": 7034
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4991049051199427e-05,
+      "loss": 0.1578,
+      "step": 7035
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4987468671679197e-05,
+      "loss": 0.2685,
+      "step": 7036
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.498388829215897e-05,
+      "loss": 0.317,
+      "step": 7037
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.498030791263874e-05,
+      "loss": 0.1651,
+      "step": 7038
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4976727533118513e-05,
+      "loss": 0.156,
+      "step": 7039
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4973147153598283e-05,
+      "loss": 0.2207,
+      "step": 7040
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4969566774078053e-05,
+      "loss": 0.3086,
+      "step": 7041
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4965986394557826e-05,
+      "loss": 0.144,
+      "step": 7042
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4962406015037596e-05,
+      "loss": 0.2601,
+      "step": 7043
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4958825635517366e-05,
+      "loss": 0.1791,
+      "step": 7044
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4955245255997136e-05,
+      "loss": 0.1643,
+      "step": 7045
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.495166487647691e-05,
+      "loss": 0.1961,
+      "step": 7046
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.494808449695668e-05,
+      "loss": 0.2408,
+      "step": 7047
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.494450411743645e-05,
+      "loss": 0.3768,
+      "step": 7048
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.4940923737916218e-05,
+      "loss": 0.3312,
+      "step": 7049
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.493734335839599e-05,
+      "loss": 0.2946,
+      "step": 7050
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4933762978875765e-05,
+      "loss": 0.211,
+      "step": 7051
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4930182599355534e-05,
+      "loss": 0.2039,
+      "step": 7052
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4926602219835304e-05,
+      "loss": 0.2433,
+      "step": 7053
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4923021840315074e-05,
+      "loss": 0.2119,
+      "step": 7054
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4919441460794847e-05,
+      "loss": 0.2663,
+      "step": 7055
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4915861081274617e-05,
+      "loss": 0.1467,
+      "step": 7056
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4912280701754387e-05,
+      "loss": 0.2129,
+      "step": 7057
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4908700322234156e-05,
+      "loss": 0.1078,
+      "step": 7058
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.490511994271393e-05,
+      "loss": 0.1282,
+      "step": 7059
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.49015395631937e-05,
+      "loss": 0.1849,
+      "step": 7060
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.489795918367347e-05,
+      "loss": 0.2651,
+      "step": 7061
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4894378804153242e-05,
+      "loss": 0.2158,
+      "step": 7062
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4890798424633012e-05,
+      "loss": 0.1185,
+      "step": 7063
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4887218045112785e-05,
+      "loss": 0.3258,
+      "step": 7064
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4883637665592555e-05,
+      "loss": 0.3019,
+      "step": 7065
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4880057286072325e-05,
+      "loss": 0.2774,
+      "step": 7066
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4876476906552095e-05,
+      "loss": 0.2098,
+      "step": 7067
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4872896527031868e-05,
+      "loss": 0.134,
+      "step": 7068
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4869316147511638e-05,
+      "loss": 0.2516,
+      "step": 7069
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4865735767991408e-05,
+      "loss": 0.2846,
+      "step": 7070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4862155388471177e-05,
+      "loss": 0.3213,
+      "step": 7071
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4858575008950947e-05,
+      "loss": 0.1139,
+      "step": 7072
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.485499462943072e-05,
+      "loss": 0.1815,
+      "step": 7073
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4851414249910494e-05,
+      "loss": 0.2298,
+      "step": 7074
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4847833870390263e-05,
+      "loss": 0.1833,
+      "step": 7075
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4844253490870033e-05,
+      "loss": 0.2427,
+      "step": 7076
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.4840673111349806e-05,
+      "loss": 0.2567,
+      "step": 7077
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4837092731829576e-05,
+      "loss": 0.2428,
+      "step": 7078
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4833512352309346e-05,
+      "loss": 0.213,
+      "step": 7079
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4829931972789116e-05,
+      "loss": 0.2743,
+      "step": 7080
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4826351593268886e-05,
+      "loss": 0.2314,
+      "step": 7081
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.482277121374866e-05,
+      "loss": 0.2748,
+      "step": 7082
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.481919083422843e-05,
+      "loss": 0.2022,
+      "step": 7083
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.48156104547082e-05,
+      "loss": 0.3122,
+      "step": 7084
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4812030075187968e-05,
+      "loss": 0.1651,
+      "step": 7085
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4808449695667745e-05,
+      "loss": 0.1565,
+      "step": 7086
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4804869316147515e-05,
+      "loss": 0.2657,
+      "step": 7087
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4801288936627284e-05,
+      "loss": 0.2845,
+      "step": 7088
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4797708557107054e-05,
+      "loss": 0.2985,
+      "step": 7089
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4794128177586824e-05,
+      "loss": 0.1922,
+      "step": 7090
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4790547798066597e-05,
+      "loss": 0.2683,
+      "step": 7091
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4786967418546367e-05,
+      "loss": 0.2864,
+      "step": 7092
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4783387039026137e-05,
+      "loss": 0.2627,
+      "step": 7093
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4779806659505907e-05,
+      "loss": 0.3872,
+      "step": 7094
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.477622627998568e-05,
+      "loss": 0.2171,
+      "step": 7095
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.477264590046545e-05,
+      "loss": 0.2797,
+      "step": 7096
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4769065520945223e-05,
+      "loss": 0.2182,
+      "step": 7097
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4765485141424993e-05,
+      "loss": 0.3488,
+      "step": 7098
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4761904761904762e-05,
+      "loss": 0.2211,
+      "step": 7099
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4758324382384536e-05,
+      "loss": 0.3232,
+      "step": 7100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4754744002864305e-05,
+      "loss": 0.2051,
+      "step": 7101
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4751163623344075e-05,
+      "loss": 0.427,
+      "step": 7102
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4747583243823845e-05,
+      "loss": 0.1128,
+      "step": 7103
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4744002864303618e-05,
+      "loss": 0.2509,
+      "step": 7104
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.4740422484783388e-05,
+      "loss": 0.3003,
+      "step": 7105
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4736842105263158e-05,
+      "loss": 0.2152,
+      "step": 7106
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4733261725742927e-05,
+      "loss": 0.2264,
+      "step": 7107
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.47296813462227e-05,
+      "loss": 0.1634,
+      "step": 7108
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4726100966702474e-05,
+      "loss": 0.1481,
+      "step": 7109
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4722520587182244e-05,
+      "loss": 0.2408,
+      "step": 7110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4718940207662013e-05,
+      "loss": 0.2725,
+      "step": 7111
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4715359828141783e-05,
+      "loss": 0.1421,
+      "step": 7112
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4711779448621556e-05,
+      "loss": 0.3694,
+      "step": 7113
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4708199069101326e-05,
+      "loss": 0.1866,
+      "step": 7114
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4704618689581096e-05,
+      "loss": 0.3124,
+      "step": 7115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4701038310060866e-05,
+      "loss": 0.2078,
+      "step": 7116
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.469745793054064e-05,
+      "loss": 0.176,
+      "step": 7117
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.469387755102041e-05,
+      "loss": 0.1028,
+      "step": 7118
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.469029717150018e-05,
+      "loss": 0.3414,
+      "step": 7119
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.468671679197995e-05,
+      "loss": 0.2203,
+      "step": 7120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.468313641245972e-05,
+      "loss": 0.2601,
+      "step": 7121
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4679556032939495e-05,
+      "loss": 0.2679,
+      "step": 7122
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4675975653419265e-05,
+      "loss": 0.2503,
+      "step": 7123
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4672395273899034e-05,
+      "loss": 0.2319,
+      "step": 7124
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4668814894378804e-05,
+      "loss": 0.2071,
+      "step": 7125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4665234514858577e-05,
+      "loss": 0.252,
+      "step": 7126
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4661654135338347e-05,
+      "loss": 0.351,
+      "step": 7127
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4658073755818117e-05,
+      "loss": 0.1274,
+      "step": 7128
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4654493376297887e-05,
+      "loss": 0.2491,
+      "step": 7129
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.465091299677766e-05,
+      "loss": 0.2269,
+      "step": 7130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.464733261725743e-05,
+      "loss": 0.2126,
+      "step": 7131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4643752237737203e-05,
+      "loss": 0.1549,
+      "step": 7132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.4640171858216973e-05,
+      "loss": 0.2602,
+      "step": 7133
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4636591478696743e-05,
+      "loss": 0.2365,
+      "step": 7134
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4633011099176516e-05,
+      "loss": 0.1811,
+      "step": 7135
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4629430719656286e-05,
+      "loss": 0.2029,
+      "step": 7136
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4625850340136055e-05,
+      "loss": 0.1704,
+      "step": 7137
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4622269960615825e-05,
+      "loss": 0.1855,
+      "step": 7138
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.46186895810956e-05,
+      "loss": 0.109,
+      "step": 7139
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4615109201575368e-05,
+      "loss": 0.15,
+      "step": 7140
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4611528822055138e-05,
+      "loss": 0.125,
+      "step": 7141
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4607948442534908e-05,
+      "loss": 0.1582,
+      "step": 7142
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.460436806301468e-05,
+      "loss": 0.2167,
+      "step": 7143
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4600787683494454e-05,
+      "loss": 0.3405,
+      "step": 7144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4597207303974224e-05,
+      "loss": 0.255,
+      "step": 7145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4593626924453994e-05,
+      "loss": 0.2338,
+      "step": 7146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4590046544933764e-05,
+      "loss": 0.1682,
+      "step": 7147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4586466165413537e-05,
+      "loss": 0.0629,
+      "step": 7148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4582885785893306e-05,
+      "loss": 0.1788,
+      "step": 7149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4579305406373076e-05,
+      "loss": 0.319,
+      "step": 7150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4575725026852846e-05,
+      "loss": 0.2915,
+      "step": 7151
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.457214464733262e-05,
+      "loss": 0.3038,
+      "step": 7152
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.456856426781239e-05,
+      "loss": 0.2446,
+      "step": 7153
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.456498388829216e-05,
+      "loss": 0.2118,
+      "step": 7154
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.456140350877193e-05,
+      "loss": 0.2377,
+      "step": 7155
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4557823129251702e-05,
+      "loss": 0.1487,
+      "step": 7156
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4554242749731475e-05,
+      "loss": 0.2008,
+      "step": 7157
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4550662370211245e-05,
+      "loss": 0.1423,
+      "step": 7158
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4547081990691015e-05,
+      "loss": 0.279,
+      "step": 7159
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4543501611170784e-05,
+      "loss": 0.1484,
+      "step": 7160
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.4539921231650558e-05,
+      "loss": 0.2973,
+      "step": 7161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4536340852130327e-05,
+      "loss": 0.2176,
+      "step": 7162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4532760472610097e-05,
+      "loss": 0.274,
+      "step": 7163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4529180093089867e-05,
+      "loss": 0.2693,
+      "step": 7164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.452559971356964e-05,
+      "loss": 0.3292,
+      "step": 7165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.452201933404941e-05,
+      "loss": 0.4284,
+      "step": 7166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4518438954529183e-05,
+      "loss": 0.2484,
+      "step": 7167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4514858575008953e-05,
+      "loss": 0.254,
+      "step": 7168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4511278195488723e-05,
+      "loss": 0.1489,
+      "step": 7169
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4507697815968496e-05,
+      "loss": 0.1327,
+      "step": 7170
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4504117436448266e-05,
+      "loss": 0.1446,
+      "step": 7171
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4500537056928036e-05,
+      "loss": 0.2539,
+      "step": 7172
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4496956677407805e-05,
+      "loss": 0.2899,
+      "step": 7173
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.449337629788758e-05,
+      "loss": 0.2204,
+      "step": 7174
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.448979591836735e-05,
+      "loss": 0.2177,
+      "step": 7175
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4486215538847118e-05,
+      "loss": 0.1855,
+      "step": 7176
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4482635159326888e-05,
+      "loss": 0.1891,
+      "step": 7177
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4479054779806658e-05,
+      "loss": 0.2077,
+      "step": 7178
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4475474400286434e-05,
+      "loss": 0.1146,
+      "step": 7179
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4471894020766204e-05,
+      "loss": 0.1287,
+      "step": 7180
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4468313641245974e-05,
+      "loss": 0.164,
+      "step": 7181
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4464733261725744e-05,
+      "loss": 0.1594,
+      "step": 7182
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4461152882205517e-05,
+      "loss": 0.2643,
+      "step": 7183
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4457572502685287e-05,
+      "loss": 0.1426,
+      "step": 7184
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4453992123165057e-05,
+      "loss": 0.228,
+      "step": 7185
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4450411743644826e-05,
+      "loss": 0.1845,
+      "step": 7186
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.4446831364124596e-05,
+      "loss": 0.2055,
+      "step": 7187
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.444325098460437e-05,
+      "loss": 0.3278,
+      "step": 7188
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.443967060508414e-05,
+      "loss": 0.455,
+      "step": 7189
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.443609022556391e-05,
+      "loss": 0.1585,
+      "step": 7190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4432509846043682e-05,
+      "loss": 0.1422,
+      "step": 7191
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4428929466523452e-05,
+      "loss": 0.321,
+      "step": 7192
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4425349087003225e-05,
+      "loss": 0.1122,
+      "step": 7193
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4421768707482995e-05,
+      "loss": 0.1796,
+      "step": 7194
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4418188327962765e-05,
+      "loss": 0.1229,
+      "step": 7195
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4414607948442535e-05,
+      "loss": 0.2326,
+      "step": 7196
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4411027568922308e-05,
+      "loss": 0.1893,
+      "step": 7197
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4407447189402077e-05,
+      "loss": 0.2904,
+      "step": 7198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4403866809881847e-05,
+      "loss": 0.1894,
+      "step": 7199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4400286430361617e-05,
+      "loss": 0.225,
+      "step": 7200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.439670605084139e-05,
+      "loss": 0.1686,
+      "step": 7201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4393125671321163e-05,
+      "loss": 0.2331,
+      "step": 7202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4389545291800933e-05,
+      "loss": 0.3046,
+      "step": 7203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4385964912280703e-05,
+      "loss": 0.2605,
+      "step": 7204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4382384532760473e-05,
+      "loss": 0.3171,
+      "step": 7205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4378804153240246e-05,
+      "loss": 0.2367,
+      "step": 7206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4375223773720016e-05,
+      "loss": 0.2661,
+      "step": 7207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4371643394199786e-05,
+      "loss": 0.3211,
+      "step": 7208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4368063014679555e-05,
+      "loss": 0.2302,
+      "step": 7209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.436448263515933e-05,
+      "loss": 0.1783,
+      "step": 7210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.43609022556391e-05,
+      "loss": 0.189,
+      "step": 7211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4357321876118868e-05,
+      "loss": 0.2089,
+      "step": 7212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4353741496598638e-05,
+      "loss": 0.1778,
+      "step": 7213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.435016111707841e-05,
+      "loss": 0.2606,
+      "step": 7214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4346580737558184e-05,
+      "loss": 0.2421,
+      "step": 7215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4343000358037954e-05,
+      "loss": 0.2434,
+      "step": 7216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.4339419978517724e-05,
+      "loss": 0.1519,
+      "step": 7217
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4335839598997494e-05,
+      "loss": 0.328,
+      "step": 7218
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4332259219477267e-05,
+      "loss": 0.1972,
+      "step": 7219
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4328678839957037e-05,
+      "loss": 0.1484,
+      "step": 7220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4325098460436807e-05,
+      "loss": 0.1698,
+      "step": 7221
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4321518080916576e-05,
+      "loss": 0.2298,
+      "step": 7222
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.431793770139635e-05,
+      "loss": 0.176,
+      "step": 7223
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.431435732187612e-05,
+      "loss": 0.1582,
+      "step": 7224
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.431077694235589e-05,
+      "loss": 0.1893,
+      "step": 7225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4307196562835662e-05,
+      "loss": 0.1259,
+      "step": 7226
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4303616183315432e-05,
+      "loss": 0.3381,
+      "step": 7227
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4300035803795205e-05,
+      "loss": 0.1915,
+      "step": 7228
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4296455424274975e-05,
+      "loss": 0.2713,
+      "step": 7229
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4292875044754745e-05,
+      "loss": 0.36,
+      "step": 7230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4289294665234515e-05,
+      "loss": 0.271,
+      "step": 7231
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4285714285714288e-05,
+      "loss": 0.3622,
+      "step": 7232
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4282133906194058e-05,
+      "loss": 0.2121,
+      "step": 7233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4278553526673828e-05,
+      "loss": 0.1632,
+      "step": 7234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4274973147153597e-05,
+      "loss": 0.1582,
+      "step": 7235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.427139276763337e-05,
+      "loss": 0.278,
+      "step": 7236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4267812388113144e-05,
+      "loss": 0.3092,
+      "step": 7237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4264232008592914e-05,
+      "loss": 0.2554,
+      "step": 7238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4260651629072683e-05,
+      "loss": 0.2132,
+      "step": 7239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4257071249552453e-05,
+      "loss": 0.1307,
+      "step": 7240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4253490870032226e-05,
+      "loss": 0.1879,
+      "step": 7241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4249910490511996e-05,
+      "loss": 0.1503,
+      "step": 7242
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4246330110991766e-05,
+      "loss": 0.1329,
+      "step": 7243
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4242749731471536e-05,
+      "loss": 0.1944,
+      "step": 7244
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.423916935195131e-05,
+      "loss": 0.1494,
+      "step": 7245
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.423558897243108e-05,
+      "loss": 0.3333,
+      "step": 7246
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.423200859291085e-05,
+      "loss": 0.186,
+      "step": 7247
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4228428213390618e-05,
+      "loss": 0.1537,
+      "step": 7248
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.422484783387039e-05,
+      "loss": 0.2831,
+      "step": 7249
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4221267454350165e-05,
+      "loss": 0.1387,
+      "step": 7250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4217687074829934e-05,
+      "loss": 0.2666,
+      "step": 7251
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4214106695309704e-05,
+      "loss": 0.2721,
+      "step": 7252
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4210526315789474e-05,
+      "loss": 0.0827,
+      "step": 7253
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4206945936269247e-05,
+      "loss": 0.1832,
+      "step": 7254
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4203365556749017e-05,
+      "loss": 0.2302,
+      "step": 7255
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4199785177228787e-05,
+      "loss": 0.1087,
+      "step": 7256
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4196204797708557e-05,
+      "loss": 0.21,
+      "step": 7257
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.419262441818833e-05,
+      "loss": 0.1232,
+      "step": 7258
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.41890440386681e-05,
+      "loss": 0.187,
+      "step": 7259
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.418546365914787e-05,
+      "loss": 0.2822,
+      "step": 7260
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4181883279627643e-05,
+      "loss": 0.2202,
+      "step": 7261
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4178302900107412e-05,
+      "loss": 0.306,
+      "step": 7262
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4174722520587186e-05,
+      "loss": 0.1512,
+      "step": 7263
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4171142141066955e-05,
+      "loss": 0.1617,
+      "step": 7264
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4167561761546725e-05,
+      "loss": 0.3032,
+      "step": 7265
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4163981382026495e-05,
+      "loss": 0.3045,
+      "step": 7266
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4160401002506268e-05,
+      "loss": 0.3147,
+      "step": 7267
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4156820622986038e-05,
+      "loss": 0.2266,
+      "step": 7268
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4153240243465808e-05,
+      "loss": 0.3235,
+      "step": 7269
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4149659863945578e-05,
+      "loss": 0.4047,
+      "step": 7270
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4146079484425347e-05,
+      "loss": 0.156,
+      "step": 7271
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4142499104905124e-05,
+      "loss": 0.1732,
+      "step": 7272
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4138918725384894e-05,
+      "loss": 0.1916,
+      "step": 7273
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4135338345864664e-05,
+      "loss": 0.2358,
+      "step": 7274
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4131757966344433e-05,
+      "loss": 0.3486,
+      "step": 7275
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4128177586824207e-05,
+      "loss": 0.3224,
+      "step": 7276
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4124597207303976e-05,
+      "loss": 0.173,
+      "step": 7277
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4121016827783746e-05,
+      "loss": 0.312,
+      "step": 7278
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4117436448263516e-05,
+      "loss": 0.264,
+      "step": 7279
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4113856068743286e-05,
+      "loss": 0.2068,
+      "step": 7280
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.411027568922306e-05,
+      "loss": 0.3588,
+      "step": 7281
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.410669530970283e-05,
+      "loss": 0.2103,
+      "step": 7282
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.41031149301826e-05,
+      "loss": 0.1886,
+      "step": 7283
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4099534550662372e-05,
+      "loss": 0.1721,
+      "step": 7284
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4095954171142145e-05,
+      "loss": 0.2903,
+      "step": 7285
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4092373791621915e-05,
+      "loss": 0.1694,
+      "step": 7286
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4088793412101685e-05,
+      "loss": 0.1703,
+      "step": 7287
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4085213032581454e-05,
+      "loss": 0.3488,
+      "step": 7288
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4081632653061224e-05,
+      "loss": 0.2606,
+      "step": 7289
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4078052273540997e-05,
+      "loss": 0.14,
+      "step": 7290
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4074471894020767e-05,
+      "loss": 0.2566,
+      "step": 7291
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4070891514500537e-05,
+      "loss": 0.3271,
+      "step": 7292
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4067311134980307e-05,
+      "loss": 0.1837,
+      "step": 7293
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.406373075546008e-05,
+      "loss": 0.2216,
+      "step": 7294
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.406015037593985e-05,
+      "loss": 0.248,
+      "step": 7295
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4056569996419623e-05,
+      "loss": 0.1504,
+      "step": 7296
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4052989616899393e-05,
+      "loss": 0.1555,
+      "step": 7297
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4049409237379162e-05,
+      "loss": 0.1863,
+      "step": 7298
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4045828857858936e-05,
+      "loss": 0.209,
+      "step": 7299
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4042248478338705e-05,
+      "loss": 0.2921,
+      "step": 7300
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4038668098818475e-05,
+      "loss": 0.1666,
+      "step": 7301
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4035087719298245e-05,
+      "loss": 0.2882,
+      "step": 7302
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4031507339778018e-05,
+      "loss": 0.2588,
+      "step": 7303
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4027926960257788e-05,
+      "loss": 0.1887,
+      "step": 7304
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4024346580737558e-05,
+      "loss": 0.2638,
+      "step": 7305
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4020766201217328e-05,
+      "loss": 0.3614,
+      "step": 7306
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.40171858216971e-05,
+      "loss": 0.1946,
+      "step": 7307
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4013605442176874e-05,
+      "loss": 0.2025,
+      "step": 7308
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4010025062656644e-05,
+      "loss": 0.2896,
+      "step": 7309
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4006444683136414e-05,
+      "loss": 0.2395,
+      "step": 7310
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4002864303616183e-05,
+      "loss": 0.1128,
+      "step": 7311
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3999283924095957e-05,
+      "loss": 0.2287,
+      "step": 7312
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3995703544575726e-05,
+      "loss": 0.1838,
+      "step": 7313
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3992123165055496e-05,
+      "loss": 0.1587,
+      "step": 7314
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3988542785535266e-05,
+      "loss": 0.2225,
+      "step": 7315
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.398496240601504e-05,
+      "loss": 0.2107,
+      "step": 7316
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.398138202649481e-05,
+      "loss": 0.2618,
+      "step": 7317
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.397780164697458e-05,
+      "loss": 0.0986,
+      "step": 7318
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3974221267454352e-05,
+      "loss": 0.1727,
+      "step": 7319
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3970640887934122e-05,
+      "loss": 0.1956,
+      "step": 7320
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3967060508413895e-05,
+      "loss": 0.2576,
+      "step": 7321
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3963480128893665e-05,
+      "loss": 0.1083,
+      "step": 7322
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3959899749373435e-05,
+      "loss": 0.183,
+      "step": 7323
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3956319369853204e-05,
+      "loss": 0.3186,
+      "step": 7324
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3952738990332978e-05,
+      "loss": 0.3204,
+      "step": 7325
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3949158610812747e-05,
+      "loss": 0.2196,
+      "step": 7326
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3945578231292517e-05,
+      "loss": 0.1483,
+      "step": 7327
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.3941997851772287e-05,
+      "loss": 0.2272,
+      "step": 7328
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.393841747225206e-05,
+      "loss": 0.2936,
+      "step": 7329
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.393483709273183e-05,
+      "loss": 0.184,
+      "step": 7330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3931256713211603e-05,
+      "loss": 0.1398,
+      "step": 7331
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3927676333691373e-05,
+      "loss": 0.2425,
+      "step": 7332
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3924095954171143e-05,
+      "loss": 0.3073,
+      "step": 7333
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3920515574650916e-05,
+      "loss": 0.3262,
+      "step": 7334
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3916935195130686e-05,
+      "loss": 0.2092,
+      "step": 7335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3913354815610456e-05,
+      "loss": 0.1881,
+      "step": 7336
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3909774436090225e-05,
+      "loss": 0.2836,
+      "step": 7337
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.390619405657e-05,
+      "loss": 0.2196,
+      "step": 7338
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3902613677049768e-05,
+      "loss": 0.257,
+      "step": 7339
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3899033297529538e-05,
+      "loss": 0.1884,
+      "step": 7340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3895452918009308e-05,
+      "loss": 0.2242,
+      "step": 7341
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.389187253848908e-05,
+      "loss": 0.149,
+      "step": 7342
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3888292158968854e-05,
+      "loss": 0.148,
+      "step": 7343
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3884711779448624e-05,
+      "loss": 0.2083,
+      "step": 7344
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3881131399928394e-05,
+      "loss": 0.3333,
+      "step": 7345
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3877551020408164e-05,
+      "loss": 0.1631,
+      "step": 7346
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3873970640887937e-05,
+      "loss": 0.1025,
+      "step": 7347
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3870390261367707e-05,
+      "loss": 0.2851,
+      "step": 7348
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3866809881847476e-05,
+      "loss": 0.2615,
+      "step": 7349
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3863229502327246e-05,
+      "loss": 0.1494,
+      "step": 7350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.385964912280702e-05,
+      "loss": 0.2315,
+      "step": 7351
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.385606874328679e-05,
+      "loss": 0.1374,
+      "step": 7352
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.385248836376656e-05,
+      "loss": 0.1785,
+      "step": 7353
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3848907984246332e-05,
+      "loss": 0.2169,
+      "step": 7354
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3845327604726102e-05,
+      "loss": 0.1081,
+      "step": 7355
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3841747225205875e-05,
+      "loss": 0.2364,
+      "step": 7356
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.3838166845685645e-05,
+      "loss": 0.3391,
+      "step": 7357
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3834586466165415e-05,
+      "loss": 0.1557,
+      "step": 7358
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3831006086645185e-05,
+      "loss": 0.177,
+      "step": 7359
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3827425707124958e-05,
+      "loss": 0.2553,
+      "step": 7360
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3823845327604728e-05,
+      "loss": 0.1617,
+      "step": 7361
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3820264948084497e-05,
+      "loss": 0.1644,
+      "step": 7362
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3816684568564267e-05,
+      "loss": 0.15,
+      "step": 7363
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.381310418904404e-05,
+      "loss": 0.2467,
+      "step": 7364
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.2041,
+      "step": 7365
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3805943430003583e-05,
+      "loss": 0.1904,
+      "step": 7366
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3802363050483353e-05,
+      "loss": 0.1856,
+      "step": 7367
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3798782670963123e-05,
+      "loss": 0.0885,
+      "step": 7368
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3795202291442896e-05,
+      "loss": 0.1448,
+      "step": 7369
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3791621911922666e-05,
+      "loss": 0.2273,
+      "step": 7370
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3788041532402436e-05,
+      "loss": 0.3006,
+      "step": 7371
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3784461152882206e-05,
+      "loss": 0.1648,
+      "step": 7372
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.378088077336198e-05,
+      "loss": 0.326,
+      "step": 7373
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.377730039384175e-05,
+      "loss": 0.0851,
+      "step": 7374
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.377372001432152e-05,
+      "loss": 0.2294,
+      "step": 7375
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3770139634801288e-05,
+      "loss": 0.269,
+      "step": 7376
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.376655925528106e-05,
+      "loss": 0.1534,
+      "step": 7377
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3762978875760835e-05,
+      "loss": 0.2197,
+      "step": 7378
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3759398496240604e-05,
+      "loss": 0.1614,
+      "step": 7379
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3755818116720374e-05,
+      "loss": 0.1817,
+      "step": 7380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3752237737200144e-05,
+      "loss": 0.2305,
+      "step": 7381
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3748657357679914e-05,
+      "loss": 0.3587,
+      "step": 7382
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3745076978159687e-05,
+      "loss": 0.2378,
+      "step": 7383
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3741496598639457e-05,
+      "loss": 0.1401,
+      "step": 7384
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.3737916219119226e-05,
+      "loss": 0.2671,
+      "step": 7385
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3734335839598996e-05,
+      "loss": 0.3067,
+      "step": 7386
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.373075546007877e-05,
+      "loss": 0.2849,
+      "step": 7387
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.372717508055854e-05,
+      "loss": 0.2653,
+      "step": 7388
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3723594701038312e-05,
+      "loss": 0.204,
+      "step": 7389
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3720014321518082e-05,
+      "loss": 0.2021,
+      "step": 7390
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3716433941997852e-05,
+      "loss": 0.113,
+      "step": 7391
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3712853562477625e-05,
+      "loss": 0.304,
+      "step": 7392
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3709273182957395e-05,
+      "loss": 0.2745,
+      "step": 7393
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3705692803437165e-05,
+      "loss": 0.2817,
+      "step": 7394
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3702112423916935e-05,
+      "loss": 0.2664,
+      "step": 7395
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3698532044396708e-05,
+      "loss": 0.2539,
+      "step": 7396
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3694951664876478e-05,
+      "loss": 0.29,
+      "step": 7397
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3691371285356247e-05,
+      "loss": 0.3178,
+      "step": 7398
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3687790905836017e-05,
+      "loss": 0.2236,
+      "step": 7399
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.1274,
+      "step": 7400
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3680630146795564e-05,
+      "loss": 0.1851,
+      "step": 7401
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3677049767275333e-05,
+      "loss": 0.1262,
+      "step": 7402
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3673469387755103e-05,
+      "loss": 0.1885,
+      "step": 7403
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3669889008234873e-05,
+      "loss": 0.2279,
+      "step": 7404
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3666308628714646e-05,
+      "loss": 0.1848,
+      "step": 7405
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3662728249194416e-05,
+      "loss": 0.1502,
+      "step": 7406
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3659147869674186e-05,
+      "loss": 0.2286,
+      "step": 7407
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3655567490153956e-05,
+      "loss": 0.2276,
+      "step": 7408
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.365198711063373e-05,
+      "loss": 0.2253,
+      "step": 7409
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.36484067311135e-05,
+      "loss": 0.3015,
+      "step": 7410
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.364482635159327e-05,
+      "loss": 0.1619,
+      "step": 7411
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.364124597207304e-05,
+      "loss": 0.1302,
+      "step": 7412
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.363766559255281e-05,
+      "loss": 0.2443,
+      "step": 7413
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3634085213032585e-05,
+      "loss": 0.2233,
+      "step": 7414
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3630504833512354e-05,
+      "loss": 0.2597,
+      "step": 7415
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3626924453992124e-05,
+      "loss": 0.2153,
+      "step": 7416
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3623344074471894e-05,
+      "loss": 0.1766,
+      "step": 7417
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3619763694951667e-05,
+      "loss": 0.2315,
+      "step": 7418
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3616183315431437e-05,
+      "loss": 0.3278,
+      "step": 7419
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3612602935911207e-05,
+      "loss": 0.2572,
+      "step": 7420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3609022556390977e-05,
+      "loss": 0.2461,
+      "step": 7421
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.360544217687075e-05,
+      "loss": 0.2927,
+      "step": 7422
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.360186179735052e-05,
+      "loss": 0.1607,
+      "step": 7423
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3598281417830293e-05,
+      "loss": 0.3327,
+      "step": 7424
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3594701038310063e-05,
+      "loss": 0.1972,
+      "step": 7425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3591120658789832e-05,
+      "loss": 0.2669,
+      "step": 7426
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3587540279269606e-05,
+      "loss": 0.2056,
+      "step": 7427
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3583959899749375e-05,
+      "loss": 0.1958,
+      "step": 7428
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3580379520229145e-05,
+      "loss": 0.2204,
+      "step": 7429
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3576799140708915e-05,
+      "loss": 0.1437,
+      "step": 7430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3573218761188688e-05,
+      "loss": 0.2336,
+      "step": 7431
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3569638381668458e-05,
+      "loss": 0.3065,
+      "step": 7432
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3566058002148228e-05,
+      "loss": 0.284,
+      "step": 7433
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3562477622627997e-05,
+      "loss": 0.3018,
+      "step": 7434
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.355889724310777e-05,
+      "loss": 0.1807,
+      "step": 7435
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3555316863587544e-05,
+      "loss": 0.2373,
+      "step": 7436
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3551736484067314e-05,
+      "loss": 0.3416,
+      "step": 7437
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3548156104547083e-05,
+      "loss": 0.1818,
+      "step": 7438
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3544575725026853e-05,
+      "loss": 0.1916,
+      "step": 7439
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3540995345506626e-05,
+      "loss": 0.2753,
+      "step": 7440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3537414965986396e-05,
+      "loss": 0.2158,
+      "step": 7441
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3533834586466166e-05,
+      "loss": 0.2076,
+      "step": 7442
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3530254206945936e-05,
+      "loss": 0.1701,
+      "step": 7443
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.352667382742571e-05,
+      "loss": 0.2448,
+      "step": 7444
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.352309344790548e-05,
+      "loss": 0.1702,
+      "step": 7445
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.351951306838525e-05,
+      "loss": 0.21,
+      "step": 7446
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.351593268886502e-05,
+      "loss": 0.2064,
+      "step": 7447
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.351235230934479e-05,
+      "loss": 0.1424,
+      "step": 7448
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3508771929824565e-05,
+      "loss": 0.2031,
+      "step": 7449
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3505191550304335e-05,
+      "loss": 0.2293,
+      "step": 7450
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3501611170784104e-05,
+      "loss": 0.2204,
+      "step": 7451
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3498030791263874e-05,
+      "loss": 0.1276,
+      "step": 7452
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3494450411743647e-05,
+      "loss": 0.3491,
+      "step": 7453
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3490870032223417e-05,
+      "loss": 0.2326,
+      "step": 7454
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3487289652703187e-05,
+      "loss": 0.1132,
+      "step": 7455
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3483709273182957e-05,
+      "loss": 0.1406,
+      "step": 7456
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.348012889366273e-05,
+      "loss": 0.2512,
+      "step": 7457
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.34765485141425e-05,
+      "loss": 0.1949,
+      "step": 7458
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3472968134622273e-05,
+      "loss": 0.2766,
+      "step": 7459
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3469387755102043e-05,
+      "loss": 0.2258,
+      "step": 7460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3465807375581813e-05,
+      "loss": 0.2004,
+      "step": 7461
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3462226996061586e-05,
+      "loss": 0.2747,
+      "step": 7462
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3458646616541356e-05,
+      "loss": 0.1642,
+      "step": 7463
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3455066237021125e-05,
+      "loss": 0.2218,
+      "step": 7464
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3451485857500895e-05,
+      "loss": 0.2433,
+      "step": 7465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.344790547798067e-05,
+      "loss": 0.2341,
+      "step": 7466
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3444325098460438e-05,
+      "loss": 0.1462,
+      "step": 7467
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3440744718940208e-05,
+      "loss": 0.4409,
+      "step": 7468
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3437164339419978e-05,
+      "loss": 0.206,
+      "step": 7469
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3433583959899748e-05,
+      "loss": 0.2729,
+      "step": 7470
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3430003580379524e-05,
+      "loss": 0.3477,
+      "step": 7471
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3426423200859294e-05,
+      "loss": 0.1255,
+      "step": 7472
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3422842821339064e-05,
+      "loss": 0.2137,
+      "step": 7473
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3419262441818834e-05,
+      "loss": 0.3048,
+      "step": 7474
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3415682062298607e-05,
+      "loss": 0.2559,
+      "step": 7475
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3412101682778377e-05,
+      "loss": 0.289,
+      "step": 7476
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3408521303258146e-05,
+      "loss": 0.1962,
+      "step": 7477
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3404940923737916e-05,
+      "loss": 0.2505,
+      "step": 7478
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3401360544217686e-05,
+      "loss": 0.154,
+      "step": 7479
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.339778016469746e-05,
+      "loss": 0.3657,
+      "step": 7480
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.339419978517723e-05,
+      "loss": 0.2032,
+      "step": 7481
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3390619405657e-05,
+      "loss": 0.1451,
+      "step": 7482
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3387039026136772e-05,
+      "loss": 0.2823,
+      "step": 7483
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3383458646616545e-05,
+      "loss": 0.2132,
+      "step": 7484
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3379878267096315e-05,
+      "loss": 0.1286,
+      "step": 7485
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3376297887576085e-05,
+      "loss": 0.2569,
+      "step": 7486
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3372717508055854e-05,
+      "loss": 0.1068,
+      "step": 7487
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3369137128535624e-05,
+      "loss": 0.1741,
+      "step": 7488
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3365556749015397e-05,
+      "loss": 0.2178,
+      "step": 7489
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3361976369495167e-05,
+      "loss": 0.1299,
+      "step": 7490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3358395989974937e-05,
+      "loss": 0.3713,
+      "step": 7491
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3354815610454707e-05,
+      "loss": 0.1866,
+      "step": 7492
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.335123523093448e-05,
+      "loss": 0.2076,
+      "step": 7493
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3347654851414253e-05,
+      "loss": 0.2332,
+      "step": 7494
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3344074471894023e-05,
+      "loss": 0.2021,
+      "step": 7495
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3340494092373793e-05,
+      "loss": 0.136,
+      "step": 7496
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3336913712853563e-05,
+      "loss": 0.2876,
+      "step": 7497
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3333333333333336e-05,
+      "loss": 0.1644,
+      "step": 7498
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3329752953813106e-05,
+      "loss": 0.2501,
+      "step": 7499
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3326172574292875e-05,
+      "loss": 0.1362,
+      "step": 7500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3322592194772645e-05,
+      "loss": 0.3912,
+      "step": 7501
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.331901181525242e-05,
+      "loss": 0.0955,
+      "step": 7502
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3315431435732188e-05,
+      "loss": 0.3473,
+      "step": 7503
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3311851056211958e-05,
+      "loss": 0.1648,
+      "step": 7504
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3308270676691728e-05,
+      "loss": 0.2288,
+      "step": 7505
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.33046902971715e-05,
+      "loss": 0.1709,
+      "step": 7506
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3301109917651274e-05,
+      "loss": 0.1424,
+      "step": 7507
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3297529538131044e-05,
+      "loss": 0.2042,
+      "step": 7508
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3293949158610814e-05,
+      "loss": 0.1644,
+      "step": 7509
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3290368779090584e-05,
+      "loss": 0.154,
+      "step": 7510
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3286788399570357e-05,
+      "loss": 0.2693,
+      "step": 7511
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3283208020050127e-05,
+      "loss": 0.2035,
+      "step": 7512
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3279627640529896e-05,
+      "loss": 0.2455,
+      "step": 7513
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3276047261009666e-05,
+      "loss": 0.254,
+      "step": 7514
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.327246688148944e-05,
+      "loss": 0.2176,
+      "step": 7515
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.326888650196921e-05,
+      "loss": 0.3405,
+      "step": 7516
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.326530612244898e-05,
+      "loss": 0.1975,
+      "step": 7517
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3261725742928752e-05,
+      "loss": 0.1644,
+      "step": 7518
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3258145363408522e-05,
+      "loss": 0.2926,
+      "step": 7519
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3254564983888295e-05,
+      "loss": 0.2587,
+      "step": 7520
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3250984604368065e-05,
+      "loss": 0.2632,
+      "step": 7521
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3247404224847835e-05,
+      "loss": 0.1584,
+      "step": 7522
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3243823845327605e-05,
+      "loss": 0.1658,
+      "step": 7523
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3240243465807378e-05,
+      "loss": 0.1906,
+      "step": 7524
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3236663086287147e-05,
+      "loss": 0.1155,
+      "step": 7525
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3233082706766917e-05,
+      "loss": 0.157,
+      "step": 7526
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3229502327246687e-05,
+      "loss": 0.1451,
+      "step": 7527
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.322592194772646e-05,
+      "loss": 0.2075,
+      "step": 7528
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3222341568206233e-05,
+      "loss": 0.1766,
+      "step": 7529
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3218761188686003e-05,
+      "loss": 0.1819,
+      "step": 7530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3215180809165773e-05,
+      "loss": 0.1732,
+      "step": 7531
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3211600429645543e-05,
+      "loss": 0.2362,
+      "step": 7532
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3208020050125316e-05,
+      "loss": 0.2498,
+      "step": 7533
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3204439670605086e-05,
+      "loss": 0.2297,
+      "step": 7534
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3200859291084856e-05,
+      "loss": 0.2397,
+      "step": 7535
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3197278911564625e-05,
+      "loss": 0.2994,
+      "step": 7536
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.31936985320444e-05,
+      "loss": 0.1329,
+      "step": 7537
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.319011815252417e-05,
+      "loss": 0.2209,
+      "step": 7538
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3186537773003938e-05,
+      "loss": 0.2423,
+      "step": 7539
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3182957393483708e-05,
+      "loss": 0.2565,
+      "step": 7540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.317937701396348e-05,
+      "loss": 0.2849,
+      "step": 7541
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3175796634443254e-05,
+      "loss": 0.1877,
+      "step": 7542
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3172216254923024e-05,
+      "loss": 0.2077,
+      "step": 7543
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3168635875402794e-05,
+      "loss": 0.2037,
+      "step": 7544
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3165055495882564e-05,
+      "loss": 0.2544,
+      "step": 7545
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3161475116362337e-05,
+      "loss": 0.2845,
+      "step": 7546
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3157894736842107e-05,
+      "loss": 0.3475,
+      "step": 7547
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3154314357321877e-05,
+      "loss": 0.2279,
+      "step": 7548
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3150733977801646e-05,
+      "loss": 0.1236,
+      "step": 7549
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.314715359828142e-05,
+      "loss": 0.3214,
+      "step": 7550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.314357321876119e-05,
+      "loss": 0.2294,
+      "step": 7551
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.313999283924096e-05,
+      "loss": 0.2628,
+      "step": 7552
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3136412459720732e-05,
+      "loss": 0.2523,
+      "step": 7553
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3132832080200502e-05,
+      "loss": 0.2258,
+      "step": 7554
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3129251700680275e-05,
+      "loss": 0.1839,
+      "step": 7555
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3125671321160045e-05,
+      "loss": 0.2957,
+      "step": 7556
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3122090941639815e-05,
+      "loss": 0.2863,
+      "step": 7557
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3118510562119585e-05,
+      "loss": 0.2632,
+      "step": 7558
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3114930182599358e-05,
+      "loss": 0.1256,
+      "step": 7559
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3111349803079128e-05,
+      "loss": 0.3493,
+      "step": 7560
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3107769423558898e-05,
+      "loss": 0.1789,
+      "step": 7561
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3104189044038667e-05,
+      "loss": 0.2912,
+      "step": 7562
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.310060866451844e-05,
+      "loss": 0.2332,
+      "step": 7563
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3097028284998214e-05,
+      "loss": 0.2151,
+      "step": 7564
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3093447905477984e-05,
+      "loss": 0.2278,
+      "step": 7565
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3089867525957753e-05,
+      "loss": 0.2795,
+      "step": 7566
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3086287146437523e-05,
+      "loss": 0.2244,
+      "step": 7567
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3082706766917296e-05,
+      "loss": 0.164,
+      "step": 7568
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3079126387397066e-05,
+      "loss": 0.1656,
+      "step": 7569
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3075546007876836e-05,
+      "loss": 0.2437,
+      "step": 7570
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3071965628356606e-05,
+      "loss": 0.2232,
+      "step": 7571
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3068385248836376e-05,
+      "loss": 0.1543,
+      "step": 7572
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.306480486931615e-05,
+      "loss": 0.1803,
+      "step": 7573
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.306122448979592e-05,
+      "loss": 0.1821,
+      "step": 7574
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3057644110275688e-05,
+      "loss": 0.2792,
+      "step": 7575
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.305406373075546e-05,
+      "loss": 0.3175,
+      "step": 7576
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3050483351235235e-05,
+      "loss": 0.1435,
+      "step": 7577
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3046902971715004e-05,
+      "loss": 0.2605,
+      "step": 7578
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3043322592194774e-05,
+      "loss": 0.1991,
+      "step": 7579
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3039742212674544e-05,
+      "loss": 0.1427,
+      "step": 7580
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3036161833154314e-05,
+      "loss": 0.1502,
+      "step": 7581
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3032581453634087e-05,
+      "loss": 0.2423,
+      "step": 7582
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3029001074113857e-05,
+      "loss": 0.2382,
+      "step": 7583
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3025420694593627e-05,
+      "loss": 0.1518,
+      "step": 7584
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3021840315073396e-05,
+      "loss": 0.1273,
+      "step": 7585
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.301825993555317e-05,
+      "loss": 0.2439,
+      "step": 7586
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.301467955603294e-05,
+      "loss": 0.2298,
+      "step": 7587
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3011099176512713e-05,
+      "loss": 0.2237,
+      "step": 7588
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3007518796992482e-05,
+      "loss": 0.1847,
+      "step": 7589
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3003938417472252e-05,
+      "loss": 0.4401,
+      "step": 7590
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.3000358037952025e-05,
+      "loss": 0.1341,
+      "step": 7591
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2996777658431795e-05,
+      "loss": 0.1122,
+      "step": 7592
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2993197278911565e-05,
+      "loss": 0.2015,
+      "step": 7593
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2989616899391335e-05,
+      "loss": 0.1535,
+      "step": 7594
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2986036519871108e-05,
+      "loss": 0.1292,
+      "step": 7595
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2982456140350878e-05,
+      "loss": 0.1512,
+      "step": 7596
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2978875760830648e-05,
+      "loss": 0.1999,
+      "step": 7597
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2975295381310417e-05,
+      "loss": 0.1314,
+      "step": 7598
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.297171500179019e-05,
+      "loss": 0.2835,
+      "step": 7599
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2968134622269964e-05,
+      "loss": 0.2424,
+      "step": 7600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2964554242749734e-05,
+      "loss": 0.2509,
+      "step": 7601
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2960973863229503e-05,
+      "loss": 0.1608,
+      "step": 7602
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2957393483709273e-05,
+      "loss": 0.1569,
+      "step": 7603
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2953813104189046e-05,
+      "loss": 0.1576,
+      "step": 7604
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2950232724668816e-05,
+      "loss": 0.2177,
+      "step": 7605
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2946652345148586e-05,
+      "loss": 0.2062,
+      "step": 7606
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2943071965628356e-05,
+      "loss": 0.4423,
+      "step": 7607
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.293949158610813e-05,
+      "loss": 0.2893,
+      "step": 7608
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.29359112065879e-05,
+      "loss": 0.2878,
+      "step": 7609
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.293233082706767e-05,
+      "loss": 0.243,
+      "step": 7610
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2928750447547442e-05,
+      "loss": 0.1132,
+      "step": 7611
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.292517006802721e-05,
+      "loss": 0.3695,
+      "step": 7612
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2921589688506985e-05,
+      "loss": 0.1339,
+      "step": 7613
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2918009308986755e-05,
+      "loss": 0.3523,
+      "step": 7614
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2914428929466524e-05,
+      "loss": 0.2068,
+      "step": 7615
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2910848549946294e-05,
+      "loss": 0.1828,
+      "step": 7616
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2907268170426067e-05,
+      "loss": 0.1296,
+      "step": 7617
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2903687790905837e-05,
+      "loss": 0.2233,
+      "step": 7618
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2900107411385607e-05,
+      "loss": 0.1752,
+      "step": 7619
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2896527031865377e-05,
+      "loss": 0.3012,
+      "step": 7620
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.289294665234515e-05,
+      "loss": 0.2676,
+      "step": 7621
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.288936627282492e-05,
+      "loss": 0.2249,
+      "step": 7622
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2885785893304693e-05,
+      "loss": 0.2085,
+      "step": 7623
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2882205513784463e-05,
+      "loss": 0.2783,
+      "step": 7624
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2878625134264232e-05,
+      "loss": 0.2537,
+      "step": 7625
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2875044754744006e-05,
+      "loss": 0.3501,
+      "step": 7626
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2871464375223775e-05,
+      "loss": 0.2053,
+      "step": 7627
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2867883995703545e-05,
+      "loss": 0.2994,
+      "step": 7628
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2864303616183315e-05,
+      "loss": 0.1949,
+      "step": 7629
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2860723236663088e-05,
+      "loss": 0.2475,
+      "step": 7630
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2857142857142858e-05,
+      "loss": 0.2421,
+      "step": 7631
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2853562477622628e-05,
+      "loss": 0.2646,
+      "step": 7632
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2849982098102398e-05,
+      "loss": 0.2501,
+      "step": 7633
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.284640171858217e-05,
+      "loss": 0.209,
+      "step": 7634
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2842821339061944e-05,
+      "loss": 0.1834,
+      "step": 7635
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2839240959541714e-05,
+      "loss": 0.1038,
+      "step": 7636
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2835660580021484e-05,
+      "loss": 0.1633,
+      "step": 7637
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2832080200501253e-05,
+      "loss": 0.2259,
+      "step": 7638
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2828499820981027e-05,
+      "loss": 0.211,
+      "step": 7639
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2824919441460796e-05,
+      "loss": 0.2668,
+      "step": 7640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2821339061940566e-05,
+      "loss": 0.1943,
+      "step": 7641
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2817758682420336e-05,
+      "loss": 0.324,
+      "step": 7642
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.281417830290011e-05,
+      "loss": 0.2517,
+      "step": 7643
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.281059792337988e-05,
+      "loss": 0.2459,
+      "step": 7644
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.280701754385965e-05,
+      "loss": 0.3213,
+      "step": 7645
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2803437164339422e-05,
+      "loss": 0.1443,
+      "step": 7646
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2799856784819192e-05,
+      "loss": 0.2921,
+      "step": 7647
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2796276405298965e-05,
+      "loss": 0.1799,
+      "step": 7648
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2792696025778735e-05,
+      "loss": 0.1738,
+      "step": 7649
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2789115646258505e-05,
+      "loss": 0.0802,
+      "step": 7650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2785535266738274e-05,
+      "loss": 0.1866,
+      "step": 7651
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2781954887218048e-05,
+      "loss": 0.366,
+      "step": 7652
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2778374507697817e-05,
+      "loss": 0.1519,
+      "step": 7653
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2774794128177587e-05,
+      "loss": 0.285,
+      "step": 7654
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2771213748657357e-05,
+      "loss": 0.2135,
+      "step": 7655
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.276763336913713e-05,
+      "loss": 0.2355,
+      "step": 7656
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.27640529896169e-05,
+      "loss": 0.3417,
+      "step": 7657
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2760472610096673e-05,
+      "loss": 0.163,
+      "step": 7658
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2756892230576443e-05,
+      "loss": 0.2739,
+      "step": 7659
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2753311851056213e-05,
+      "loss": 0.1309,
+      "step": 7660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2749731471535986e-05,
+      "loss": 0.2673,
+      "step": 7661
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2746151092015756e-05,
+      "loss": 0.2085,
+      "step": 7662
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2742570712495526e-05,
+      "loss": 0.2253,
+      "step": 7663
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2738990332975295e-05,
+      "loss": 0.1421,
+      "step": 7664
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.273540995345507e-05,
+      "loss": 0.142,
+      "step": 7665
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2731829573934838e-05,
+      "loss": 0.1806,
+      "step": 7666
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2728249194414608e-05,
+      "loss": 0.3269,
+      "step": 7667
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2724668814894378e-05,
+      "loss": 0.1186,
+      "step": 7668
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.272108843537415e-05,
+      "loss": 0.1969,
+      "step": 7669
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2717508055853924e-05,
+      "loss": 0.1701,
+      "step": 7670
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2713927676333694e-05,
+      "loss": 0.3354,
+      "step": 7671
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2710347296813464e-05,
+      "loss": 0.2519,
+      "step": 7672
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2706766917293234e-05,
+      "loss": 0.233,
+      "step": 7673
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2703186537773007e-05,
+      "loss": 0.1952,
+      "step": 7674
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2699606158252777e-05,
+      "loss": 0.1418,
+      "step": 7675
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2696025778732546e-05,
+      "loss": 0.3644,
+      "step": 7676
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2692445399212316e-05,
+      "loss": 0.2639,
+      "step": 7677
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2688865019692086e-05,
+      "loss": 0.3314,
+      "step": 7678
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.268528464017186e-05,
+      "loss": 0.1689,
+      "step": 7679
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.268170426065163e-05,
+      "loss": 0.1628,
+      "step": 7680
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2678123881131402e-05,
+      "loss": 0.2579,
+      "step": 7681
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2674543501611172e-05,
+      "loss": 0.1856,
+      "step": 7682
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2670963122090945e-05,
+      "loss": 0.1967,
+      "step": 7683
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2667382742570715e-05,
+      "loss": 0.1992,
+      "step": 7684
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2663802363050485e-05,
+      "loss": 0.2265,
+      "step": 7685
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2660221983530255e-05,
+      "loss": 0.2031,
+      "step": 7686
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2656641604010024e-05,
+      "loss": 0.2159,
+      "step": 7687
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2653061224489798e-05,
+      "loss": 0.089,
+      "step": 7688
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2649480844969567e-05,
+      "loss": 0.2408,
+      "step": 7689
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2645900465449337e-05,
+      "loss": 0.149,
+      "step": 7690
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2642320085929107e-05,
+      "loss": 0.1369,
+      "step": 7691
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.263873970640888e-05,
+      "loss": 0.3271,
+      "step": 7692
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2635159326888653e-05,
+      "loss": 0.3172,
+      "step": 7693
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2631578947368423e-05,
+      "loss": 0.1693,
+      "step": 7694
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2627998567848193e-05,
+      "loss": 0.1824,
+      "step": 7695
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2624418188327963e-05,
+      "loss": 0.2002,
+      "step": 7696
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2620837808807736e-05,
+      "loss": 0.1299,
+      "step": 7697
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2617257429287506e-05,
+      "loss": 0.1566,
+      "step": 7698
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2613677049767276e-05,
+      "loss": 0.0959,
+      "step": 7699
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2610096670247045e-05,
+      "loss": 0.0892,
+      "step": 7700
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.260651629072682e-05,
+      "loss": 0.1577,
+      "step": 7701
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.260293591120659e-05,
+      "loss": 0.2105,
+      "step": 7702
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2599355531686358e-05,
+      "loss": 0.1012,
+      "step": 7703
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.259577515216613e-05,
+      "loss": 0.408,
+      "step": 7704
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.25921947726459e-05,
+      "loss": 0.1758,
+      "step": 7705
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2588614393125674e-05,
+      "loss": 0.1683,
+      "step": 7706
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2585034013605444e-05,
+      "loss": 0.3715,
+      "step": 7707
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2581453634085214e-05,
+      "loss": 0.1602,
+      "step": 7708
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2577873254564984e-05,
+      "loss": 0.2517,
+      "step": 7709
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2574292875044757e-05,
+      "loss": 0.1851,
+      "step": 7710
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2570712495524527e-05,
+      "loss": 0.1734,
+      "step": 7711
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2567132116004297e-05,
+      "loss": 0.2003,
+      "step": 7712
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2563551736484066e-05,
+      "loss": 0.2297,
+      "step": 7713
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.255997135696384e-05,
+      "loss": 0.1294,
+      "step": 7714
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.255639097744361e-05,
+      "loss": 0.2398,
+      "step": 7715
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2552810597923382e-05,
+      "loss": 0.1023,
+      "step": 7716
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2549230218403152e-05,
+      "loss": 0.2221,
+      "step": 7717
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2545649838882922e-05,
+      "loss": 0.2408,
+      "step": 7718
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2542069459362695e-05,
+      "loss": 0.161,
+      "step": 7719
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2538489079842465e-05,
+      "loss": 0.2785,
+      "step": 7720
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2534908700322235e-05,
+      "loss": 0.168,
+      "step": 7721
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2531328320802005e-05,
+      "loss": 0.1268,
+      "step": 7722
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2527747941281778e-05,
+      "loss": 0.2013,
+      "step": 7723
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2524167561761548e-05,
+      "loss": 0.1984,
+      "step": 7724
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2520587182241317e-05,
+      "loss": 0.2351,
+      "step": 7725
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2517006802721087e-05,
+      "loss": 0.2025,
+      "step": 7726
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.251342642320086e-05,
+      "loss": 0.2505,
+      "step": 7727
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2509846043680634e-05,
+      "loss": 0.2647,
+      "step": 7728
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2506265664160403e-05,
+      "loss": 0.2476,
+      "step": 7729
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2502685284640173e-05,
+      "loss": 0.1704,
+      "step": 7730
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2499104905119943e-05,
+      "loss": 0.2633,
+      "step": 7731
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2495524525599716e-05,
+      "loss": 0.356,
+      "step": 7732
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2491944146079486e-05,
+      "loss": 0.3264,
+      "step": 7733
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2488363766559256e-05,
+      "loss": 0.2623,
+      "step": 7734
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2484783387039026e-05,
+      "loss": 0.1473,
+      "step": 7735
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.24812030075188e-05,
+      "loss": 0.1959,
+      "step": 7736
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.247762262799857e-05,
+      "loss": 0.2696,
+      "step": 7737
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.247404224847834e-05,
+      "loss": 0.1817,
+      "step": 7738
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.247046186895811e-05,
+      "loss": 0.1536,
+      "step": 7739
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.246688148943788e-05,
+      "loss": 0.1983,
+      "step": 7740
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2463301109917655e-05,
+      "loss": 0.1106,
+      "step": 7741
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2459720730397424e-05,
+      "loss": 0.2166,
+      "step": 7742
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2456140350877194e-05,
+      "loss": 0.2538,
+      "step": 7743
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2452559971356964e-05,
+      "loss": 0.182,
+      "step": 7744
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2448979591836737e-05,
+      "loss": 0.24,
+      "step": 7745
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2445399212316507e-05,
+      "loss": 0.2529,
+      "step": 7746
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2441818832796277e-05,
+      "loss": 0.1106,
+      "step": 7747
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2438238453276047e-05,
+      "loss": 0.2031,
+      "step": 7748
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.243465807375582e-05,
+      "loss": 0.175,
+      "step": 7749
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.243107769423559e-05,
+      "loss": 0.2378,
+      "step": 7750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2427497314715363e-05,
+      "loss": 0.1252,
+      "step": 7751
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2423916935195133e-05,
+      "loss": 0.151,
+      "step": 7752
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2420336555674902e-05,
+      "loss": 0.2401,
+      "step": 7753
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2416756176154676e-05,
+      "loss": 0.2515,
+      "step": 7754
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2413175796634445e-05,
+      "loss": 0.1814,
+      "step": 7755
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2409595417114215e-05,
+      "loss": 0.1408,
+      "step": 7756
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2406015037593985e-05,
+      "loss": 0.1237,
+      "step": 7757
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2402434658073758e-05,
+      "loss": 0.2272,
+      "step": 7758
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2398854278553528e-05,
+      "loss": 0.1371,
+      "step": 7759
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2395273899033298e-05,
+      "loss": 0.1457,
+      "step": 7760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2391693519513067e-05,
+      "loss": 0.3492,
+      "step": 7761
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.238811313999284e-05,
+      "loss": 0.2683,
+      "step": 7762
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2384532760472614e-05,
+      "loss": 0.1943,
+      "step": 7763
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2380952380952384e-05,
+      "loss": 0.1413,
+      "step": 7764
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2377372001432153e-05,
+      "loss": 0.1895,
+      "step": 7765
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2373791621911923e-05,
+      "loss": 0.2177,
+      "step": 7766
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2370211242391696e-05,
+      "loss": 0.1633,
+      "step": 7767
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2366630862871466e-05,
+      "loss": 0.2935,
+      "step": 7768
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2363050483351236e-05,
+      "loss": 0.273,
+      "step": 7769
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2359470103831006e-05,
+      "loss": 0.2466,
+      "step": 7770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2355889724310776e-05,
+      "loss": 0.1113,
+      "step": 7771
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.235230934479055e-05,
+      "loss": 0.2008,
+      "step": 7772
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.234872896527032e-05,
+      "loss": 0.145,
+      "step": 7773
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2345148585750092e-05,
+      "loss": 0.0996,
+      "step": 7774
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.234156820622986e-05,
+      "loss": 0.2546,
+      "step": 7775
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2337987826709635e-05,
+      "loss": 0.16,
+      "step": 7776
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2334407447189405e-05,
+      "loss": 0.2727,
+      "step": 7777
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2330827067669174e-05,
+      "loss": 0.1506,
+      "step": 7778
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2327246688148944e-05,
+      "loss": 0.2344,
+      "step": 7779
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2323666308628714e-05,
+      "loss": 0.2897,
+      "step": 7780
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2320085929108487e-05,
+      "loss": 0.1773,
+      "step": 7781
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2316505549588257e-05,
+      "loss": 0.2663,
+      "step": 7782
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2312925170068027e-05,
+      "loss": 0.2592,
+      "step": 7783
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2309344790547797e-05,
+      "loss": 0.2763,
+      "step": 7784
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.230576441102757e-05,
+      "loss": 0.0849,
+      "step": 7785
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2302184031507343e-05,
+      "loss": 0.2719,
+      "step": 7786
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2298603651987113e-05,
+      "loss": 0.1094,
+      "step": 7787
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2295023272466883e-05,
+      "loss": 0.2983,
+      "step": 7788
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2291442892946652e-05,
+      "loss": 0.1438,
+      "step": 7789
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2287862513426426e-05,
+      "loss": 0.3062,
+      "step": 7790
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2284282133906195e-05,
+      "loss": 0.1113,
+      "step": 7791
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2280701754385965e-05,
+      "loss": 0.2065,
+      "step": 7792
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2277121374865735e-05,
+      "loss": 0.3145,
+      "step": 7793
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2273540995345508e-05,
+      "loss": 0.1451,
+      "step": 7794
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2269960615825278e-05,
+      "loss": 0.1161,
+      "step": 7795
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2266380236305048e-05,
+      "loss": 0.1628,
+      "step": 7796
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2262799856784818e-05,
+      "loss": 0.2873,
+      "step": 7797
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.225921947726459e-05,
+      "loss": 0.2364,
+      "step": 7798
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2255639097744364e-05,
+      "loss": 0.3157,
+      "step": 7799
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2252058718224134e-05,
+      "loss": 0.2693,
+      "step": 7800
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2248478338703904e-05,
+      "loss": 0.3431,
+      "step": 7801
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2244897959183673e-05,
+      "loss": 0.2326,
+      "step": 7802
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2241317579663447e-05,
+      "loss": 0.3296,
+      "step": 7803
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2237737200143216e-05,
+      "loss": 0.5205,
+      "step": 7804
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2234156820622986e-05,
+      "loss": 0.2411,
+      "step": 7805
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2230576441102756e-05,
+      "loss": 0.2671,
+      "step": 7806
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.222699606158253e-05,
+      "loss": 0.2666,
+      "step": 7807
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.22234156820623e-05,
+      "loss": 0.226,
+      "step": 7808
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2219835302542072e-05,
+      "loss": 0.214,
+      "step": 7809
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2216254923021842e-05,
+      "loss": 0.2275,
+      "step": 7810
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.221267454350161e-05,
+      "loss": 0.2132,
+      "step": 7811
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2209094163981385e-05,
+      "loss": 0.3033,
+      "step": 7812
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2205513784461155e-05,
+      "loss": 0.2773,
+      "step": 7813
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2201933404940924e-05,
+      "loss": 0.2483,
+      "step": 7814
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2198353025420694e-05,
+      "loss": 0.199,
+      "step": 7815
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2194772645900467e-05,
+      "loss": 0.1234,
+      "step": 7816
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2191192266380237e-05,
+      "loss": 0.2022,
+      "step": 7817
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2187611886860007e-05,
+      "loss": 0.2826,
+      "step": 7818
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2184031507339777e-05,
+      "loss": 0.266,
+      "step": 7819
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.218045112781955e-05,
+      "loss": 0.1955,
+      "step": 7820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2176870748299323e-05,
+      "loss": 0.1506,
+      "step": 7821
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2173290368779093e-05,
+      "loss": 0.3278,
+      "step": 7822
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2169709989258863e-05,
+      "loss": 0.2055,
+      "step": 7823
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2166129609738633e-05,
+      "loss": 0.155,
+      "step": 7824
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2162549230218406e-05,
+      "loss": 0.1239,
+      "step": 7825
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2158968850698176e-05,
+      "loss": 0.1932,
+      "step": 7826
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2155388471177945e-05,
+      "loss": 0.2023,
+      "step": 7827
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2151808091657715e-05,
+      "loss": 0.1547,
+      "step": 7828
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.214822771213749e-05,
+      "loss": 0.2666,
+      "step": 7829
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2144647332617258e-05,
+      "loss": 0.1815,
+      "step": 7830
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2141066953097028e-05,
+      "loss": 0.2049,
+      "step": 7831
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2137486573576798e-05,
+      "loss": 0.2572,
+      "step": 7832
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.213390619405657e-05,
+      "loss": 0.2157,
+      "step": 7833
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.2130325814536344e-05,
+      "loss": 0.2079,
+      "step": 7834
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2126745435016114e-05,
+      "loss": 0.1216,
+      "step": 7835
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2123165055495884e-05,
+      "loss": 0.1748,
+      "step": 7836
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2119584675975654e-05,
+      "loss": 0.2103,
+      "step": 7837
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2116004296455427e-05,
+      "loss": 0.2426,
+      "step": 7838
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2112423916935197e-05,
+      "loss": 0.248,
+      "step": 7839
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2108843537414966e-05,
+      "loss": 0.1506,
+      "step": 7840
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2105263157894736e-05,
+      "loss": 0.2918,
+      "step": 7841
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.210168277837451e-05,
+      "loss": 0.2799,
+      "step": 7842
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.209810239885428e-05,
+      "loss": 0.1591,
+      "step": 7843
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.209452201933405e-05,
+      "loss": 0.1573,
+      "step": 7844
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2090941639813822e-05,
+      "loss": 0.1705,
+      "step": 7845
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2087361260293592e-05,
+      "loss": 0.285,
+      "step": 7846
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2083780880773365e-05,
+      "loss": 0.2174,
+      "step": 7847
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2080200501253135e-05,
+      "loss": 0.2349,
+      "step": 7848
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2076620121732905e-05,
+      "loss": 0.3323,
+      "step": 7849
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2073039742212675e-05,
+      "loss": 0.2134,
+      "step": 7850
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2069459362692448e-05,
+      "loss": 0.2885,
+      "step": 7851
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2065878983172218e-05,
+      "loss": 0.2207,
+      "step": 7852
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2062298603651987e-05,
+      "loss": 0.3527,
+      "step": 7853
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2058718224131757e-05,
+      "loss": 0.2406,
+      "step": 7854
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.205513784461153e-05,
+      "loss": 0.1637,
+      "step": 7855
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2051557465091303e-05,
+      "loss": 0.2227,
+      "step": 7856
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2047977085571073e-05,
+      "loss": 0.2272,
+      "step": 7857
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2044396706050843e-05,
+      "loss": 0.1137,
+      "step": 7858
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2040816326530613e-05,
+      "loss": 0.1916,
+      "step": 7859
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2037235947010386e-05,
+      "loss": 0.2849,
+      "step": 7860
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2033655567490156e-05,
+      "loss": 0.1266,
+      "step": 7861
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2030075187969926e-05,
+      "loss": 0.268,
+      "step": 7862
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2026494808449695e-05,
+      "loss": 0.2096,
+      "step": 7863
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.202291442892947e-05,
+      "loss": 0.3026,
+      "step": 7864
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.201933404940924e-05,
+      "loss": 0.2591,
+      "step": 7865
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2015753669889008e-05,
+      "loss": 0.1909,
+      "step": 7866
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2012173290368778e-05,
+      "loss": 0.2174,
+      "step": 7867
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.200859291084855e-05,
+      "loss": 0.2651,
+      "step": 7868
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2005012531328324e-05,
+      "loss": 0.25,
+      "step": 7869
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2001432151808094e-05,
+      "loss": 0.1862,
+      "step": 7870
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1997851772287864e-05,
+      "loss": 0.2281,
+      "step": 7871
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1994271392767634e-05,
+      "loss": 0.2318,
+      "step": 7872
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1990691013247407e-05,
+      "loss": 0.1898,
+      "step": 7873
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1987110633727177e-05,
+      "loss": 0.3488,
+      "step": 7874
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1983530254206947e-05,
+      "loss": 0.1807,
+      "step": 7875
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1979949874686716e-05,
+      "loss": 0.2755,
+      "step": 7876
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1976369495166486e-05,
+      "loss": 0.1571,
+      "step": 7877
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.197278911564626e-05,
+      "loss": 0.2043,
+      "step": 7878
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.196920873612603e-05,
+      "loss": 0.2732,
+      "step": 7879
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1965628356605802e-05,
+      "loss": 0.1986,
+      "step": 7880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1962047977085572e-05,
+      "loss": 0.2615,
+      "step": 7881
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1958467597565342e-05,
+      "loss": 0.1444,
+      "step": 7882
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1954887218045115e-05,
+      "loss": 0.1774,
+      "step": 7883
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1951306838524885e-05,
+      "loss": 0.2848,
+      "step": 7884
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1947726459004655e-05,
+      "loss": 0.1493,
+      "step": 7885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1944146079484425e-05,
+      "loss": 0.1633,
+      "step": 7886
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1940565699964198e-05,
+      "loss": 0.1908,
+      "step": 7887
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1936985320443968e-05,
+      "loss": 0.2487,
+      "step": 7888
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1933404940923737e-05,
+      "loss": 0.363,
+      "step": 7889
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1929824561403507e-05,
+      "loss": 0.3061,
+      "step": 7890
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.192624418188328e-05,
+      "loss": 0.2238,
+      "step": 7891
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1922663802363054e-05,
+      "loss": 0.3722,
+      "step": 7892
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1919083422842823e-05,
+      "loss": 0.1673,
+      "step": 7893
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1915503043322593e-05,
+      "loss": 0.2072,
+      "step": 7894
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1911922663802363e-05,
+      "loss": 0.244,
+      "step": 7895
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1908342284282136e-05,
+      "loss": 0.1518,
+      "step": 7896
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1904761904761906e-05,
+      "loss": 0.1162,
+      "step": 7897
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1901181525241676e-05,
+      "loss": 0.3276,
+      "step": 7898
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1897601145721446e-05,
+      "loss": 0.3364,
+      "step": 7899
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.189402076620122e-05,
+      "loss": 0.0777,
+      "step": 7900
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.189044038668099e-05,
+      "loss": 0.2227,
+      "step": 7901
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1886860007160758e-05,
+      "loss": 0.2704,
+      "step": 7902
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.188327962764053e-05,
+      "loss": 0.2179,
+      "step": 7903
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.18796992481203e-05,
+      "loss": 0.1151,
+      "step": 7904
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1876118868600074e-05,
+      "loss": 0.1786,
+      "step": 7905
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1872538489079844e-05,
+      "loss": 0.2858,
+      "step": 7906
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1868958109559614e-05,
+      "loss": 0.1659,
+      "step": 7907
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1865377730039384e-05,
+      "loss": 0.2129,
+      "step": 7908
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1861797350519157e-05,
+      "loss": 0.175,
+      "step": 7909
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1858216970998927e-05,
+      "loss": 0.202,
+      "step": 7910
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1854636591478697e-05,
+      "loss": 0.1956,
+      "step": 7911
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1851056211958466e-05,
+      "loss": 0.1784,
+      "step": 7912
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.184747583243824e-05,
+      "loss": 0.1514,
+      "step": 7913
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.184389545291801e-05,
+      "loss": 0.1621,
+      "step": 7914
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1840315073397783e-05,
+      "loss": 0.1634,
+      "step": 7915
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1836734693877552e-05,
+      "loss": 0.181,
+      "step": 7916
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1833154314357322e-05,
+      "loss": 0.4378,
+      "step": 7917
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1829573934837095e-05,
+      "loss": 0.1946,
+      "step": 7918
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1825993555316865e-05,
+      "loss": 0.1968,
+      "step": 7919
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1822413175796635e-05,
+      "loss": 0.3941,
+      "step": 7920
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1818832796276405e-05,
+      "loss": 0.1547,
+      "step": 7921
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1815252416756178e-05,
+      "loss": 0.3249,
+      "step": 7922
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1811672037235948e-05,
+      "loss": 0.1123,
+      "step": 7923
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1808091657715718e-05,
+      "loss": 0.1575,
+      "step": 7924
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1804511278195487e-05,
+      "loss": 0.34,
+      "step": 7925
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.180093089867526e-05,
+      "loss": 0.2085,
+      "step": 7926
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1797350519155034e-05,
+      "loss": 0.3865,
+      "step": 7927
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1793770139634804e-05,
+      "loss": 0.1993,
+      "step": 7928
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1790189760114573e-05,
+      "loss": 0.208,
+      "step": 7929
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1786609380594343e-05,
+      "loss": 0.2784,
+      "step": 7930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1783029001074116e-05,
+      "loss": 0.2238,
+      "step": 7931
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1779448621553886e-05,
+      "loss": 0.1766,
+      "step": 7932
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1775868242033656e-05,
+      "loss": 0.118,
+      "step": 7933
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1772287862513426e-05,
+      "loss": 0.2111,
+      "step": 7934
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.17687074829932e-05,
+      "loss": 0.263,
+      "step": 7935
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.176512710347297e-05,
+      "loss": 0.122,
+      "step": 7936
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.176154672395274e-05,
+      "loss": 0.2526,
+      "step": 7937
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1757966344432512e-05,
+      "loss": 0.1178,
+      "step": 7938
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.175438596491228e-05,
+      "loss": 0.1505,
+      "step": 7939
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1750805585392055e-05,
+      "loss": 0.2174,
+      "step": 7940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1747225205871825e-05,
+      "loss": 0.2808,
+      "step": 7941
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1743644826351594e-05,
+      "loss": 0.3293,
+      "step": 7942
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1740064446831364e-05,
+      "loss": 0.1836,
+      "step": 7943
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1736484067311137e-05,
+      "loss": 0.1778,
+      "step": 7944
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1732903687790907e-05,
+      "loss": 0.2145,
+      "step": 7945
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.1729323308270677e-05,
+      "loss": 0.2001,
+      "step": 7946
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1725742928750447e-05,
+      "loss": 0.151,
+      "step": 7947
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.172216254923022e-05,
+      "loss": 0.3539,
+      "step": 7948
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.171858216970999e-05,
+      "loss": 0.1473,
+      "step": 7949
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1715001790189763e-05,
+      "loss": 0.313,
+      "step": 7950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1711421410669533e-05,
+      "loss": 0.176,
+      "step": 7951
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1707841031149302e-05,
+      "loss": 0.1661,
+      "step": 7952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1704260651629076e-05,
+      "loss": 0.1531,
+      "step": 7953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1700680272108845e-05,
+      "loss": 0.1222,
+      "step": 7954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1697099892588615e-05,
+      "loss": 0.2068,
+      "step": 7955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1693519513068385e-05,
+      "loss": 0.198,
+      "step": 7956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1689939133548158e-05,
+      "loss": 0.252,
+      "step": 7957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1686358754027928e-05,
+      "loss": 0.3442,
+      "step": 7958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1682778374507698e-05,
+      "loss": 0.1388,
+      "step": 7959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1679197994987468e-05,
+      "loss": 0.1819,
+      "step": 7960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.167561761546724e-05,
+      "loss": 0.1632,
+      "step": 7961
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1672037235947014e-05,
+      "loss": 0.2641,
+      "step": 7962
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1668456856426784e-05,
+      "loss": 0.1409,
+      "step": 7963
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1664876476906554e-05,
+      "loss": 0.2371,
+      "step": 7964
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1661296097386323e-05,
+      "loss": 0.2203,
+      "step": 7965
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1657715717866097e-05,
+      "loss": 0.181,
+      "step": 7966
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1654135338345866e-05,
+      "loss": 0.1706,
+      "step": 7967
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1650554958825636e-05,
+      "loss": 0.1919,
+      "step": 7968
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1646974579305406e-05,
+      "loss": 0.1561,
+      "step": 7969
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1643394199785176e-05,
+      "loss": 0.2113,
+      "step": 7970
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.163981382026495e-05,
+      "loss": 0.1272,
+      "step": 7971
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.163623344074472e-05,
+      "loss": 0.23,
+      "step": 7972
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1632653061224492e-05,
+      "loss": 0.3039,
+      "step": 7973
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.1629072681704262e-05,
+      "loss": 0.2262,
+      "step": 7974
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1625492302184035e-05,
+      "loss": 0.1666,
+      "step": 7975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1621911922663805e-05,
+      "loss": 0.3034,
+      "step": 7976
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1618331543143575e-05,
+      "loss": 0.1697,
+      "step": 7977
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1614751163623344e-05,
+      "loss": 0.1493,
+      "step": 7978
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1611170784103114e-05,
+      "loss": 0.1991,
+      "step": 7979
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1607590404582887e-05,
+      "loss": 0.358,
+      "step": 7980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1604010025062657e-05,
+      "loss": 0.2938,
+      "step": 7981
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1600429645542427e-05,
+      "loss": 0.2177,
+      "step": 7982
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1596849266022197e-05,
+      "loss": 0.2325,
+      "step": 7983
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.159326888650197e-05,
+      "loss": 0.2522,
+      "step": 7984
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1589688506981743e-05,
+      "loss": 0.2703,
+      "step": 7985
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1586108127461513e-05,
+      "loss": 0.2765,
+      "step": 7986
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1582527747941283e-05,
+      "loss": 0.1787,
+      "step": 7987
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1578947368421053e-05,
+      "loss": 0.2152,
+      "step": 7988
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1575366988900826e-05,
+      "loss": 0.2448,
+      "step": 7989
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1571786609380596e-05,
+      "loss": 0.129,
+      "step": 7990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1568206229860365e-05,
+      "loss": 0.2208,
+      "step": 7991
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1564625850340135e-05,
+      "loss": 0.1842,
+      "step": 7992
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1561045470819908e-05,
+      "loss": 0.1941,
+      "step": 7993
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1557465091299678e-05,
+      "loss": 0.1443,
+      "step": 7994
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1553884711779448e-05,
+      "loss": 0.1455,
+      "step": 7995
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.155030433225922e-05,
+      "loss": 0.2783,
+      "step": 7996
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.154672395273899e-05,
+      "loss": 0.2472,
+      "step": 7997
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1543143573218764e-05,
+      "loss": 0.2015,
+      "step": 7998
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1539563193698534e-05,
+      "loss": 0.2561,
+      "step": 7999
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1535982814178304e-05,
+      "loss": 0.2998,
+      "step": 8000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1532402434658073e-05,
+      "loss": 0.1751,
+      "step": 8001
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.1528822055137847e-05,
+      "loss": 0.2728,
+      "step": 8002
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1525241675617616e-05,
+      "loss": 0.2536,
+      "step": 8003
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1521661296097386e-05,
+      "loss": 0.2821,
+      "step": 8004
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1518080916577156e-05,
+      "loss": 0.233,
+      "step": 8005
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.151450053705693e-05,
+      "loss": 0.2767,
+      "step": 8006
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.15109201575367e-05,
+      "loss": 0.1415,
+      "step": 8007
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1507339778016472e-05,
+      "loss": 0.2093,
+      "step": 8008
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1503759398496242e-05,
+      "loss": 0.3724,
+      "step": 8009
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1500179018976012e-05,
+      "loss": 0.2008,
+      "step": 8010
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1496598639455785e-05,
+      "loss": 0.2005,
+      "step": 8011
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1493018259935555e-05,
+      "loss": 0.1996,
+      "step": 8012
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1489437880415325e-05,
+      "loss": 0.1245,
+      "step": 8013
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1485857500895094e-05,
+      "loss": 0.148,
+      "step": 8014
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1482277121374868e-05,
+      "loss": 0.2493,
+      "step": 8015
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1478696741854637e-05,
+      "loss": 0.1562,
+      "step": 8016
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1475116362334407e-05,
+      "loss": 0.1787,
+      "step": 8017
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1471535982814177e-05,
+      "loss": 0.1296,
+      "step": 8018
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.146795560329395e-05,
+      "loss": 0.1697,
+      "step": 8019
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1464375223773723e-05,
+      "loss": 0.1766,
+      "step": 8020
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1460794844253493e-05,
+      "loss": 0.2406,
+      "step": 8021
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1457214464733263e-05,
+      "loss": 0.2598,
+      "step": 8022
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1453634085213033e-05,
+      "loss": 0.1565,
+      "step": 8023
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1450053705692806e-05,
+      "loss": 0.1515,
+      "step": 8024
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1446473326172576e-05,
+      "loss": 0.0769,
+      "step": 8025
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1442892946652346e-05,
+      "loss": 0.2825,
+      "step": 8026
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1439312567132115e-05,
+      "loss": 0.2627,
+      "step": 8027
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.143573218761189e-05,
+      "loss": 0.248,
+      "step": 8028
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.143215180809166e-05,
+      "loss": 0.2069,
+      "step": 8029
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.1144,
+      "step": 8030
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.14249910490512e-05,
+      "loss": 0.1167,
+      "step": 8031
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.142141066953097e-05,
+      "loss": 0.2543,
+      "step": 8032
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1417830290010744e-05,
+      "loss": 0.2534,
+      "step": 8033
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1414249910490514e-05,
+      "loss": 0.2109,
+      "step": 8034
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1410669530970284e-05,
+      "loss": 0.1969,
+      "step": 8035
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1407089151450054e-05,
+      "loss": 0.2483,
+      "step": 8036
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1403508771929827e-05,
+      "loss": 0.1186,
+      "step": 8037
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1399928392409597e-05,
+      "loss": 0.2563,
+      "step": 8038
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1396348012889367e-05,
+      "loss": 0.1342,
+      "step": 8039
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1392767633369136e-05,
+      "loss": 0.3949,
+      "step": 8040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.138918725384891e-05,
+      "loss": 0.269,
+      "step": 8041
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.138560687432868e-05,
+      "loss": 0.1182,
+      "step": 8042
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1382026494808452e-05,
+      "loss": 0.1893,
+      "step": 8043
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1378446115288222e-05,
+      "loss": 0.217,
+      "step": 8044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1374865735767992e-05,
+      "loss": 0.1573,
+      "step": 8045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1371285356247765e-05,
+      "loss": 0.1826,
+      "step": 8046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1367704976727535e-05,
+      "loss": 0.2895,
+      "step": 8047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1364124597207305e-05,
+      "loss": 0.2172,
+      "step": 8048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1360544217687075e-05,
+      "loss": 0.1394,
+      "step": 8049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1356963838166848e-05,
+      "loss": 0.2312,
+      "step": 8050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1353383458646618e-05,
+      "loss": 0.1404,
+      "step": 8051
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1349803079126387e-05,
+      "loss": 0.251,
+      "step": 8052
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1346222699606157e-05,
+      "loss": 0.1756,
+      "step": 8053
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.134264232008593e-05,
+      "loss": 0.1933,
+      "step": 8054
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1339061940565704e-05,
+      "loss": 0.1539,
+      "step": 8055
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1335481561045473e-05,
+      "loss": 0.1845,
+      "step": 8056
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1331901181525243e-05,
+      "loss": 0.1313,
+      "step": 8057
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1328320802005013e-05,
+      "loss": 0.1591,
+      "step": 8058
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1324740422484786e-05,
+      "loss": 0.2276,
+      "step": 8059
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1321160042964556e-05,
+      "loss": 0.2516,
+      "step": 8060
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1317579663444326e-05,
+      "loss": 0.1713,
+      "step": 8061
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1313999283924096e-05,
+      "loss": 0.165,
+      "step": 8062
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.131041890440387e-05,
+      "loss": 0.2587,
+      "step": 8063
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.130683852488364e-05,
+      "loss": 0.1362,
+      "step": 8064
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.130325814536341e-05,
+      "loss": 0.2675,
+      "step": 8065
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.129967776584318e-05,
+      "loss": 0.1705,
+      "step": 8066
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.129609738632295e-05,
+      "loss": 0.1822,
+      "step": 8067
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1292517006802725e-05,
+      "loss": 0.1969,
+      "step": 8068
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1288936627282494e-05,
+      "loss": 0.1216,
+      "step": 8069
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1285356247762264e-05,
+      "loss": 0.1862,
+      "step": 8070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1281775868242034e-05,
+      "loss": 0.2526,
+      "step": 8071
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1278195488721804e-05,
+      "loss": 0.1935,
+      "step": 8072
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1274615109201577e-05,
+      "loss": 0.3222,
+      "step": 8073
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1271034729681347e-05,
+      "loss": 0.218,
+      "step": 8074
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1267454350161117e-05,
+      "loss": 0.0938,
+      "step": 8075
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1263873970640886e-05,
+      "loss": 0.3481,
+      "step": 8076
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.126029359112066e-05,
+      "loss": 0.1627,
+      "step": 8077
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1256713211600433e-05,
+      "loss": 0.2662,
+      "step": 8078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1253132832080203e-05,
+      "loss": 0.2817,
+      "step": 8079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1249552452559972e-05,
+      "loss": 0.417,
+      "step": 8080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1245972073039742e-05,
+      "loss": 0.1638,
+      "step": 8081
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1242391693519515e-05,
+      "loss": 0.2218,
+      "step": 8082
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1238811313999285e-05,
+      "loss": 0.1062,
+      "step": 8083
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1235230934479055e-05,
+      "loss": 0.252,
+      "step": 8084
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1231650554958825e-05,
+      "loss": 0.3052,
+      "step": 8085
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.1228070175438598e-05,
+      "loss": 0.2318,
+      "step": 8086
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1224489795918368e-05,
+      "loss": 0.2403,
+      "step": 8087
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1220909416398137e-05,
+      "loss": 0.3469,
+      "step": 8088
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1217329036877907e-05,
+      "loss": 0.2069,
+      "step": 8089
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.121374865735768e-05,
+      "loss": 0.2733,
+      "step": 8090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1210168277837454e-05,
+      "loss": 0.2006,
+      "step": 8091
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1206587898317223e-05,
+      "loss": 0.2318,
+      "step": 8092
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1203007518796993e-05,
+      "loss": 0.3011,
+      "step": 8093
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1199427139276763e-05,
+      "loss": 0.1622,
+      "step": 8094
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1195846759756536e-05,
+      "loss": 0.1489,
+      "step": 8095
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1192266380236306e-05,
+      "loss": 0.2158,
+      "step": 8096
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1188686000716076e-05,
+      "loss": 0.282,
+      "step": 8097
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1185105621195846e-05,
+      "loss": 0.2502,
+      "step": 8098
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.118152524167562e-05,
+      "loss": 0.2433,
+      "step": 8099
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.117794486215539e-05,
+      "loss": 0.0694,
+      "step": 8100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1174364482635162e-05,
+      "loss": 0.0702,
+      "step": 8101
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.117078410311493e-05,
+      "loss": 0.3045,
+      "step": 8102
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.11672037235947e-05,
+      "loss": 0.2146,
+      "step": 8103
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1163623344074475e-05,
+      "loss": 0.26,
+      "step": 8104
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1160042964554244e-05,
+      "loss": 0.2336,
+      "step": 8105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1156462585034014e-05,
+      "loss": 0.194,
+      "step": 8106
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1152882205513784e-05,
+      "loss": 0.2714,
+      "step": 8107
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1149301825993557e-05,
+      "loss": 0.1424,
+      "step": 8108
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1145721446473327e-05,
+      "loss": 0.2467,
+      "step": 8109
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1142141066953097e-05,
+      "loss": 0.2202,
+      "step": 8110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1138560687432867e-05,
+      "loss": 0.3465,
+      "step": 8111
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.113498030791264e-05,
+      "loss": 0.192,
+      "step": 8112
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1131399928392413e-05,
+      "loss": 0.1903,
+      "step": 8113
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.1127819548872183e-05,
+      "loss": 0.2513,
+      "step": 8114
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1124239169351953e-05,
+      "loss": 0.3414,
+      "step": 8115
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1120658789831722e-05,
+      "loss": 0.178,
+      "step": 8116
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1117078410311496e-05,
+      "loss": 0.1159,
+      "step": 8117
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1113498030791265e-05,
+      "loss": 0.1873,
+      "step": 8118
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1109917651271035e-05,
+      "loss": 0.1711,
+      "step": 8119
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1106337271750805e-05,
+      "loss": 0.1929,
+      "step": 8120
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1102756892230578e-05,
+      "loss": 0.1342,
+      "step": 8121
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1099176512710348e-05,
+      "loss": 0.2324,
+      "step": 8122
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1095596133190118e-05,
+      "loss": 0.1329,
+      "step": 8123
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1092015753669888e-05,
+      "loss": 0.3009,
+      "step": 8124
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.108843537414966e-05,
+      "loss": 0.1443,
+      "step": 8125
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1084854994629434e-05,
+      "loss": 0.233,
+      "step": 8126
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1081274615109204e-05,
+      "loss": 0.3796,
+      "step": 8127
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1077694235588974e-05,
+      "loss": 0.1399,
+      "step": 8128
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1074113856068743e-05,
+      "loss": 0.1279,
+      "step": 8129
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1070533476548517e-05,
+      "loss": 0.1394,
+      "step": 8130
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1066953097028286e-05,
+      "loss": 0.1176,
+      "step": 8131
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1063372717508056e-05,
+      "loss": 0.3242,
+      "step": 8132
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1059792337987826e-05,
+      "loss": 0.2871,
+      "step": 8133
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.10562119584676e-05,
+      "loss": 0.3786,
+      "step": 8134
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.1694,
+      "step": 8135
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1049051199427142e-05,
+      "loss": 0.1833,
+      "step": 8136
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1045470819906912e-05,
+      "loss": 0.1608,
+      "step": 8137
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.104189044038668e-05,
+      "loss": 0.264,
+      "step": 8138
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1038310060866455e-05,
+      "loss": 0.1392,
+      "step": 8139
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1034729681346225e-05,
+      "loss": 0.2704,
+      "step": 8140
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1031149301825994e-05,
+      "loss": 0.0815,
+      "step": 8141
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 2.1027568922305764e-05,
+      "loss": 0.3089,
+      "step": 8142
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1023988542785537e-05,
+      "loss": 0.2257,
+      "step": 8143
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1020408163265307e-05,
+      "loss": 0.1498,
+      "step": 8144
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1016827783745077e-05,
+      "loss": 0.2641,
+      "step": 8145
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1013247404224847e-05,
+      "loss": 0.198,
+      "step": 8146
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.100966702470462e-05,
+      "loss": 0.2656,
+      "step": 8147
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1006086645184393e-05,
+      "loss": 0.3043,
+      "step": 8148
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.1002506265664163e-05,
+      "loss": 0.1526,
+      "step": 8149
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0998925886143933e-05,
+      "loss": 0.1671,
+      "step": 8150
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0995345506623703e-05,
+      "loss": 0.3554,
+      "step": 8151
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0991765127103476e-05,
+      "loss": 0.1372,
+      "step": 8152
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0988184747583246e-05,
+      "loss": 0.1659,
+      "step": 8153
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0984604368063015e-05,
+      "loss": 0.1843,
+      "step": 8154
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0981023988542785e-05,
+      "loss": 0.2475,
+      "step": 8155
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.097744360902256e-05,
+      "loss": 0.2267,
+      "step": 8156
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0973863229502328e-05,
+      "loss": 0.0806,
+      "step": 8157
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0970282849982098e-05,
+      "loss": 0.222,
+      "step": 8158
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0966702470461868e-05,
+      "loss": 0.2204,
+      "step": 8159
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.096312209094164e-05,
+      "loss": 0.2385,
+      "step": 8160
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0959541711421414e-05,
+      "loss": 0.2314,
+      "step": 8161
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0955961331901184e-05,
+      "loss": 0.1779,
+      "step": 8162
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0952380952380954e-05,
+      "loss": 0.229,
+      "step": 8163
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0948800572860724e-05,
+      "loss": 0.1377,
+      "step": 8164
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0945220193340497e-05,
+      "loss": 0.1429,
+      "step": 8165
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0941639813820267e-05,
+      "loss": 0.2975,
+      "step": 8166
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0938059434300036e-05,
+      "loss": 0.2036,
+      "step": 8167
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0934479054779806e-05,
+      "loss": 0.125,
+      "step": 8168
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.0930898675259576e-05,
+      "loss": 0.2085,
+      "step": 8169
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.092731829573935e-05,
+      "loss": 0.2052,
+      "step": 8170
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0923737916219122e-05,
+      "loss": 0.1659,
+      "step": 8171
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0920157536698892e-05,
+      "loss": 0.228,
+      "step": 8172
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0916577157178662e-05,
+      "loss": 0.2489,
+      "step": 8173
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0912996777658435e-05,
+      "loss": 0.3112,
+      "step": 8174
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0909416398138205e-05,
+      "loss": 0.2239,
+      "step": 8175
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0905836018617975e-05,
+      "loss": 0.1974,
+      "step": 8176
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0902255639097745e-05,
+      "loss": 0.2888,
+      "step": 8177
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0898675259577514e-05,
+      "loss": 0.1628,
+      "step": 8178
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0895094880057288e-05,
+      "loss": 0.1707,
+      "step": 8179
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0891514500537057e-05,
+      "loss": 0.3495,
+      "step": 8180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0887934121016827e-05,
+      "loss": 0.1867,
+      "step": 8181
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0884353741496597e-05,
+      "loss": 0.1481,
+      "step": 8182
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0880773361976373e-05,
+      "loss": 0.307,
+      "step": 8183
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0877192982456143e-05,
+      "loss": 0.2568,
+      "step": 8184
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0873612602935913e-05,
+      "loss": 0.229,
+      "step": 8185
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0870032223415683e-05,
+      "loss": 0.1964,
+      "step": 8186
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0866451843895453e-05,
+      "loss": 0.1836,
+      "step": 8187
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0862871464375226e-05,
+      "loss": 0.123,
+      "step": 8188
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0859291084854996e-05,
+      "loss": 0.1315,
+      "step": 8189
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0855710705334765e-05,
+      "loss": 0.0808,
+      "step": 8190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0852130325814535e-05,
+      "loss": 0.1662,
+      "step": 8191
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.084854994629431e-05,
+      "loss": 0.2048,
+      "step": 8192
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0844969566774078e-05,
+      "loss": 0.2692,
+      "step": 8193
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0841389187253848e-05,
+      "loss": 0.2069,
+      "step": 8194
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.083780880773362e-05,
+      "loss": 0.2619,
+      "step": 8195
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.083422842821339e-05,
+      "loss": 0.1869,
+      "step": 8196
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0830648048693164e-05,
+      "loss": 0.1136,
+      "step": 8197
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.0827067669172934e-05,
+      "loss": 0.1334,
+      "step": 8198
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0823487289652704e-05,
+      "loss": 0.2492,
+      "step": 8199
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0819906910132474e-05,
+      "loss": 0.2495,
+      "step": 8200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0816326530612247e-05,
+      "loss": 0.2137,
+      "step": 8201
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0812746151092017e-05,
+      "loss": 0.1722,
+      "step": 8202
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0809165771571786e-05,
+      "loss": 0.1066,
+      "step": 8203
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0805585392051556e-05,
+      "loss": 0.1777,
+      "step": 8204
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.080200501253133e-05,
+      "loss": 0.1519,
+      "step": 8205
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0798424633011103e-05,
+      "loss": 0.0967,
+      "step": 8206
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0794844253490872e-05,
+      "loss": 0.1528,
+      "step": 8207
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0791263873970642e-05,
+      "loss": 0.2033,
+      "step": 8208
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0787683494450412e-05,
+      "loss": 0.3542,
+      "step": 8209
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0784103114930185e-05,
+      "loss": 0.3753,
+      "step": 8210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0780522735409955e-05,
+      "loss": 0.1543,
+      "step": 8211
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0776942355889725e-05,
+      "loss": 0.2368,
+      "step": 8212
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0773361976369495e-05,
+      "loss": 0.1589,
+      "step": 8213
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0769781596849268e-05,
+      "loss": 0.2273,
+      "step": 8214
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0766201217329038e-05,
+      "loss": 0.2457,
+      "step": 8215
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0762620837808807e-05,
+      "loss": 0.3545,
+      "step": 8216
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0759040458288577e-05,
+      "loss": 0.1978,
+      "step": 8217
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.075546007876835e-05,
+      "loss": 0.2963,
+      "step": 8218
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0751879699248124e-05,
+      "loss": 0.1834,
+      "step": 8219
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0748299319727893e-05,
+      "loss": 0.1706,
+      "step": 8220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0744718940207663e-05,
+      "loss": 0.2334,
+      "step": 8221
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0741138560687433e-05,
+      "loss": 0.2326,
+      "step": 8222
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0737558181167206e-05,
+      "loss": 0.0836,
+      "step": 8223
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0733977801646976e-05,
+      "loss": 0.226,
+      "step": 8224
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0730397422126746e-05,
+      "loss": 0.2271,
+      "step": 8225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.0726817042606516e-05,
+      "loss": 0.2444,
+      "step": 8226
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.072323666308629e-05,
+      "loss": 0.107,
+      "step": 8227
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.071965628356606e-05,
+      "loss": 0.2102,
+      "step": 8228
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0716075904045828e-05,
+      "loss": 0.3973,
+      "step": 8229
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.07124955245256e-05,
+      "loss": 0.204,
+      "step": 8230
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.070891514500537e-05,
+      "loss": 0.2202,
+      "step": 8231
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0705334765485144e-05,
+      "loss": 0.2286,
+      "step": 8232
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0701754385964914e-05,
+      "loss": 0.213,
+      "step": 8233
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0698174006444684e-05,
+      "loss": 0.2369,
+      "step": 8234
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0694593626924454e-05,
+      "loss": 0.2803,
+      "step": 8235
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0691013247404227e-05,
+      "loss": 0.1663,
+      "step": 8236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0687432867883997e-05,
+      "loss": 0.1185,
+      "step": 8237
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0683852488363767e-05,
+      "loss": 0.3354,
+      "step": 8238
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0680272108843536e-05,
+      "loss": 0.3073,
+      "step": 8239
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.067669172932331e-05,
+      "loss": 0.1767,
+      "step": 8240
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.067311134980308e-05,
+      "loss": 0.1338,
+      "step": 8241
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0669530970282853e-05,
+      "loss": 0.3755,
+      "step": 8242
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0665950590762622e-05,
+      "loss": 0.3139,
+      "step": 8243
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0662370211242392e-05,
+      "loss": 0.2186,
+      "step": 8244
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0658789831722165e-05,
+      "loss": 0.2207,
+      "step": 8245
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0655209452201935e-05,
+      "loss": 0.1878,
+      "step": 8246
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0651629072681705e-05,
+      "loss": 0.2947,
+      "step": 8247
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0648048693161475e-05,
+      "loss": 0.3899,
+      "step": 8248
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0644468313641248e-05,
+      "loss": 0.2785,
+      "step": 8249
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0640887934121018e-05,
+      "loss": 0.1723,
+      "step": 8250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0637307554600788e-05,
+      "loss": 0.1601,
+      "step": 8251
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0633727175080557e-05,
+      "loss": 0.2375,
+      "step": 8252
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.063014679556033e-05,
+      "loss": 0.2607,
+      "step": 8253
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0626566416040104e-05,
+      "loss": 0.2907,
+      "step": 8254
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0622986036519874e-05,
+      "loss": 0.0933,
+      "step": 8255
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0619405656999643e-05,
+      "loss": 0.245,
+      "step": 8256
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0615825277479413e-05,
+      "loss": 0.2699,
+      "step": 8257
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0612244897959186e-05,
+      "loss": 0.1919,
+      "step": 8258
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0608664518438956e-05,
+      "loss": 0.1634,
+      "step": 8259
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0605084138918726e-05,
+      "loss": 0.1593,
+      "step": 8260
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0601503759398496e-05,
+      "loss": 0.2841,
+      "step": 8261
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.059792337987827e-05,
+      "loss": 0.1787,
+      "step": 8262
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.059434300035804e-05,
+      "loss": 0.1266,
+      "step": 8263
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.059076262083781e-05,
+      "loss": 0.2224,
+      "step": 8264
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0587182241317582e-05,
+      "loss": 0.124,
+      "step": 8265
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.058360186179735e-05,
+      "loss": 0.2059,
+      "step": 8266
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0580021482277125e-05,
+      "loss": 0.1923,
+      "step": 8267
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0576441102756895e-05,
+      "loss": 0.199,
+      "step": 8268
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0572860723236664e-05,
+      "loss": 0.3397,
+      "step": 8269
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0569280343716434e-05,
+      "loss": 0.1853,
+      "step": 8270
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0565699964196204e-05,
+      "loss": 0.1605,
+      "step": 8271
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0562119584675977e-05,
+      "loss": 0.2095,
+      "step": 8272
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0558539205155747e-05,
+      "loss": 0.2789,
+      "step": 8273
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0554958825635517e-05,
+      "loss": 0.107,
+      "step": 8274
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0551378446115287e-05,
+      "loss": 0.136,
+      "step": 8275
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.054779806659506e-05,
+      "loss": 0.2013,
+      "step": 8276
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0544217687074833e-05,
+      "loss": 0.1595,
+      "step": 8277
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0540637307554603e-05,
+      "loss": 0.2212,
+      "step": 8278
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0537056928034372e-05,
+      "loss": 0.2215,
+      "step": 8279
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0533476548514142e-05,
+      "loss": 0.1605,
+      "step": 8280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0529896168993915e-05,
+      "loss": 0.192,
+      "step": 8281
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 2.0526315789473685e-05,
+      "loss": 0.1793,
+      "step": 8282
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0522735409953455e-05,
+      "loss": 0.1802,
+      "step": 8283
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0519155030433225e-05,
+      "loss": 0.1317,
+      "step": 8284
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0515574650912998e-05,
+      "loss": 0.3058,
+      "step": 8285
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0511994271392768e-05,
+      "loss": 0.2315,
+      "step": 8286
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0508413891872538e-05,
+      "loss": 0.1619,
+      "step": 8287
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.050483351235231e-05,
+      "loss": 0.2899,
+      "step": 8288
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.050125313283208e-05,
+      "loss": 0.1443,
+      "step": 8289
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0497672753311854e-05,
+      "loss": 0.1098,
+      "step": 8290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0494092373791624e-05,
+      "loss": 0.4459,
+      "step": 8291
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0490511994271393e-05,
+      "loss": 0.3237,
+      "step": 8292
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0486931614751163e-05,
+      "loss": 0.2441,
+      "step": 8293
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0483351235230936e-05,
+      "loss": 0.2368,
+      "step": 8294
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0479770855710706e-05,
+      "loss": 0.2472,
+      "step": 8295
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0476190476190476e-05,
+      "loss": 0.1307,
+      "step": 8296
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0472610096670246e-05,
+      "loss": 0.2371,
+      "step": 8297
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.046902971715002e-05,
+      "loss": 0.2428,
+      "step": 8298
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.046544933762979e-05,
+      "loss": 0.163,
+      "step": 8299
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0461868958109562e-05,
+      "loss": 0.3097,
+      "step": 8300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0458288578589332e-05,
+      "loss": 0.2146,
+      "step": 8301
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.04547081990691e-05,
+      "loss": 0.257,
+      "step": 8302
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0451127819548875e-05,
+      "loss": 0.1287,
+      "step": 8303
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0447547440028645e-05,
+      "loss": 0.1685,
+      "step": 8304
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0443967060508414e-05,
+      "loss": 0.344,
+      "step": 8305
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0440386680988184e-05,
+      "loss": 0.1815,
+      "step": 8306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0436806301467957e-05,
+      "loss": 0.1164,
+      "step": 8307
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0433225921947727e-05,
+      "loss": 0.2159,
+      "step": 8308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0429645542427497e-05,
+      "loss": 0.1396,
+      "step": 8309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.0426065162907267e-05,
+      "loss": 0.2505,
+      "step": 8310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.042248478338704e-05,
+      "loss": 0.2433,
+      "step": 8311
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0418904403866813e-05,
+      "loss": 0.2897,
+      "step": 8312
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0415324024346583e-05,
+      "loss": 0.2126,
+      "step": 8313
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0411743644826353e-05,
+      "loss": 0.2469,
+      "step": 8314
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0408163265306123e-05,
+      "loss": 0.1686,
+      "step": 8315
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0404582885785896e-05,
+      "loss": 0.2689,
+      "step": 8316
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0401002506265666e-05,
+      "loss": 0.1822,
+      "step": 8317
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0397422126745435e-05,
+      "loss": 0.2858,
+      "step": 8318
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0393841747225205e-05,
+      "loss": 0.281,
+      "step": 8319
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0390261367704978e-05,
+      "loss": 0.2769,
+      "step": 8320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0386680988184748e-05,
+      "loss": 0.1595,
+      "step": 8321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0383100608664518e-05,
+      "loss": 0.1532,
+      "step": 8322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.037952022914429e-05,
+      "loss": 0.2503,
+      "step": 8323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.037593984962406e-05,
+      "loss": 0.1284,
+      "step": 8324
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0372359470103834e-05,
+      "loss": 0.0949,
+      "step": 8325
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0368779090583604e-05,
+      "loss": 0.2844,
+      "step": 8326
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0365198711063374e-05,
+      "loss": 0.128,
+      "step": 8327
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0361618331543143e-05,
+      "loss": 0.1603,
+      "step": 8328
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0358037952022917e-05,
+      "loss": 0.1034,
+      "step": 8329
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0354457572502686e-05,
+      "loss": 0.1565,
+      "step": 8330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0350877192982456e-05,
+      "loss": 0.1658,
+      "step": 8331
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0347296813462226e-05,
+      "loss": 0.2152,
+      "step": 8332
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0343716433942e-05,
+      "loss": 0.2098,
+      "step": 8333
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.034013605442177e-05,
+      "loss": 0.3264,
+      "step": 8334
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0336555674901542e-05,
+      "loss": 0.3053,
+      "step": 8335
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0332975295381312e-05,
+      "loss": 0.2383,
+      "step": 8336
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0329394915861082e-05,
+      "loss": 0.1798,
+      "step": 8337
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0325814536340855e-05,
+      "loss": 0.1467,
+      "step": 8338
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0322234156820625e-05,
+      "loss": 0.2481,
+      "step": 8339
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0318653777300395e-05,
+      "loss": 0.155,
+      "step": 8340
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0315073397780164e-05,
+      "loss": 0.2662,
+      "step": 8341
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0311493018259938e-05,
+      "loss": 0.2555,
+      "step": 8342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0307912638739707e-05,
+      "loss": 0.2289,
+      "step": 8343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0304332259219477e-05,
+      "loss": 0.2297,
+      "step": 8344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0300751879699247e-05,
+      "loss": 0.207,
+      "step": 8345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.029717150017902e-05,
+      "loss": 0.0617,
+      "step": 8346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0293591120658793e-05,
+      "loss": 0.1754,
+      "step": 8347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0290010741138563e-05,
+      "loss": 0.2045,
+      "step": 8348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0286430361618333e-05,
+      "loss": 0.238,
+      "step": 8349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0282849982098103e-05,
+      "loss": 0.1928,
+      "step": 8350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0279269602577876e-05,
+      "loss": 0.377,
+      "step": 8351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0275689223057646e-05,
+      "loss": 0.2015,
+      "step": 8352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0272108843537416e-05,
+      "loss": 0.0914,
+      "step": 8353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0268528464017185e-05,
+      "loss": 0.2243,
+      "step": 8354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.026494808449696e-05,
+      "loss": 0.1632,
+      "step": 8355
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.026136770497673e-05,
+      "loss": 0.231,
+      "step": 8356
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0257787325456498e-05,
+      "loss": 0.2313,
+      "step": 8357
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.025420694593627e-05,
+      "loss": 0.1437,
+      "step": 8358
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.025062656641604e-05,
+      "loss": 0.1821,
+      "step": 8359
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0247046186895814e-05,
+      "loss": 0.1498,
+      "step": 8360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0243465807375584e-05,
+      "loss": 0.2049,
+      "step": 8361
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0239885427855354e-05,
+      "loss": 0.1769,
+      "step": 8362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0236305048335124e-05,
+      "loss": 0.1522,
+      "step": 8363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0232724668814897e-05,
+      "loss": 0.101,
+      "step": 8364
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0229144289294667e-05,
+      "loss": 0.2078,
+      "step": 8365
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0225563909774437e-05,
+      "loss": 0.134,
+      "step": 8366
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0221983530254206e-05,
+      "loss": 0.2112,
+      "step": 8367
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0218403150733976e-05,
+      "loss": 0.0954,
+      "step": 8368
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.021482277121375e-05,
+      "loss": 0.0736,
+      "step": 8369
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0211242391693522e-05,
+      "loss": 0.2123,
+      "step": 8370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0207662012173292e-05,
+      "loss": 0.1868,
+      "step": 8371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0204081632653062e-05,
+      "loss": 0.1773,
+      "step": 8372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0200501253132835e-05,
+      "loss": 0.0815,
+      "step": 8373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0196920873612605e-05,
+      "loss": 0.1657,
+      "step": 8374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0193340494092375e-05,
+      "loss": 0.4061,
+      "step": 8375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0189760114572145e-05,
+      "loss": 0.1783,
+      "step": 8376
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0186179735051914e-05,
+      "loss": 0.2426,
+      "step": 8377
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0182599355531688e-05,
+      "loss": 0.2784,
+      "step": 8378
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0179018976011457e-05,
+      "loss": 0.1829,
+      "step": 8379
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0175438596491227e-05,
+      "loss": 0.2454,
+      "step": 8380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0171858216970997e-05,
+      "loss": 0.1722,
+      "step": 8381
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.016827783745077e-05,
+      "loss": 0.3276,
+      "step": 8382
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0164697457930543e-05,
+      "loss": 0.1726,
+      "step": 8383
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0161117078410313e-05,
+      "loss": 0.1686,
+      "step": 8384
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0157536698890083e-05,
+      "loss": 0.275,
+      "step": 8385
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0153956319369853e-05,
+      "loss": 0.187,
+      "step": 8386
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0150375939849626e-05,
+      "loss": 0.2066,
+      "step": 8387
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0146795560329396e-05,
+      "loss": 0.2551,
+      "step": 8388
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0143215180809166e-05,
+      "loss": 0.2679,
+      "step": 8389
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0139634801288935e-05,
+      "loss": 0.1104,
+      "step": 8390
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.013605442176871e-05,
+      "loss": 0.1959,
+      "step": 8391
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.013247404224848e-05,
+      "loss": 0.2162,
+      "step": 8392
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.012889366272825e-05,
+      "loss": 0.2755,
+      "step": 8393
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.012531328320802e-05,
+      "loss": 0.344,
+      "step": 8394
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.012173290368779e-05,
+      "loss": 0.2613,
+      "step": 8395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0118152524167564e-05,
+      "loss": 0.2434,
+      "step": 8396
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0114572144647334e-05,
+      "loss": 0.128,
+      "step": 8397
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0110991765127104e-05,
+      "loss": 0.1968,
+      "step": 8398
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0107411385606874e-05,
+      "loss": 0.1815,
+      "step": 8399
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0103831006086647e-05,
+      "loss": 0.2021,
+      "step": 8400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0100250626566417e-05,
+      "loss": 0.2463,
+      "step": 8401
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0096670247046187e-05,
+      "loss": 0.184,
+      "step": 8402
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0093089867525956e-05,
+      "loss": 0.2321,
+      "step": 8403
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.008950948800573e-05,
+      "loss": 0.1818,
+      "step": 8404
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0085929108485503e-05,
+      "loss": 0.1407,
+      "step": 8405
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0082348728965273e-05,
+      "loss": 0.2732,
+      "step": 8406
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0078768349445042e-05,
+      "loss": 0.1368,
+      "step": 8407
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0075187969924812e-05,
+      "loss": 0.2287,
+      "step": 8408
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0071607590404585e-05,
+      "loss": 0.3006,
+      "step": 8409
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.8700189541754934,
+      "eval_loss": 0.34804627299308777,
+      "eval_runtime": 24.8025,
+      "eval_samples_per_second": 1808.083,
+      "eval_steps_per_second": 28.263,
+      "step": 8409
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0068027210884355e-05,
+      "loss": 0.0934,
+      "step": 8410
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0064446831364125e-05,
+      "loss": 0.0943,
+      "step": 8411
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0060866451843895e-05,
+      "loss": 0.1536,
+      "step": 8412
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0057286072323668e-05,
+      "loss": 0.124,
+      "step": 8413
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0053705692803438e-05,
+      "loss": 0.0797,
+      "step": 8414
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0050125313283208e-05,
+      "loss": 0.1162,
+      "step": 8415
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0046544933762977e-05,
+      "loss": 0.1141,
+      "step": 8416
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.004296455424275e-05,
+      "loss": 0.1655,
+      "step": 8417
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0039384174722524e-05,
+      "loss": 0.2862,
+      "step": 8418
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0035803795202293e-05,
+      "loss": 0.0724,
+      "step": 8419
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0032223415682063e-05,
+      "loss": 0.097,
+      "step": 8420
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0028643036161833e-05,
+      "loss": 0.0726,
+      "step": 8421
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0025062656641606e-05,
+      "loss": 0.0622,
+      "step": 8422
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0021482277121376e-05,
+      "loss": 0.1601,
+      "step": 8423
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0017901897601146e-05,
+      "loss": 0.0705,
+      "step": 8424
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0014321518080916e-05,
+      "loss": 0.1158,
+      "step": 8425
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.001074113856069e-05,
+      "loss": 0.0865,
+      "step": 8426
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.000716075904046e-05,
+      "loss": 0.1333,
+      "step": 8427
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2.0003580379520232e-05,
+      "loss": 0.0979,
+      "step": 8428
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 2e-05,
+      "loss": 0.1107,
+      "step": 8429
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.999641962047977e-05,
+      "loss": 0.0692,
+      "step": 8430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9992839240959545e-05,
+      "loss": 0.0671,
+      "step": 8431
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9989258861439314e-05,
+      "loss": 0.0908,
+      "step": 8432
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9985678481919084e-05,
+      "loss": 0.1001,
+      "step": 8433
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9982098102398854e-05,
+      "loss": 0.0603,
+      "step": 8434
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9978517722878627e-05,
+      "loss": 0.2136,
+      "step": 8435
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9974937343358397e-05,
+      "loss": 0.1124,
+      "step": 8436
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9971356963838167e-05,
+      "loss": 0.0912,
+      "step": 8437
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9967776584317937e-05,
+      "loss": 0.0254,
+      "step": 8438
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.996419620479771e-05,
+      "loss": 0.2212,
+      "step": 8439
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9960615825277483e-05,
+      "loss": 0.1811,
+      "step": 8440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9957035445757253e-05,
+      "loss": 0.2139,
+      "step": 8441
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9953455066237023e-05,
+      "loss": 0.1242,
+      "step": 8442
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9949874686716792e-05,
+      "loss": 0.3167,
+      "step": 8443
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9946294307196566e-05,
+      "loss": 0.0443,
+      "step": 8444
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9942713927676335e-05,
+      "loss": 0.1985,
+      "step": 8445
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9939133548156105e-05,
+      "loss": 0.1018,
+      "step": 8446
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9935553168635875e-05,
+      "loss": 0.2013,
+      "step": 8447
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9931972789115648e-05,
+      "loss": 0.0826,
+      "step": 8448
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9928392409595418e-05,
+      "loss": 0.1334,
+      "step": 8449
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9924812030075188e-05,
+      "loss": 0.1504,
+      "step": 8450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.9921231650554958e-05,
+      "loss": 0.0335,
+      "step": 8451
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.991765127103473e-05,
+      "loss": 0.0992,
+      "step": 8452
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9914070891514504e-05,
+      "loss": 0.1093,
+      "step": 8453
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9910490511994274e-05,
+      "loss": 0.1044,
+      "step": 8454
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9906910132474044e-05,
+      "loss": 0.0254,
+      "step": 8455
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9903329752953813e-05,
+      "loss": 0.1709,
+      "step": 8456
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9899749373433587e-05,
+      "loss": 0.1548,
+      "step": 8457
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9896168993913356e-05,
+      "loss": 0.2945,
+      "step": 8458
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9892588614393126e-05,
+      "loss": 0.2372,
+      "step": 8459
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9889008234872896e-05,
+      "loss": 0.0818,
+      "step": 8460
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9885427855352666e-05,
+      "loss": 0.1169,
+      "step": 8461
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.988184747583244e-05,
+      "loss": 0.0963,
+      "step": 8462
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9878267096312212e-05,
+      "loss": 0.1587,
+      "step": 8463
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9874686716791982e-05,
+      "loss": 0.1684,
+      "step": 8464
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.987110633727175e-05,
+      "loss": 0.1294,
+      "step": 8465
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9867525957751525e-05,
+      "loss": 0.1113,
+      "step": 8466
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9863945578231295e-05,
+      "loss": 0.2498,
+      "step": 8467
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9860365198711064e-05,
+      "loss": 0.2387,
+      "step": 8468
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9856784819190834e-05,
+      "loss": 0.0532,
+      "step": 8469
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9853204439670604e-05,
+      "loss": 0.1035,
+      "step": 8470
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9849624060150377e-05,
+      "loss": 0.3041,
+      "step": 8471
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9846043680630147e-05,
+      "loss": 0.3456,
+      "step": 8472
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9842463301109917e-05,
+      "loss": 0.1097,
+      "step": 8473
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9838882921589687e-05,
+      "loss": 0.0688,
+      "step": 8474
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9835302542069463e-05,
+      "loss": 0.1922,
+      "step": 8475
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9831722162549233e-05,
+      "loss": 0.0695,
+      "step": 8476
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9828141783029003e-05,
+      "loss": 0.1222,
+      "step": 8477
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9824561403508773e-05,
+      "loss": 0.1003,
+      "step": 8478
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9820981023988542e-05,
+      "loss": 0.0755,
+      "step": 8479
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9817400644468316e-05,
+      "loss": 0.1108,
+      "step": 8480
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9813820264948085e-05,
+      "loss": 0.1374,
+      "step": 8481
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9810239885427855e-05,
+      "loss": 0.1939,
+      "step": 8482
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9806659505907625e-05,
+      "loss": 0.0609,
+      "step": 8483
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9803079126387398e-05,
+      "loss": 0.1409,
+      "step": 8484
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9799498746867168e-05,
+      "loss": 0.1315,
+      "step": 8485
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9795918367346938e-05,
+      "loss": 0.1925,
+      "step": 8486
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.979233798782671e-05,
+      "loss": 0.2051,
+      "step": 8487
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.978875760830648e-05,
+      "loss": 0.1613,
+      "step": 8488
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9785177228786254e-05,
+      "loss": 0.1048,
+      "step": 8489
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9781596849266024e-05,
+      "loss": 0.077,
+      "step": 8490
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9778016469745794e-05,
+      "loss": 0.1328,
+      "step": 8491
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9774436090225563e-05,
+      "loss": 0.1332,
+      "step": 8492
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9770855710705337e-05,
+      "loss": 0.1298,
+      "step": 8493
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9767275331185106e-05,
+      "loss": 0.1671,
+      "step": 8494
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9763694951664876e-05,
+      "loss": 0.0886,
+      "step": 8495
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9760114572144646e-05,
+      "loss": 0.152,
+      "step": 8496
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.975653419262442e-05,
+      "loss": 0.1802,
+      "step": 8497
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9752953813104192e-05,
+      "loss": 0.0932,
+      "step": 8498
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9749373433583962e-05,
+      "loss": 0.1189,
+      "step": 8499
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9745793054063732e-05,
+      "loss": 0.0871,
+      "step": 8500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9742212674543502e-05,
+      "loss": 0.1221,
+      "step": 8501
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9738632295023275e-05,
+      "loss": 0.2534,
+      "step": 8502
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9735051915503045e-05,
+      "loss": 0.1132,
+      "step": 8503
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9731471535982815e-05,
+      "loss": 0.1409,
+      "step": 8504
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9727891156462584e-05,
+      "loss": 0.1124,
+      "step": 8505
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9724310776942358e-05,
+      "loss": 0.1084,
+      "step": 8506
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.9720730397422127e-05,
+      "loss": 0.0548,
+      "step": 8507
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9717150017901897e-05,
+      "loss": 0.0786,
+      "step": 8508
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9713569638381667e-05,
+      "loss": 0.116,
+      "step": 8509
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.970998925886144e-05,
+      "loss": 0.2278,
+      "step": 8510
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9706408879341213e-05,
+      "loss": 0.0708,
+      "step": 8511
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9702828499820983e-05,
+      "loss": 0.1874,
+      "step": 8512
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9699248120300753e-05,
+      "loss": 0.0879,
+      "step": 8513
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9695667740780523e-05,
+      "loss": 0.1272,
+      "step": 8514
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9692087361260296e-05,
+      "loss": 0.0935,
+      "step": 8515
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9688506981740066e-05,
+      "loss": 0.1057,
+      "step": 8516
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9684926602219835e-05,
+      "loss": 0.0806,
+      "step": 8517
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9681346222699605e-05,
+      "loss": 0.0512,
+      "step": 8518
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.967776584317938e-05,
+      "loss": 0.0922,
+      "step": 8519
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9674185463659148e-05,
+      "loss": 0.0739,
+      "step": 8520
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9670605084138918e-05,
+      "loss": 0.1583,
+      "step": 8521
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.966702470461869e-05,
+      "loss": 0.2099,
+      "step": 8522
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.966344432509846e-05,
+      "loss": 0.2814,
+      "step": 8523
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9659863945578234e-05,
+      "loss": 0.1228,
+      "step": 8524
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9656283566058004e-05,
+      "loss": 0.0838,
+      "step": 8525
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9652703186537774e-05,
+      "loss": 0.0706,
+      "step": 8526
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.2542,
+      "step": 8527
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9645542427497317e-05,
+      "loss": 0.0771,
+      "step": 8528
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9641962047977087e-05,
+      "loss": 0.0835,
+      "step": 8529
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9638381668456856e-05,
+      "loss": 0.0463,
+      "step": 8530
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9634801288936626e-05,
+      "loss": 0.0613,
+      "step": 8531
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.96312209094164e-05,
+      "loss": 0.2075,
+      "step": 8532
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9627640529896173e-05,
+      "loss": 0.0367,
+      "step": 8533
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9624060150375942e-05,
+      "loss": 0.1192,
+      "step": 8534
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.9620479770855712e-05,
+      "loss": 0.2535,
+      "step": 8535
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9616899391335482e-05,
+      "loss": 0.3132,
+      "step": 8536
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9613319011815255e-05,
+      "loss": 0.0283,
+      "step": 8537
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9609738632295025e-05,
+      "loss": 0.1172,
+      "step": 8538
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9606158252774795e-05,
+      "loss": 0.0448,
+      "step": 8539
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9602577873254565e-05,
+      "loss": 0.1888,
+      "step": 8540
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9598997493734338e-05,
+      "loss": 0.2487,
+      "step": 8541
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9595417114214108e-05,
+      "loss": 0.0799,
+      "step": 8542
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 0.2016,
+      "step": 8543
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9588256355173647e-05,
+      "loss": 0.2776,
+      "step": 8544
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.958467597565342e-05,
+      "loss": 0.2334,
+      "step": 8545
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9581095596133194e-05,
+      "loss": 0.3564,
+      "step": 8546
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9577515216612963e-05,
+      "loss": 0.2065,
+      "step": 8547
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9573934837092733e-05,
+      "loss": 0.1602,
+      "step": 8548
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9570354457572503e-05,
+      "loss": 0.2151,
+      "step": 8549
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9566774078052276e-05,
+      "loss": 0.1109,
+      "step": 8550
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9563193698532046e-05,
+      "loss": 0.2318,
+      "step": 8551
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9559613319011816e-05,
+      "loss": 0.132,
+      "step": 8552
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9556032939491586e-05,
+      "loss": 0.145,
+      "step": 8553
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.955245255997136e-05,
+      "loss": 0.1145,
+      "step": 8554
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.954887218045113e-05,
+      "loss": 0.0726,
+      "step": 8555
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9545291800930898e-05,
+      "loss": 0.1913,
+      "step": 8556
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.954171142141067e-05,
+      "loss": 0.0863,
+      "step": 8557
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.953813104189044e-05,
+      "loss": 0.1403,
+      "step": 8558
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9534550662370214e-05,
+      "loss": 0.1645,
+      "step": 8559
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9530970282849984e-05,
+      "loss": 0.0882,
+      "step": 8560
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9527389903329754e-05,
+      "loss": 0.2329,
+      "step": 8561
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9523809523809524e-05,
+      "loss": 0.1269,
+      "step": 8562
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.9520229144289297e-05,
+      "loss": 0.0731,
+      "step": 8563
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9516648764769067e-05,
+      "loss": 0.205,
+      "step": 8564
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9513068385248837e-05,
+      "loss": 0.1245,
+      "step": 8565
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9509488005728606e-05,
+      "loss": 0.1501,
+      "step": 8566
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9505907626208376e-05,
+      "loss": 0.1568,
+      "step": 8567
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9502327246688153e-05,
+      "loss": 0.2141,
+      "step": 8568
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9498746867167923e-05,
+      "loss": 0.0706,
+      "step": 8569
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9495166487647692e-05,
+      "loss": 0.112,
+      "step": 8570
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9491586108127462e-05,
+      "loss": 0.2021,
+      "step": 8571
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9488005728607232e-05,
+      "loss": 0.2201,
+      "step": 8572
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9484425349087005e-05,
+      "loss": 0.1113,
+      "step": 8573
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9480844969566775e-05,
+      "loss": 0.1283,
+      "step": 8574
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9477264590046545e-05,
+      "loss": 0.1584,
+      "step": 8575
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9473684210526315e-05,
+      "loss": 0.1907,
+      "step": 8576
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9470103831006088e-05,
+      "loss": 0.0525,
+      "step": 8577
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9466523451485858e-05,
+      "loss": 0.0634,
+      "step": 8578
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9462943071965627e-05,
+      "loss": 0.1808,
+      "step": 8579
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.94593626924454e-05,
+      "loss": 0.209,
+      "step": 8580
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.945578231292517e-05,
+      "loss": 0.2738,
+      "step": 8581
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9452201933404944e-05,
+      "loss": 0.1136,
+      "step": 8582
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9448621553884713e-05,
+      "loss": 0.236,
+      "step": 8583
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9445041174364483e-05,
+      "loss": 0.2086,
+      "step": 8584
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9441460794844253e-05,
+      "loss": 0.2503,
+      "step": 8585
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9437880415324026e-05,
+      "loss": 0.1166,
+      "step": 8586
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9434300035803796e-05,
+      "loss": 0.195,
+      "step": 8587
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9430719656283566e-05,
+      "loss": 0.0519,
+      "step": 8588
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9427139276763336e-05,
+      "loss": 0.1438,
+      "step": 8589
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.942355889724311e-05,
+      "loss": 0.1988,
+      "step": 8590
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.941997851772288e-05,
+      "loss": 0.2255,
+      "step": 8591
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9416398138202652e-05,
+      "loss": 0.1078,
+      "step": 8592
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.941281775868242e-05,
+      "loss": 0.1097,
+      "step": 8593
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.940923737916219e-05,
+      "loss": 0.1181,
+      "step": 8594
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9405656999641965e-05,
+      "loss": 0.1019,
+      "step": 8595
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9402076620121734e-05,
+      "loss": 0.2052,
+      "step": 8596
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9398496240601504e-05,
+      "loss": 0.1599,
+      "step": 8597
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9394915861081274e-05,
+      "loss": 0.0941,
+      "step": 8598
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9391335481561047e-05,
+      "loss": 0.1224,
+      "step": 8599
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9387755102040817e-05,
+      "loss": 0.1003,
+      "step": 8600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9384174722520587e-05,
+      "loss": 0.1575,
+      "step": 8601
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9380594343000357e-05,
+      "loss": 0.1147,
+      "step": 8602
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.937701396348013e-05,
+      "loss": 0.1318,
+      "step": 8603
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9373433583959903e-05,
+      "loss": 0.1089,
+      "step": 8604
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9369853204439673e-05,
+      "loss": 0.2299,
+      "step": 8605
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9366272824919442e-05,
+      "loss": 0.1399,
+      "step": 8606
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9362692445399212e-05,
+      "loss": 0.0295,
+      "step": 8607
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9359112065878985e-05,
+      "loss": 0.2056,
+      "step": 8608
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9355531686358755e-05,
+      "loss": 0.0917,
+      "step": 8609
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9351951306838525e-05,
+      "loss": 0.1042,
+      "step": 8610
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9348370927318295e-05,
+      "loss": 0.228,
+      "step": 8611
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9344790547798068e-05,
+      "loss": 0.1681,
+      "step": 8612
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9341210168277838e-05,
+      "loss": 0.0652,
+      "step": 8613
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9337629788757608e-05,
+      "loss": 0.0714,
+      "step": 8614
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.933404940923738e-05,
+      "loss": 0.0959,
+      "step": 8615
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.933046902971715e-05,
+      "loss": 0.1834,
+      "step": 8616
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9326888650196924e-05,
+      "loss": 0.1444,
+      "step": 8617
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9323308270676694e-05,
+      "loss": 0.2006,
+      "step": 8618
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.9319727891156463e-05,
+      "loss": 0.0616,
+      "step": 8619
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9316147511636233e-05,
+      "loss": 0.084,
+      "step": 8620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9312567132116006e-05,
+      "loss": 0.1164,
+      "step": 8621
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9308986752595776e-05,
+      "loss": 0.0748,
+      "step": 8622
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9305406373075546e-05,
+      "loss": 0.1145,
+      "step": 8623
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9301825993555316e-05,
+      "loss": 0.1954,
+      "step": 8624
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.0558,
+      "step": 8625
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.929466523451486e-05,
+      "loss": 0.2134,
+      "step": 8626
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9291084854994632e-05,
+      "loss": 0.1456,
+      "step": 8627
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9287504475474402e-05,
+      "loss": 0.2433,
+      "step": 8628
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.928392409595417e-05,
+      "loss": 0.1548,
+      "step": 8629
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9280343716433945e-05,
+      "loss": 0.1454,
+      "step": 8630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9276763336913715e-05,
+      "loss": 0.1459,
+      "step": 8631
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9273182957393484e-05,
+      "loss": 0.1845,
+      "step": 8632
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9269602577873254e-05,
+      "loss": 0.2438,
+      "step": 8633
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9266022198353027e-05,
+      "loss": 0.2154,
+      "step": 8634
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9262441818832797e-05,
+      "loss": 0.0674,
+      "step": 8635
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9258861439312567e-05,
+      "loss": 0.1795,
+      "step": 8636
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9255281059792337e-05,
+      "loss": 0.0884,
+      "step": 8637
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.925170068027211e-05,
+      "loss": 0.1487,
+      "step": 8638
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9248120300751883e-05,
+      "loss": 0.1197,
+      "step": 8639
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9244539921231653e-05,
+      "loss": 0.1389,
+      "step": 8640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9240959541711423e-05,
+      "loss": 0.063,
+      "step": 8641
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9237379162191193e-05,
+      "loss": 0.1836,
+      "step": 8642
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9233798782670966e-05,
+      "loss": 0.1311,
+      "step": 8643
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9230218403150736e-05,
+      "loss": 0.0551,
+      "step": 8644
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9226638023630505e-05,
+      "loss": 0.1889,
+      "step": 8645
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9223057644110275e-05,
+      "loss": 0.1633,
+      "step": 8646
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.9219477264590048e-05,
+      "loss": 0.1968,
+      "step": 8647
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9215896885069818e-05,
+      "loss": 0.3294,
+      "step": 8648
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9212316505549588e-05,
+      "loss": 0.069,
+      "step": 8649
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.920873612602936e-05,
+      "loss": 0.1912,
+      "step": 8650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.920515574650913e-05,
+      "loss": 0.0743,
+      "step": 8651
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9201575366988904e-05,
+      "loss": 0.1425,
+      "step": 8652
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9197994987468674e-05,
+      "loss": 0.0393,
+      "step": 8653
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9194414607948444e-05,
+      "loss": 0.0492,
+      "step": 8654
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9190834228428213e-05,
+      "loss": 0.1211,
+      "step": 8655
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9187253848907987e-05,
+      "loss": 0.1324,
+      "step": 8656
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 0.0936,
+      "step": 8657
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9180093089867526e-05,
+      "loss": 0.0606,
+      "step": 8658
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9176512710347296e-05,
+      "loss": 0.2284,
+      "step": 8659
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9172932330827066e-05,
+      "loss": 0.3295,
+      "step": 8660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.916935195130684e-05,
+      "loss": 0.2013,
+      "step": 8661
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9165771571786612e-05,
+      "loss": 0.1372,
+      "step": 8662
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9162191192266382e-05,
+      "loss": 0.1425,
+      "step": 8663
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9158610812746152e-05,
+      "loss": 0.1498,
+      "step": 8664
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9155030433225925e-05,
+      "loss": 0.0516,
+      "step": 8665
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9151450053705695e-05,
+      "loss": 0.1542,
+      "step": 8666
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9147869674185465e-05,
+      "loss": 0.1848,
+      "step": 8667
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9144289294665234e-05,
+      "loss": 0.1209,
+      "step": 8668
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9140708915145004e-05,
+      "loss": 0.1141,
+      "step": 8669
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9137128535624777e-05,
+      "loss": 0.1692,
+      "step": 8670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9133548156104547e-05,
+      "loss": 0.1013,
+      "step": 8671
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9129967776584317e-05,
+      "loss": 0.1084,
+      "step": 8672
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9126387397064087e-05,
+      "loss": 0.1432,
+      "step": 8673
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9122807017543863e-05,
+      "loss": 0.1147,
+      "step": 8674
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.9119226638023633e-05,
+      "loss": 0.193,
+      "step": 8675
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9115646258503403e-05,
+      "loss": 0.0912,
+      "step": 8676
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9112065878983173e-05,
+      "loss": 0.1306,
+      "step": 8677
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9108485499462943e-05,
+      "loss": 0.2213,
+      "step": 8678
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9104905119942716e-05,
+      "loss": 0.1503,
+      "step": 8679
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9101324740422486e-05,
+      "loss": 0.1238,
+      "step": 8680
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9097744360902255e-05,
+      "loss": 0.1113,
+      "step": 8681
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9094163981382025e-05,
+      "loss": 0.1457,
+      "step": 8682
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.90905836018618e-05,
+      "loss": 0.1125,
+      "step": 8683
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9087003222341568e-05,
+      "loss": 0.1598,
+      "step": 8684
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.908342284282134e-05,
+      "loss": 0.0918,
+      "step": 8685
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.907984246330111e-05,
+      "loss": 0.0853,
+      "step": 8686
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.907626208378088e-05,
+      "loss": 0.1531,
+      "step": 8687
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9072681704260654e-05,
+      "loss": 0.0546,
+      "step": 8688
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9069101324740424e-05,
+      "loss": 0.149,
+      "step": 8689
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9065520945220194e-05,
+      "loss": 0.0554,
+      "step": 8690
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9061940565699964e-05,
+      "loss": 0.0553,
+      "step": 8691
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9058360186179737e-05,
+      "loss": 0.1808,
+      "step": 8692
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9054779806659507e-05,
+      "loss": 0.1909,
+      "step": 8693
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9051199427139276e-05,
+      "loss": 0.2037,
+      "step": 8694
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.0747,
+      "step": 8695
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.904403866809882e-05,
+      "loss": 0.2204,
+      "step": 8696
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9040458288578592e-05,
+      "loss": 0.1359,
+      "step": 8697
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9036877909058362e-05,
+      "loss": 0.1506,
+      "step": 8698
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9033297529538132e-05,
+      "loss": 0.0807,
+      "step": 8699
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9029717150017902e-05,
+      "loss": 0.1339,
+      "step": 8700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9026136770497675e-05,
+      "loss": 0.0506,
+      "step": 8701
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9022556390977445e-05,
+      "loss": 0.1692,
+      "step": 8702
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.9018976011457215e-05,
+      "loss": 0.1458,
+      "step": 8703
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9015395631936984e-05,
+      "loss": 0.0788,
+      "step": 8704
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9011815252416758e-05,
+      "loss": 0.198,
+      "step": 8705
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9008234872896527e-05,
+      "loss": 0.2646,
+      "step": 8706
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9004654493376297e-05,
+      "loss": 0.1285,
+      "step": 8707
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.9001074113856067e-05,
+      "loss": 0.1328,
+      "step": 8708
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.899749373433584e-05,
+      "loss": 0.1377,
+      "step": 8709
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8993913354815613e-05,
+      "loss": 0.0694,
+      "step": 8710
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8990332975295383e-05,
+      "loss": 0.1089,
+      "step": 8711
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8986752595775153e-05,
+      "loss": 0.0738,
+      "step": 8712
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8983172216254923e-05,
+      "loss": 0.2036,
+      "step": 8713
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 0.1227,
+      "step": 8714
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8976011457214466e-05,
+      "loss": 0.2177,
+      "step": 8715
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8972431077694236e-05,
+      "loss": 0.1221,
+      "step": 8716
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8968850698174005e-05,
+      "loss": 0.2364,
+      "step": 8717
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.896527031865378e-05,
+      "loss": 0.1219,
+      "step": 8718
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.896168993913355e-05,
+      "loss": 0.1152,
+      "step": 8719
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.895810955961332e-05,
+      "loss": 0.0854,
+      "step": 8720
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.895452918009309e-05,
+      "loss": 0.1127,
+      "step": 8721
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.895094880057286e-05,
+      "loss": 0.1115,
+      "step": 8722
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8947368421052634e-05,
+      "loss": 0.1083,
+      "step": 8723
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8943788041532404e-05,
+      "loss": 0.183,
+      "step": 8724
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8940207662012174e-05,
+      "loss": 0.1462,
+      "step": 8725
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8936627282491944e-05,
+      "loss": 0.1615,
+      "step": 8726
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8933046902971717e-05,
+      "loss": 0.0676,
+      "step": 8727
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8929466523451487e-05,
+      "loss": 0.1049,
+      "step": 8728
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8925886143931257e-05,
+      "loss": 0.0735,
+      "step": 8729
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8922305764411026e-05,
+      "loss": 0.1657,
+      "step": 8730
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.89187253848908e-05,
+      "loss": 0.264,
+      "step": 8731
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8915145005370573e-05,
+      "loss": 0.0373,
+      "step": 8732
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8911564625850343e-05,
+      "loss": 0.1172,
+      "step": 8733
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8907984246330112e-05,
+      "loss": 0.1224,
+      "step": 8734
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8904403866809882e-05,
+      "loss": 0.0654,
+      "step": 8735
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8900823487289655e-05,
+      "loss": 0.0459,
+      "step": 8736
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8897243107769425e-05,
+      "loss": 0.2581,
+      "step": 8737
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8893662728249195e-05,
+      "loss": 0.1495,
+      "step": 8738
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8890082348728965e-05,
+      "loss": 0.2417,
+      "step": 8739
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8886501969208738e-05,
+      "loss": 0.2094,
+      "step": 8740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8882921589688508e-05,
+      "loss": 0.2319,
+      "step": 8741
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8879341210168278e-05,
+      "loss": 0.0977,
+      "step": 8742
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8875760830648047e-05,
+      "loss": 0.1275,
+      "step": 8743
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.887218045112782e-05,
+      "loss": 0.1983,
+      "step": 8744
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8868600071607594e-05,
+      "loss": 0.3127,
+      "step": 8745
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8865019692087363e-05,
+      "loss": 0.2227,
+      "step": 8746
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8861439312567133e-05,
+      "loss": 0.1941,
+      "step": 8747
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8857858933046903e-05,
+      "loss": 0.0937,
+      "step": 8748
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8854278553526676e-05,
+      "loss": 0.1619,
+      "step": 8749
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8850698174006446e-05,
+      "loss": 0.1415,
+      "step": 8750
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8847117794486216e-05,
+      "loss": 0.1694,
+      "step": 8751
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8843537414965986e-05,
+      "loss": 0.179,
+      "step": 8752
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.883995703544576e-05,
+      "loss": 0.1251,
+      "step": 8753
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.883637665592553e-05,
+      "loss": 0.1782,
+      "step": 8754
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8832796276405302e-05,
+      "loss": 0.0728,
+      "step": 8755
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.882921589688507e-05,
+      "loss": 0.0978,
+      "step": 8756
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.882563551736484e-05,
+      "loss": 0.227,
+      "step": 8757
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8822055137844615e-05,
+      "loss": 0.1986,
+      "step": 8758
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.8818474758324384e-05,
+      "loss": 0.1143,
+      "step": 8759
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8814894378804154e-05,
+      "loss": 0.1162,
+      "step": 8760
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8811313999283924e-05,
+      "loss": 0.1575,
+      "step": 8761
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8807733619763697e-05,
+      "loss": 0.2827,
+      "step": 8762
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8804153240243467e-05,
+      "loss": 0.0885,
+      "step": 8763
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8800572860723237e-05,
+      "loss": 0.1557,
+      "step": 8764
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8796992481203007e-05,
+      "loss": 0.1079,
+      "step": 8765
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8793412101682776e-05,
+      "loss": 0.1457,
+      "step": 8766
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8789831722162553e-05,
+      "loss": 0.2065,
+      "step": 8767
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8786251342642323e-05,
+      "loss": 0.1624,
+      "step": 8768
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8782670963122093e-05,
+      "loss": 0.1784,
+      "step": 8769
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8779090583601862e-05,
+      "loss": 0.1056,
+      "step": 8770
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8775510204081632e-05,
+      "loss": 0.1647,
+      "step": 8771
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8771929824561405e-05,
+      "loss": 0.1338,
+      "step": 8772
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8768349445041175e-05,
+      "loss": 0.1345,
+      "step": 8773
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8764769065520945e-05,
+      "loss": 0.1594,
+      "step": 8774
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8761188686000715e-05,
+      "loss": 0.0949,
+      "step": 8775
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8757608306480488e-05,
+      "loss": 0.0767,
+      "step": 8776
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8754027926960258e-05,
+      "loss": 0.0449,
+      "step": 8777
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8750447547440028e-05,
+      "loss": 0.1562,
+      "step": 8778
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.87468671679198e-05,
+      "loss": 0.0817,
+      "step": 8779
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.874328678839957e-05,
+      "loss": 0.1289,
+      "step": 8780
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8739706408879344e-05,
+      "loss": 0.0977,
+      "step": 8781
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8736126029359114e-05,
+      "loss": 0.098,
+      "step": 8782
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8732545649838883e-05,
+      "loss": 0.1888,
+      "step": 8783
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8728965270318653e-05,
+      "loss": 0.0791,
+      "step": 8784
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8725384890798426e-05,
+      "loss": 0.2275,
+      "step": 8785
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8721804511278196e-05,
+      "loss": 0.1175,
+      "step": 8786
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.8718224131757966e-05,
+      "loss": 0.2825,
+      "step": 8787
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8714643752237736e-05,
+      "loss": 0.2241,
+      "step": 8788
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.871106337271751e-05,
+      "loss": 0.0389,
+      "step": 8789
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8707482993197282e-05,
+      "loss": 0.2415,
+      "step": 8790
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8703902613677052e-05,
+      "loss": 0.0527,
+      "step": 8791
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.870032223415682e-05,
+      "loss": 0.2082,
+      "step": 8792
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.869674185463659e-05,
+      "loss": 0.0842,
+      "step": 8793
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8693161475116365e-05,
+      "loss": 0.0507,
+      "step": 8794
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8689581095596134e-05,
+      "loss": 0.1562,
+      "step": 8795
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8686000716075904e-05,
+      "loss": 0.1299,
+      "step": 8796
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8682420336555674e-05,
+      "loss": 0.0964,
+      "step": 8797
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8678839957035447e-05,
+      "loss": 0.2519,
+      "step": 8798
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8675259577515217e-05,
+      "loss": 0.1701,
+      "step": 8799
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8671679197994987e-05,
+      "loss": 0.0843,
+      "step": 8800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8668098818474757e-05,
+      "loss": 0.183,
+      "step": 8801
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.866451843895453e-05,
+      "loss": 0.0846,
+      "step": 8802
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8660938059434303e-05,
+      "loss": 0.1892,
+      "step": 8803
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8657357679914073e-05,
+      "loss": 0.0922,
+      "step": 8804
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8653777300393843e-05,
+      "loss": 0.1081,
+      "step": 8805
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8650196920873612e-05,
+      "loss": 0.1094,
+      "step": 8806
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8646616541353386e-05,
+      "loss": 0.08,
+      "step": 8807
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8643036161833155e-05,
+      "loss": 0.1741,
+      "step": 8808
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8639455782312925e-05,
+      "loss": 0.1064,
+      "step": 8809
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8635875402792695e-05,
+      "loss": 0.0516,
+      "step": 8810
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8632295023272468e-05,
+      "loss": 0.0899,
+      "step": 8811
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8628714643752238e-05,
+      "loss": 0.203,
+      "step": 8812
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8625134264232008e-05,
+      "loss": 0.0811,
+      "step": 8813
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.862155388471178e-05,
+      "loss": 0.1313,
+      "step": 8814
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.861797350519155e-05,
+      "loss": 0.0567,
+      "step": 8815
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8614393125671324e-05,
+      "loss": 0.2833,
+      "step": 8816
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8610812746151094e-05,
+      "loss": 0.0944,
+      "step": 8817
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8607232366630864e-05,
+      "loss": 0.1881,
+      "step": 8818
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8603651987110633e-05,
+      "loss": 0.0662,
+      "step": 8819
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8600071607590407e-05,
+      "loss": 0.166,
+      "step": 8820
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.0974,
+      "step": 8821
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8592910848549946e-05,
+      "loss": 0.1431,
+      "step": 8822
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8589330469029716e-05,
+      "loss": 0.268,
+      "step": 8823
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.858575008950949e-05,
+      "loss": 0.1416,
+      "step": 8824
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8582169709989262e-05,
+      "loss": 0.1098,
+      "step": 8825
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8578589330469032e-05,
+      "loss": 0.2475,
+      "step": 8826
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8575008950948802e-05,
+      "loss": 0.3034,
+      "step": 8827
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8571428571428572e-05,
+      "loss": 0.2505,
+      "step": 8828
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8567848191908345e-05,
+      "loss": 0.2831,
+      "step": 8829
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8564267812388115e-05,
+      "loss": 0.1455,
+      "step": 8830
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8560687432867885e-05,
+      "loss": 0.139,
+      "step": 8831
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8557107053347654e-05,
+      "loss": 0.1209,
+      "step": 8832
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8553526673827428e-05,
+      "loss": 0.1488,
+      "step": 8833
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8549946294307197e-05,
+      "loss": 0.1173,
+      "step": 8834
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8546365914786967e-05,
+      "loss": 0.0902,
+      "step": 8835
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8542785535266737e-05,
+      "loss": 0.1879,
+      "step": 8836
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.853920515574651e-05,
+      "loss": 0.2327,
+      "step": 8837
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8535624776226283e-05,
+      "loss": 0.1501,
+      "step": 8838
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8532044396706053e-05,
+      "loss": 0.1754,
+      "step": 8839
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8528464017185823e-05,
+      "loss": 0.0944,
+      "step": 8840
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8524883637665593e-05,
+      "loss": 0.1257,
+      "step": 8841
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8521303258145366e-05,
+      "loss": 0.1658,
+      "step": 8842
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.8517722878625136e-05,
+      "loss": 0.0704,
+      "step": 8843
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8514142499104905e-05,
+      "loss": 0.101,
+      "step": 8844
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8510562119584675e-05,
+      "loss": 0.08,
+      "step": 8845
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.850698174006445e-05,
+      "loss": 0.0532,
+      "step": 8846
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8503401360544218e-05,
+      "loss": 0.1024,
+      "step": 8847
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8499820981023988e-05,
+      "loss": 0.176,
+      "step": 8848
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.849624060150376e-05,
+      "loss": 0.1319,
+      "step": 8849
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.849266022198353e-05,
+      "loss": 0.1643,
+      "step": 8850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8489079842463304e-05,
+      "loss": 0.1265,
+      "step": 8851
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8485499462943074e-05,
+      "loss": 0.1689,
+      "step": 8852
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8481919083422844e-05,
+      "loss": 0.1311,
+      "step": 8853
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8478338703902614e-05,
+      "loss": 0.1604,
+      "step": 8854
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8474758324382387e-05,
+      "loss": 0.1354,
+      "step": 8855
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8471177944862157e-05,
+      "loss": 0.1028,
+      "step": 8856
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8467597565341926e-05,
+      "loss": 0.2223,
+      "step": 8857
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8464017185821696e-05,
+      "loss": 0.1495,
+      "step": 8858
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8460436806301466e-05,
+      "loss": 0.0669,
+      "step": 8859
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8456856426781243e-05,
+      "loss": 0.1609,
+      "step": 8860
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8453276047261012e-05,
+      "loss": 0.1666,
+      "step": 8861
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8449695667740782e-05,
+      "loss": 0.1993,
+      "step": 8862
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8446115288220552e-05,
+      "loss": 0.1789,
+      "step": 8863
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8442534908700325e-05,
+      "loss": 0.2524,
+      "step": 8864
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8438954529180095e-05,
+      "loss": 0.0722,
+      "step": 8865
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8435374149659865e-05,
+      "loss": 0.1289,
+      "step": 8866
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8431793770139635e-05,
+      "loss": 0.1066,
+      "step": 8867
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8428213390619404e-05,
+      "loss": 0.0822,
+      "step": 8868
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8424633011099178e-05,
+      "loss": 0.1131,
+      "step": 8869
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.1151,
+      "step": 8870
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8417472252058717e-05,
+      "loss": 0.0747,
+      "step": 8871
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.841389187253849e-05,
+      "loss": 0.0634,
+      "step": 8872
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8410311493018264e-05,
+      "loss": 0.094,
+      "step": 8873
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8406731113498033e-05,
+      "loss": 0.1235,
+      "step": 8874
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8403150733977803e-05,
+      "loss": 0.06,
+      "step": 8875
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8399570354457573e-05,
+      "loss": 0.1294,
+      "step": 8876
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8395989974937343e-05,
+      "loss": 0.2254,
+      "step": 8877
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8392409595417116e-05,
+      "loss": 0.2003,
+      "step": 8878
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8388829215896886e-05,
+      "loss": 0.152,
+      "step": 8879
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8385248836376656e-05,
+      "loss": 0.1706,
+      "step": 8880
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8381668456856425e-05,
+      "loss": 0.0524,
+      "step": 8881
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.83780880773362e-05,
+      "loss": 0.1627,
+      "step": 8882
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8374507697815968e-05,
+      "loss": 0.1114,
+      "step": 8883
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.837092731829574e-05,
+      "loss": 0.173,
+      "step": 8884
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 0.1228,
+      "step": 8885
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.836376655925528e-05,
+      "loss": 0.1045,
+      "step": 8886
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8360186179735054e-05,
+      "loss": 0.1704,
+      "step": 8887
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8356605800214824e-05,
+      "loss": 0.1453,
+      "step": 8888
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8353025420694594e-05,
+      "loss": 0.144,
+      "step": 8889
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8349445041174364e-05,
+      "loss": 0.133,
+      "step": 8890
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8345864661654137e-05,
+      "loss": 0.0654,
+      "step": 8891
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8342284282133907e-05,
+      "loss": 0.1146,
+      "step": 8892
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8338703902613676e-05,
+      "loss": 0.1533,
+      "step": 8893
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8335123523093446e-05,
+      "loss": 0.2447,
+      "step": 8894
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.833154314357322e-05,
+      "loss": 0.115,
+      "step": 8895
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8327962764052993e-05,
+      "loss": 0.0712,
+      "step": 8896
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8324382384532762e-05,
+      "loss": 0.1689,
+      "step": 8897
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8320802005012532e-05,
+      "loss": 0.1156,
+      "step": 8898
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.8317221625492302e-05,
+      "loss": 0.1786,
+      "step": 8899
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8313641245972075e-05,
+      "loss": 0.0804,
+      "step": 8900
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8310060866451845e-05,
+      "loss": 0.0975,
+      "step": 8901
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8306480486931615e-05,
+      "loss": 0.0273,
+      "step": 8902
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8302900107411385e-05,
+      "loss": 0.2044,
+      "step": 8903
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8299319727891158e-05,
+      "loss": 0.1385,
+      "step": 8904
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8295739348370928e-05,
+      "loss": 0.0689,
+      "step": 8905
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8292158968850697e-05,
+      "loss": 0.2701,
+      "step": 8906
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.828857858933047e-05,
+      "loss": 0.1777,
+      "step": 8907
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.828499820981024e-05,
+      "loss": 0.1316,
+      "step": 8908
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8281417830290014e-05,
+      "loss": 0.0775,
+      "step": 8909
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8277837450769783e-05,
+      "loss": 0.0563,
+      "step": 8910
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8274257071249553e-05,
+      "loss": 0.1315,
+      "step": 8911
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8270676691729323e-05,
+      "loss": 0.1322,
+      "step": 8912
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8267096312209096e-05,
+      "loss": 0.1776,
+      "step": 8913
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8263515932688866e-05,
+      "loss": 0.3033,
+      "step": 8914
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8259935553168636e-05,
+      "loss": 0.1104,
+      "step": 8915
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8256355173648406e-05,
+      "loss": 0.1884,
+      "step": 8916
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.825277479412818e-05,
+      "loss": 0.1197,
+      "step": 8917
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.824919441460795e-05,
+      "loss": 0.2364,
+      "step": 8918
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8245614035087722e-05,
+      "loss": 0.1655,
+      "step": 8919
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.824203365556749e-05,
+      "loss": 0.1386,
+      "step": 8920
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.823845327604726e-05,
+      "loss": 0.0463,
+      "step": 8921
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8234872896527035e-05,
+      "loss": 0.1251,
+      "step": 8922
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8231292517006804e-05,
+      "loss": 0.1579,
+      "step": 8923
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8227712137486574e-05,
+      "loss": 0.0515,
+      "step": 8924
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8224131757966344e-05,
+      "loss": 0.0715,
+      "step": 8925
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8220551378446117e-05,
+      "loss": 0.1524,
+      "step": 8926
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8216970998925887e-05,
+      "loss": 0.1368,
+      "step": 8927
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8213390619405657e-05,
+      "loss": 0.1699,
+      "step": 8928
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8209810239885427e-05,
+      "loss": 0.1606,
+      "step": 8929
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.82062298603652e-05,
+      "loss": 0.1729,
+      "step": 8930
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8202649480844973e-05,
+      "loss": 0.1935,
+      "step": 8931
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8199069101324743e-05,
+      "loss": 0.0688,
+      "step": 8932
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8195488721804512e-05,
+      "loss": 0.1543,
+      "step": 8933
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8191908342284282e-05,
+      "loss": 0.0935,
+      "step": 8934
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8188327962764055e-05,
+      "loss": 0.1038,
+      "step": 8935
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8184747583243825e-05,
+      "loss": 0.118,
+      "step": 8936
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8181167203723595e-05,
+      "loss": 0.0597,
+      "step": 8937
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8177586824203365e-05,
+      "loss": 0.1384,
+      "step": 8938
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8174006444683138e-05,
+      "loss": 0.2177,
+      "step": 8939
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8170426065162908e-05,
+      "loss": 0.1346,
+      "step": 8940
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8166845685642678e-05,
+      "loss": 0.0624,
+      "step": 8941
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.816326530612245e-05,
+      "loss": 0.1291,
+      "step": 8942
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.815968492660222e-05,
+      "loss": 0.1667,
+      "step": 8943
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8156104547081994e-05,
+      "loss": 0.066,
+      "step": 8944
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8152524167561764e-05,
+      "loss": 0.0665,
+      "step": 8945
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8148943788041533e-05,
+      "loss": 0.0598,
+      "step": 8946
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8145363408521303e-05,
+      "loss": 0.0255,
+      "step": 8947
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8141783029001076e-05,
+      "loss": 0.2287,
+      "step": 8948
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8138202649480846e-05,
+      "loss": 0.148,
+      "step": 8949
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8134622269960616e-05,
+      "loss": 0.0807,
+      "step": 8950
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8131041890440386e-05,
+      "loss": 0.2073,
+      "step": 8951
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.812746151092016e-05,
+      "loss": 0.0906,
+      "step": 8952
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.812388113139993e-05,
+      "loss": 0.188,
+      "step": 8953
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8120300751879702e-05,
+      "loss": 0.1609,
+      "step": 8954
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.8116720372359472e-05,
+      "loss": 0.1141,
+      "step": 8955
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.811313999283924e-05,
+      "loss": 0.1705,
+      "step": 8956
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8109559613319015e-05,
+      "loss": 0.0687,
+      "step": 8957
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8105979233798785e-05,
+      "loss": 0.1101,
+      "step": 8958
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8102398854278554e-05,
+      "loss": 0.0714,
+      "step": 8959
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8098818474758324e-05,
+      "loss": 0.2225,
+      "step": 8960
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8095238095238094e-05,
+      "loss": 0.381,
+      "step": 8961
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8091657715717867e-05,
+      "loss": 0.1242,
+      "step": 8962
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8088077336197637e-05,
+      "loss": 0.1002,
+      "step": 8963
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8084496956677407e-05,
+      "loss": 0.143,
+      "step": 8964
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.808091657715718e-05,
+      "loss": 0.0737,
+      "step": 8965
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8077336197636953e-05,
+      "loss": 0.2575,
+      "step": 8966
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8073755818116723e-05,
+      "loss": 0.2441,
+      "step": 8967
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8070175438596493e-05,
+      "loss": 0.1181,
+      "step": 8968
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8066595059076263e-05,
+      "loss": 0.0706,
+      "step": 8969
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8063014679556032e-05,
+      "loss": 0.2428,
+      "step": 8970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8059434300035806e-05,
+      "loss": 0.1426,
+      "step": 8971
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8055853920515575e-05,
+      "loss": 0.0494,
+      "step": 8972
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8052273540995345e-05,
+      "loss": 0.2441,
+      "step": 8973
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8048693161475115e-05,
+      "loss": 0.0575,
+      "step": 8974
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8045112781954888e-05,
+      "loss": 0.2586,
+      "step": 8975
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8041532402434658e-05,
+      "loss": 0.1024,
+      "step": 8976
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.803795202291443e-05,
+      "loss": 0.1724,
+      "step": 8977
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.80343716433942e-05,
+      "loss": 0.0818,
+      "step": 8978
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.803079126387397e-05,
+      "loss": 0.2457,
+      "step": 8979
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8027210884353744e-05,
+      "loss": 0.072,
+      "step": 8980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8023630504833514e-05,
+      "loss": 0.2495,
+      "step": 8981
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8020050125313283e-05,
+      "loss": 0.2074,
+      "step": 8982
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8016469745793053e-05,
+      "loss": 0.1206,
+      "step": 8983
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8012889366272826e-05,
+      "loss": 0.0987,
+      "step": 8984
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8009308986752596e-05,
+      "loss": 0.1378,
+      "step": 8985
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8005728607232366e-05,
+      "loss": 0.1666,
+      "step": 8986
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.8002148227712136e-05,
+      "loss": 0.2163,
+      "step": 8987
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.799856784819191e-05,
+      "loss": 0.1343,
+      "step": 8988
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7994987468671682e-05,
+      "loss": 0.1036,
+      "step": 8989
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7991407089151452e-05,
+      "loss": 0.0757,
+      "step": 8990
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7987826709631222e-05,
+      "loss": 0.1191,
+      "step": 8991
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.798424633011099e-05,
+      "loss": 0.2064,
+      "step": 8992
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7980665950590765e-05,
+      "loss": 0.1722,
+      "step": 8993
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7977085571070535e-05,
+      "loss": 0.0955,
+      "step": 8994
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7973505191550304e-05,
+      "loss": 0.1057,
+      "step": 8995
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7969924812030074e-05,
+      "loss": 0.1628,
+      "step": 8996
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7966344432509847e-05,
+      "loss": 0.1058,
+      "step": 8997
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7962764052989617e-05,
+      "loss": 0.1158,
+      "step": 8998
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7959183673469387e-05,
+      "loss": 0.1329,
+      "step": 8999
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.795560329394916e-05,
+      "loss": 0.1653,
+      "step": 9000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.795202291442893e-05,
+      "loss": 0.1206,
+      "step": 9001
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7948442534908703e-05,
+      "loss": 0.1279,
+      "step": 9002
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7944862155388473e-05,
+      "loss": 0.1398,
+      "step": 9003
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7941281775868243e-05,
+      "loss": 0.1596,
+      "step": 9004
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7937701396348013e-05,
+      "loss": 0.0587,
+      "step": 9005
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7934121016827786e-05,
+      "loss": 0.0472,
+      "step": 9006
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7930540637307556e-05,
+      "loss": 0.0948,
+      "step": 9007
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7926960257787325e-05,
+      "loss": 0.1242,
+      "step": 9008
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7923379878267095e-05,
+      "loss": 0.1616,
+      "step": 9009
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.791979949874687e-05,
+      "loss": 0.2027,
+      "step": 9010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7916219119226638e-05,
+      "loss": 0.0895,
+      "step": 9011
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.791263873970641e-05,
+      "loss": 0.2569,
+      "step": 9012
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.790905836018618e-05,
+      "loss": 0.2076,
+      "step": 9013
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.790547798066595e-05,
+      "loss": 0.1488,
+      "step": 9014
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7901897601145724e-05,
+      "loss": 0.0619,
+      "step": 9015
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7898317221625494e-05,
+      "loss": 0.1248,
+      "step": 9016
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.0805,
+      "step": 9017
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7891156462585034e-05,
+      "loss": 0.167,
+      "step": 9018
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7887576083064807e-05,
+      "loss": 0.179,
+      "step": 9019
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7883995703544577e-05,
+      "loss": 0.0775,
+      "step": 9020
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7880415324024346e-05,
+      "loss": 0.1357,
+      "step": 9021
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7876834944504116e-05,
+      "loss": 0.0798,
+      "step": 9022
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.787325456498389e-05,
+      "loss": 0.1692,
+      "step": 9023
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7869674185463662e-05,
+      "loss": 0.2097,
+      "step": 9024
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7866093805943432e-05,
+      "loss": 0.1882,
+      "step": 9025
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7862513426423202e-05,
+      "loss": 0.1559,
+      "step": 9026
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7858933046902972e-05,
+      "loss": 0.1667,
+      "step": 9027
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7855352667382745e-05,
+      "loss": 0.1112,
+      "step": 9028
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7851772287862515e-05,
+      "loss": 0.1309,
+      "step": 9029
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7848191908342285e-05,
+      "loss": 0.3538,
+      "step": 9030
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7844611528822054e-05,
+      "loss": 0.134,
+      "step": 9031
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7841031149301828e-05,
+      "loss": 0.0769,
+      "step": 9032
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7837450769781597e-05,
+      "loss": 0.1446,
+      "step": 9033
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7833870390261367e-05,
+      "loss": 0.1195,
+      "step": 9034
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7830290010741137e-05,
+      "loss": 0.2489,
+      "step": 9035
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.782670963122091e-05,
+      "loss": 0.0653,
+      "step": 9036
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7823129251700683e-05,
+      "loss": 0.1302,
+      "step": 9037
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7819548872180453e-05,
+      "loss": 0.173,
+      "step": 9038
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7815968492660223e-05,
+      "loss": 0.0819,
+      "step": 9039
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7812388113139993e-05,
+      "loss": 0.1694,
+      "step": 9040
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7808807733619766e-05,
+      "loss": 0.1651,
+      "step": 9041
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7805227354099536e-05,
+      "loss": 0.2063,
+      "step": 9042
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7801646974579306e-05,
+      "loss": 0.2743,
+      "step": 9043
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7798066595059075e-05,
+      "loss": 0.2992,
+      "step": 9044
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.779448621553885e-05,
+      "loss": 0.1569,
+      "step": 9045
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.779090583601862e-05,
+      "loss": 0.1204,
+      "step": 9046
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.778732545649839e-05,
+      "loss": 0.1738,
+      "step": 9047
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.778374507697816e-05,
+      "loss": 0.0909,
+      "step": 9048
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.778016469745793e-05,
+      "loss": 0.1512,
+      "step": 9049
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7776584317937704e-05,
+      "loss": 0.1795,
+      "step": 9050
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7773003938417474e-05,
+      "loss": 0.2631,
+      "step": 9051
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7769423558897244e-05,
+      "loss": 0.0789,
+      "step": 9052
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7765843179377014e-05,
+      "loss": 0.1466,
+      "step": 9053
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7762262799856787e-05,
+      "loss": 0.2751,
+      "step": 9054
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7758682420336557e-05,
+      "loss": 0.1104,
+      "step": 9055
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 0.1261,
+      "step": 9056
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7751521661296096e-05,
+      "loss": 0.1265,
+      "step": 9057
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7747941281775866e-05,
+      "loss": 0.0828,
+      "step": 9058
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7744360902255643e-05,
+      "loss": 0.0935,
+      "step": 9059
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7740780522735413e-05,
+      "loss": 0.1039,
+      "step": 9060
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7737200143215182e-05,
+      "loss": 0.1145,
+      "step": 9061
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7733619763694952e-05,
+      "loss": 0.2178,
+      "step": 9062
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7730039384174725e-05,
+      "loss": 0.0976,
+      "step": 9063
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7726459004654495e-05,
+      "loss": 0.2469,
+      "step": 9064
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7722878625134265e-05,
+      "loss": 0.1125,
+      "step": 9065
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7719298245614035e-05,
+      "loss": 0.0598,
+      "step": 9066
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7715717866093805e-05,
+      "loss": 0.2599,
+      "step": 9067
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7712137486573578e-05,
+      "loss": 0.1139,
+      "step": 9068
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7708557107053348e-05,
+      "loss": 0.1421,
+      "step": 9069
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7704976727533117e-05,
+      "loss": 0.1135,
+      "step": 9070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.770139634801289e-05,
+      "loss": 0.1407,
+      "step": 9071
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.769781596849266e-05,
+      "loss": 0.1611,
+      "step": 9072
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7694235588972433e-05,
+      "loss": 0.0486,
+      "step": 9073
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7690655209452203e-05,
+      "loss": 0.1298,
+      "step": 9074
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7687074829931973e-05,
+      "loss": 0.0912,
+      "step": 9075
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7683494450411743e-05,
+      "loss": 0.1123,
+      "step": 9076
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7679914070891516e-05,
+      "loss": 0.1262,
+      "step": 9077
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7676333691371286e-05,
+      "loss": 0.0832,
+      "step": 9078
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7672753311851056e-05,
+      "loss": 0.1093,
+      "step": 9079
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7669172932330825e-05,
+      "loss": 0.2974,
+      "step": 9080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.76655925528106e-05,
+      "loss": 0.0757,
+      "step": 9081
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7662012173290372e-05,
+      "loss": 0.1176,
+      "step": 9082
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.765843179377014e-05,
+      "loss": 0.1608,
+      "step": 9083
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.765485141424991e-05,
+      "loss": 0.0738,
+      "step": 9084
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.765127103472968e-05,
+      "loss": 0.1454,
+      "step": 9085
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7647690655209454e-05,
+      "loss": 0.1548,
+      "step": 9086
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7644110275689224e-05,
+      "loss": 0.0429,
+      "step": 9087
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7640529896168994e-05,
+      "loss": 0.1399,
+      "step": 9088
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7636949516648764e-05,
+      "loss": 0.1217,
+      "step": 9089
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7633369137128537e-05,
+      "loss": 0.062,
+      "step": 9090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7629788757608307e-05,
+      "loss": 0.2326,
+      "step": 9091
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7626208378088077e-05,
+      "loss": 0.2023,
+      "step": 9092
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7622627998567846e-05,
+      "loss": 0.0782,
+      "step": 9093
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.761904761904762e-05,
+      "loss": 0.2799,
+      "step": 9094
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.7615467239527393e-05,
+      "loss": 0.1501,
+      "step": 9095
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7611886860007163e-05,
+      "loss": 0.0852,
+      "step": 9096
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7608306480486932e-05,
+      "loss": 0.1571,
+      "step": 9097
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7604726100966702e-05,
+      "loss": 0.1589,
+      "step": 9098
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7601145721446475e-05,
+      "loss": 0.0854,
+      "step": 9099
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7597565341926245e-05,
+      "loss": 0.212,
+      "step": 9100
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7593984962406015e-05,
+      "loss": 0.1716,
+      "step": 9101
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7590404582885785e-05,
+      "loss": 0.0918,
+      "step": 9102
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7586824203365558e-05,
+      "loss": 0.2865,
+      "step": 9103
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7583243823845328e-05,
+      "loss": 0.1439,
+      "step": 9104
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7579663444325098e-05,
+      "loss": 0.2002,
+      "step": 9105
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.757608306480487e-05,
+      "loss": 0.1451,
+      "step": 9106
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.757250268528464e-05,
+      "loss": 0.2156,
+      "step": 9107
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7568922305764414e-05,
+      "loss": 0.1797,
+      "step": 9108
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7565341926244184e-05,
+      "loss": 0.0948,
+      "step": 9109
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7561761546723953e-05,
+      "loss": 0.1486,
+      "step": 9110
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7558181167203723e-05,
+      "loss": 0.0857,
+      "step": 9111
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7554600787683496e-05,
+      "loss": 0.1186,
+      "step": 9112
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 0.1891,
+      "step": 9113
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7547440028643036e-05,
+      "loss": 0.1469,
+      "step": 9114
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 0.2324,
+      "step": 9115
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.754027926960258e-05,
+      "loss": 0.2288,
+      "step": 9116
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7536698890082352e-05,
+      "loss": 0.0871,
+      "step": 9117
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7533118510562122e-05,
+      "loss": 0.2022,
+      "step": 9118
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.752953813104189e-05,
+      "loss": 0.1049,
+      "step": 9119
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.752595775152166e-05,
+      "loss": 0.2398,
+      "step": 9120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7522377372001435e-05,
+      "loss": 0.1901,
+      "step": 9121
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7518796992481204e-05,
+      "loss": 0.1042,
+      "step": 9122
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7515216612960974e-05,
+      "loss": 0.1082,
+      "step": 9123
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7511636233440744e-05,
+      "loss": 0.1641,
+      "step": 9124
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7508055853920517e-05,
+      "loss": 0.1562,
+      "step": 9125
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7504475474400287e-05,
+      "loss": 0.1929,
+      "step": 9126
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7500895094880057e-05,
+      "loss": 0.1549,
+      "step": 9127
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7497314715359827e-05,
+      "loss": 0.2004,
+      "step": 9128
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.74937343358396e-05,
+      "loss": 0.2401,
+      "step": 9129
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7490153956319373e-05,
+      "loss": 0.1454,
+      "step": 9130
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7486573576799143e-05,
+      "loss": 0.1187,
+      "step": 9131
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7482993197278913e-05,
+      "loss": 0.1389,
+      "step": 9132
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7479412817758682e-05,
+      "loss": 0.1003,
+      "step": 9133
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7475832438238456e-05,
+      "loss": 0.1695,
+      "step": 9134
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7472252058718225e-05,
+      "loss": 0.3426,
+      "step": 9135
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7468671679197995e-05,
+      "loss": 0.1543,
+      "step": 9136
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7465091299677765e-05,
+      "loss": 0.0803,
+      "step": 9137
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7461510920157538e-05,
+      "loss": 0.0972,
+      "step": 9138
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7457930540637308e-05,
+      "loss": 0.0935,
+      "step": 9139
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7454350161117078e-05,
+      "loss": 0.2064,
+      "step": 9140
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.745076978159685e-05,
+      "loss": 0.1214,
+      "step": 9141
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.744718940207662e-05,
+      "loss": 0.2356,
+      "step": 9142
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7443609022556394e-05,
+      "loss": 0.2453,
+      "step": 9143
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7440028643036164e-05,
+      "loss": 0.0843,
+      "step": 9144
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7436448263515934e-05,
+      "loss": 0.162,
+      "step": 9145
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7432867883995703e-05,
+      "loss": 0.0998,
+      "step": 9146
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7429287504475477e-05,
+      "loss": 0.0738,
+      "step": 9147
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7425707124955246e-05,
+      "loss": 0.2353,
+      "step": 9148
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7422126745435016e-05,
+      "loss": 0.1257,
+      "step": 9149
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7418546365914786e-05,
+      "loss": 0.1019,
+      "step": 9150
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7414965986394556e-05,
+      "loss": 0.308,
+      "step": 9151
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7411385606874332e-05,
+      "loss": 0.1507,
+      "step": 9152
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7407805227354102e-05,
+      "loss": 0.1693,
+      "step": 9153
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7404224847833872e-05,
+      "loss": 0.1524,
+      "step": 9154
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7400644468313642e-05,
+      "loss": 0.1057,
+      "step": 9155
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7397064088793415e-05,
+      "loss": 0.1373,
+      "step": 9156
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7393483709273185e-05,
+      "loss": 0.0643,
+      "step": 9157
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7389903329752955e-05,
+      "loss": 0.161,
+      "step": 9158
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7386322950232724e-05,
+      "loss": 0.0668,
+      "step": 9159
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7382742570712494e-05,
+      "loss": 0.081,
+      "step": 9160
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7379162191192267e-05,
+      "loss": 0.222,
+      "step": 9161
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7375581811672037e-05,
+      "loss": 0.2,
+      "step": 9162
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7372001432151807e-05,
+      "loss": 0.2435,
+      "step": 9163
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.736842105263158e-05,
+      "loss": 0.0963,
+      "step": 9164
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7364840673111353e-05,
+      "loss": 0.1687,
+      "step": 9165
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7361260293591123e-05,
+      "loss": 0.153,
+      "step": 9166
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7357679914070893e-05,
+      "loss": 0.0903,
+      "step": 9167
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7354099534550663e-05,
+      "loss": 0.0945,
+      "step": 9168
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7350519155030432e-05,
+      "loss": 0.1275,
+      "step": 9169
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7346938775510206e-05,
+      "loss": 0.144,
+      "step": 9170
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7343358395989975e-05,
+      "loss": 0.0696,
+      "step": 9171
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7339778016469745e-05,
+      "loss": 0.0936,
+      "step": 9172
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7336197636949515e-05,
+      "loss": 0.1727,
+      "step": 9173
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7332617257429288e-05,
+      "loss": 0.2174,
+      "step": 9174
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7329036877909058e-05,
+      "loss": 0.0733,
+      "step": 9175
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.732545649838883e-05,
+      "loss": 0.2065,
+      "step": 9176
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.73218761188686e-05,
+      "loss": 0.1298,
+      "step": 9177
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.731829573934837e-05,
+      "loss": 0.1018,
+      "step": 9178
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7314715359828144e-05,
+      "loss": 0.1652,
+      "step": 9179
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7311134980307914e-05,
+      "loss": 0.0735,
+      "step": 9180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7307554600787684e-05,
+      "loss": 0.1529,
+      "step": 9181
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7303974221267453e-05,
+      "loss": 0.2453,
+      "step": 9182
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7300393841747227e-05,
+      "loss": 0.2036,
+      "step": 9183
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7296813462226996e-05,
+      "loss": 0.1594,
+      "step": 9184
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7293233082706766e-05,
+      "loss": 0.1261,
+      "step": 9185
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7289652703186536e-05,
+      "loss": 0.2386,
+      "step": 9186
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.728607232366631e-05,
+      "loss": 0.0954,
+      "step": 9187
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7282491944146082e-05,
+      "loss": 0.1256,
+      "step": 9188
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7278911564625852e-05,
+      "loss": 0.1043,
+      "step": 9189
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7275331185105622e-05,
+      "loss": 0.0645,
+      "step": 9190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7271750805585392e-05,
+      "loss": 0.1061,
+      "step": 9191
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7268170426065165e-05,
+      "loss": 0.1055,
+      "step": 9192
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7264590046544935e-05,
+      "loss": 0.0792,
+      "step": 9193
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7261009667024705e-05,
+      "loss": 0.122,
+      "step": 9194
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7257429287504474e-05,
+      "loss": 0.1838,
+      "step": 9195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7253848907984248e-05,
+      "loss": 0.23,
+      "step": 9196
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7250268528464017e-05,
+      "loss": 0.0397,
+      "step": 9197
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7246688148943787e-05,
+      "loss": 0.1977,
+      "step": 9198
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.724310776942356e-05,
+      "loss": 0.1785,
+      "step": 9199
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.723952738990333e-05,
+      "loss": 0.0687,
+      "step": 9200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7235947010383103e-05,
+      "loss": 0.0968,
+      "step": 9201
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7232366630862873e-05,
+      "loss": 0.0854,
+      "step": 9202
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7228786251342643e-05,
+      "loss": 0.076,
+      "step": 9203
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7225205871822413e-05,
+      "loss": 0.1906,
+      "step": 9204
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7221625492302186e-05,
+      "loss": 0.1484,
+      "step": 9205
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7218045112781956e-05,
+      "loss": 0.1169,
+      "step": 9206
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7214464733261726e-05,
+      "loss": 0.1461,
+      "step": 9207
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7210884353741495e-05,
+      "loss": 0.2178,
+      "step": 9208
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.720730397422127e-05,
+      "loss": 0.1602,
+      "step": 9209
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.720372359470104e-05,
+      "loss": 0.1724,
+      "step": 9210
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.720014321518081e-05,
+      "loss": 0.0908,
+      "step": 9211
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.719656283566058e-05,
+      "loss": 0.1192,
+      "step": 9212
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.12,
+      "step": 9213
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7189402076620124e-05,
+      "loss": 0.071,
+      "step": 9214
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7185821697099894e-05,
+      "loss": 0.1254,
+      "step": 9215
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7182241317579664e-05,
+      "loss": 0.1738,
+      "step": 9216
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7178660938059434e-05,
+      "loss": 0.1665,
+      "step": 9217
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7175080558539207e-05,
+      "loss": 0.2312,
+      "step": 9218
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7171500179018977e-05,
+      "loss": 0.1291,
+      "step": 9219
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7167919799498746e-05,
+      "loss": 0.1308,
+      "step": 9220
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7164339419978516e-05,
+      "loss": 0.1131,
+      "step": 9221
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.716075904045829e-05,
+      "loss": 0.2128,
+      "step": 9222
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7157178660938063e-05,
+      "loss": 0.0963,
+      "step": 9223
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7153598281417832e-05,
+      "loss": 0.1256,
+      "step": 9224
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7150017901897602e-05,
+      "loss": 0.1334,
+      "step": 9225
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7146437522377372e-05,
+      "loss": 0.0895,
+      "step": 9226
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7142857142857145e-05,
+      "loss": 0.1274,
+      "step": 9227
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7139276763336915e-05,
+      "loss": 0.2728,
+      "step": 9228
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7135696383816685e-05,
+      "loss": 0.1651,
+      "step": 9229
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7132116004296455e-05,
+      "loss": 0.0782,
+      "step": 9230
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7128535624776228e-05,
+      "loss": 0.0985,
+      "step": 9231
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7124955245255998e-05,
+      "loss": 0.1673,
+      "step": 9232
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.7121374865735767e-05,
+      "loss": 0.2169,
+      "step": 9233
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.711779448621554e-05,
+      "loss": 0.1088,
+      "step": 9234
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.711421410669531e-05,
+      "loss": 0.0785,
+      "step": 9235
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7110633727175084e-05,
+      "loss": 0.1607,
+      "step": 9236
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7107053347654853e-05,
+      "loss": 0.1071,
+      "step": 9237
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7103472968134623e-05,
+      "loss": 0.1314,
+      "step": 9238
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7099892588614393e-05,
+      "loss": 0.2283,
+      "step": 9239
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7096312209094166e-05,
+      "loss": 0.2193,
+      "step": 9240
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7092731829573936e-05,
+      "loss": 0.1352,
+      "step": 9241
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7089151450053706e-05,
+      "loss": 0.2435,
+      "step": 9242
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7085571070533476e-05,
+      "loss": 0.1137,
+      "step": 9243
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.708199069101325e-05,
+      "loss": 0.0974,
+      "step": 9244
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.707841031149302e-05,
+      "loss": 0.1818,
+      "step": 9245
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7074829931972792e-05,
+      "loss": 0.2709,
+      "step": 9246
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.707124955245256e-05,
+      "loss": 0.2586,
+      "step": 9247
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.706766917293233e-05,
+      "loss": 0.1537,
+      "step": 9248
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7064088793412105e-05,
+      "loss": 0.104,
+      "step": 9249
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7060508413891874e-05,
+      "loss": 0.1403,
+      "step": 9250
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7056928034371644e-05,
+      "loss": 0.1558,
+      "step": 9251
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7053347654851414e-05,
+      "loss": 0.0563,
+      "step": 9252
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7049767275331187e-05,
+      "loss": 0.2302,
+      "step": 9253
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7046186895810957e-05,
+      "loss": 0.1064,
+      "step": 9254
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7042606516290727e-05,
+      "loss": 0.0989,
+      "step": 9255
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7039026136770497e-05,
+      "loss": 0.1341,
+      "step": 9256
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.703544575725027e-05,
+      "loss": 0.0988,
+      "step": 9257
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7031865377730043e-05,
+      "loss": 0.0846,
+      "step": 9258
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7028284998209813e-05,
+      "loss": 0.122,
+      "step": 9259
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7024704618689582e-05,
+      "loss": 0.1187,
+      "step": 9260
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7021124239169352e-05,
+      "loss": 0.1823,
+      "step": 9261
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7017543859649125e-05,
+      "loss": 0.1322,
+      "step": 9262
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.7013963480128895e-05,
+      "loss": 0.2298,
+      "step": 9263
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7010383100608665e-05,
+      "loss": 0.0834,
+      "step": 9264
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7006802721088435e-05,
+      "loss": 0.1357,
+      "step": 9265
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.7003222341568205e-05,
+      "loss": 0.0712,
+      "step": 9266
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6999641962047978e-05,
+      "loss": 0.1683,
+      "step": 9267
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6996061582527748e-05,
+      "loss": 0.0763,
+      "step": 9268
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.699248120300752e-05,
+      "loss": 0.1046,
+      "step": 9269
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.698890082348729e-05,
+      "loss": 0.22,
+      "step": 9270
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.698532044396706e-05,
+      "loss": 0.2123,
+      "step": 9271
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6981740064446834e-05,
+      "loss": 0.1327,
+      "step": 9272
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6978159684926603e-05,
+      "loss": 0.1076,
+      "step": 9273
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6974579305406373e-05,
+      "loss": 0.2416,
+      "step": 9274
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6970998925886143e-05,
+      "loss": 0.1375,
+      "step": 9275
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6967418546365916e-05,
+      "loss": 0.1858,
+      "step": 9276
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6963838166845686e-05,
+      "loss": 0.0623,
+      "step": 9277
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6960257787325456e-05,
+      "loss": 0.1303,
+      "step": 9278
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6956677407805226e-05,
+      "loss": 0.163,
+      "step": 9279
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6953097028285e-05,
+      "loss": 0.167,
+      "step": 9280
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6949516648764772e-05,
+      "loss": 0.153,
+      "step": 9281
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6945936269244542e-05,
+      "loss": 0.0717,
+      "step": 9282
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.694235588972431e-05,
+      "loss": 0.1243,
+      "step": 9283
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.693877551020408e-05,
+      "loss": 0.063,
+      "step": 9284
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6935195130683855e-05,
+      "loss": 0.1136,
+      "step": 9285
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6931614751163624e-05,
+      "loss": 0.1788,
+      "step": 9286
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6928034371643394e-05,
+      "loss": 0.0882,
+      "step": 9287
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6924453992123164e-05,
+      "loss": 0.1311,
+      "step": 9288
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6920873612602937e-05,
+      "loss": 0.1928,
+      "step": 9289
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6917293233082707e-05,
+      "loss": 0.1628,
+      "step": 9290
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6913712853562477e-05,
+      "loss": 0.1067,
+      "step": 9291
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.691013247404225e-05,
+      "loss": 0.067,
+      "step": 9292
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.690655209452202e-05,
+      "loss": 0.2112,
+      "step": 9293
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6902971715001793e-05,
+      "loss": 0.135,
+      "step": 9294
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6899391335481563e-05,
+      "loss": 0.207,
+      "step": 9295
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6895810955961333e-05,
+      "loss": 0.1912,
+      "step": 9296
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6892230576441102e-05,
+      "loss": 0.2106,
+      "step": 9297
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6888650196920876e-05,
+      "loss": 0.0769,
+      "step": 9298
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6885069817400645e-05,
+      "loss": 0.0805,
+      "step": 9299
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6881489437880415e-05,
+      "loss": 0.1679,
+      "step": 9300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6877909058360185e-05,
+      "loss": 0.104,
+      "step": 9301
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6874328678839958e-05,
+      "loss": 0.2087,
+      "step": 9302
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6870748299319728e-05,
+      "loss": 0.2142,
+      "step": 9303
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.68671679197995e-05,
+      "loss": 0.1464,
+      "step": 9304
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.686358754027927e-05,
+      "loss": 0.1279,
+      "step": 9305
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.686000716075904e-05,
+      "loss": 0.1217,
+      "step": 9306
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6856426781238814e-05,
+      "loss": 0.1159,
+      "step": 9307
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6852846401718584e-05,
+      "loss": 0.1857,
+      "step": 9308
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6849266022198353e-05,
+      "loss": 0.2274,
+      "step": 9309
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6845685642678123e-05,
+      "loss": 0.1526,
+      "step": 9310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.0814,
+      "step": 9311
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6838524883637666e-05,
+      "loss": 0.0919,
+      "step": 9312
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6834944504117436e-05,
+      "loss": 0.1819,
+      "step": 9313
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6831364124597206e-05,
+      "loss": 0.0619,
+      "step": 9314
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.682778374507698e-05,
+      "loss": 0.2732,
+      "step": 9315
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6824203365556752e-05,
+      "loss": 0.2112,
+      "step": 9316
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6820622986036522e-05,
+      "loss": 0.1221,
+      "step": 9317
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6817042606516292e-05,
+      "loss": 0.1924,
+      "step": 9318
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.681346222699606e-05,
+      "loss": 0.0926,
+      "step": 9319
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6809881847475835e-05,
+      "loss": 0.0993,
+      "step": 9320
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6806301467955605e-05,
+      "loss": 0.0866,
+      "step": 9321
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6802721088435374e-05,
+      "loss": 0.0951,
+      "step": 9322
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6799140708915144e-05,
+      "loss": 0.1318,
+      "step": 9323
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6795560329394917e-05,
+      "loss": 0.117,
+      "step": 9324
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6791979949874687e-05,
+      "loss": 0.1959,
+      "step": 9325
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6788399570354457e-05,
+      "loss": 0.1577,
+      "step": 9326
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.678481919083423e-05,
+      "loss": 0.1137,
+      "step": 9327
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6781238811314e-05,
+      "loss": 0.1428,
+      "step": 9328
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6777658431793773e-05,
+      "loss": 0.1947,
+      "step": 9329
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6774078052273543e-05,
+      "loss": 0.1363,
+      "step": 9330
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6770497672753313e-05,
+      "loss": 0.1069,
+      "step": 9331
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6766917293233083e-05,
+      "loss": 0.1297,
+      "step": 9332
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6763336913712856e-05,
+      "loss": 0.2238,
+      "step": 9333
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6759756534192626e-05,
+      "loss": 0.1195,
+      "step": 9334
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6756176154672395e-05,
+      "loss": 0.1939,
+      "step": 9335
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6752595775152165e-05,
+      "loss": 0.1085,
+      "step": 9336
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.674901539563194e-05,
+      "loss": 0.031,
+      "step": 9337
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6745435016111708e-05,
+      "loss": 0.1309,
+      "step": 9338
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.674185463659148e-05,
+      "loss": 0.1812,
+      "step": 9339
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.673827425707125e-05,
+      "loss": 0.1348,
+      "step": 9340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 0.1146,
+      "step": 9341
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6731113498030794e-05,
+      "loss": 0.0439,
+      "step": 9342
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6727533118510564e-05,
+      "loss": 0.0711,
+      "step": 9343
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6723952738990334e-05,
+      "loss": 0.1831,
+      "step": 9344
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6720372359470104e-05,
+      "loss": 0.1459,
+      "step": 9345
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6716791979949877e-05,
+      "loss": 0.4047,
+      "step": 9346
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6713211600429647e-05,
+      "loss": 0.0635,
+      "step": 9347
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6709631220909416e-05,
+      "loss": 0.1684,
+      "step": 9348
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6706050841389186e-05,
+      "loss": 0.1413,
+      "step": 9349
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6702470461868956e-05,
+      "loss": 0.1242,
+      "step": 9350
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6698890082348732e-05,
+      "loss": 0.1042,
+      "step": 9351
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6695309702828502e-05,
+      "loss": 0.0601,
+      "step": 9352
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6691729323308272e-05,
+      "loss": 0.1502,
+      "step": 9353
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6688148943788042e-05,
+      "loss": 0.0427,
+      "step": 9354
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6684568564267815e-05,
+      "loss": 0.1407,
+      "step": 9355
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6680988184747585e-05,
+      "loss": 0.0464,
+      "step": 9356
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6677407805227355e-05,
+      "loss": 0.129,
+      "step": 9357
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6673827425707124e-05,
+      "loss": 0.1351,
+      "step": 9358
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6670247046186894e-05,
+      "loss": 0.1479,
+      "step": 9359
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.1431,
+      "step": 9360
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6663086287146437e-05,
+      "loss": 0.0608,
+      "step": 9361
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.665950590762621e-05,
+      "loss": 0.1596,
+      "step": 9362
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.665592552810598e-05,
+      "loss": 0.1831,
+      "step": 9363
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6652345148585753e-05,
+      "loss": 0.0607,
+      "step": 9364
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6648764769065523e-05,
+      "loss": 0.0843,
+      "step": 9365
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6645184389545293e-05,
+      "loss": 0.1125,
+      "step": 9366
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6641604010025063e-05,
+      "loss": 0.1449,
+      "step": 9367
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6638023630504833e-05,
+      "loss": 0.2447,
+      "step": 9368
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6634443250984606e-05,
+      "loss": 0.104,
+      "step": 9369
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6630862871464376e-05,
+      "loss": 0.269,
+      "step": 9370
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6627282491944145e-05,
+      "loss": 0.1178,
+      "step": 9371
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6623702112423915e-05,
+      "loss": 0.1514,
+      "step": 9372
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.662012173290369e-05,
+      "loss": 0.1251,
+      "step": 9373
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.661654135338346e-05,
+      "loss": 0.1942,
+      "step": 9374
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.661296097386323e-05,
+      "loss": 0.231,
+      "step": 9375
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.6609380594343e-05,
+      "loss": 0.0758,
+      "step": 9376
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.660580021482277e-05,
+      "loss": 0.0865,
+      "step": 9377
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6602219835302544e-05,
+      "loss": 0.1394,
+      "step": 9378
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6598639455782314e-05,
+      "loss": 0.2584,
+      "step": 9379
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6595059076262084e-05,
+      "loss": 0.1745,
+      "step": 9380
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6591478696741854e-05,
+      "loss": 0.1821,
+      "step": 9381
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6587898317221627e-05,
+      "loss": 0.2208,
+      "step": 9382
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6584317937701397e-05,
+      "loss": 0.0933,
+      "step": 9383
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6580737558181166e-05,
+      "loss": 0.2226,
+      "step": 9384
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6577157178660936e-05,
+      "loss": 0.1077,
+      "step": 9385
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.657357679914071e-05,
+      "loss": 0.0558,
+      "step": 9386
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6569996419620483e-05,
+      "loss": 0.1233,
+      "step": 9387
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6566416040100252e-05,
+      "loss": 0.2238,
+      "step": 9388
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6562835660580022e-05,
+      "loss": 0.0726,
+      "step": 9389
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6559255281059792e-05,
+      "loss": 0.1434,
+      "step": 9390
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6555674901539565e-05,
+      "loss": 0.1758,
+      "step": 9391
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6552094522019335e-05,
+      "loss": 0.2326,
+      "step": 9392
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6548514142499105e-05,
+      "loss": 0.1251,
+      "step": 9393
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6544933762978875e-05,
+      "loss": 0.2072,
+      "step": 9394
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6541353383458648e-05,
+      "loss": 0.0963,
+      "step": 9395
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6537773003938418e-05,
+      "loss": 0.2856,
+      "step": 9396
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.653419262441819e-05,
+      "loss": 0.1315,
+      "step": 9397
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.653061224489796e-05,
+      "loss": 0.1961,
+      "step": 9398
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.652703186537773e-05,
+      "loss": 0.1545,
+      "step": 9399
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6523451485857503e-05,
+      "loss": 0.0625,
+      "step": 9400
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6519871106337273e-05,
+      "loss": 0.0627,
+      "step": 9401
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6516290726817043e-05,
+      "loss": 0.1866,
+      "step": 9402
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6512710347296813e-05,
+      "loss": 0.0892,
+      "step": 9403
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6509129967776586e-05,
+      "loss": 0.1065,
+      "step": 9404
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6505549588256356e-05,
+      "loss": 0.0724,
+      "step": 9405
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6501969208736126e-05,
+      "loss": 0.1426,
+      "step": 9406
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6498388829215895e-05,
+      "loss": 0.1964,
+      "step": 9407
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.649480844969567e-05,
+      "loss": 0.1044,
+      "step": 9408
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6491228070175442e-05,
+      "loss": 0.0719,
+      "step": 9409
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.648764769065521e-05,
+      "loss": 0.1009,
+      "step": 9410
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.648406731113498e-05,
+      "loss": 0.2018,
+      "step": 9411
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.648048693161475e-05,
+      "loss": 0.2837,
+      "step": 9412
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6476906552094524e-05,
+      "loss": 0.0595,
+      "step": 9413
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6473326172574294e-05,
+      "loss": 0.0752,
+      "step": 9414
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6469745793054064e-05,
+      "loss": 0.1751,
+      "step": 9415
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6466165413533834e-05,
+      "loss": 0.0846,
+      "step": 9416
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6462585034013607e-05,
+      "loss": 0.1691,
+      "step": 9417
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6459004654493377e-05,
+      "loss": 0.1568,
+      "step": 9418
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6455424274973147e-05,
+      "loss": 0.0727,
+      "step": 9419
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6451843895452916e-05,
+      "loss": 0.1712,
+      "step": 9420
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.644826351593269e-05,
+      "loss": 0.0936,
+      "step": 9421
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6444683136412463e-05,
+      "loss": 0.2818,
+      "step": 9422
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6441102756892233e-05,
+      "loss": 0.0211,
+      "step": 9423
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6437522377372002e-05,
+      "loss": 0.0961,
+      "step": 9424
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6433941997851772e-05,
+      "loss": 0.157,
+      "step": 9425
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6430361618331545e-05,
+      "loss": 0.1538,
+      "step": 9426
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6426781238811315e-05,
+      "loss": 0.0637,
+      "step": 9427
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6423200859291085e-05,
+      "loss": 0.1278,
+      "step": 9428
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6419620479770855e-05,
+      "loss": 0.1038,
+      "step": 9429
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6416040100250628e-05,
+      "loss": 0.086,
+      "step": 9430
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6412459720730398e-05,
+      "loss": 0.064,
+      "step": 9431
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6408879341210168e-05,
+      "loss": 0.0657,
+      "step": 9432
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.640529896168994e-05,
+      "loss": 0.2593,
+      "step": 9433
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.640171858216971e-05,
+      "loss": 0.2822,
+      "step": 9434
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6398138202649484e-05,
+      "loss": 0.0487,
+      "step": 9435
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6394557823129254e-05,
+      "loss": 0.0777,
+      "step": 9436
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6390977443609023e-05,
+      "loss": 0.0953,
+      "step": 9437
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6387397064088793e-05,
+      "loss": 0.3774,
+      "step": 9438
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6383816684568566e-05,
+      "loss": 0.0758,
+      "step": 9439
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6380236305048336e-05,
+      "loss": 0.1545,
+      "step": 9440
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6376655925528106e-05,
+      "loss": 0.2291,
+      "step": 9441
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6373075546007876e-05,
+      "loss": 0.1938,
+      "step": 9442
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.636949516648765e-05,
+      "loss": 0.1205,
+      "step": 9443
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6365914786967422e-05,
+      "loss": 0.1621,
+      "step": 9444
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6362334407447192e-05,
+      "loss": 0.1792,
+      "step": 9445
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6358754027926962e-05,
+      "loss": 0.1587,
+      "step": 9446
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.635517364840673e-05,
+      "loss": 0.092,
+      "step": 9447
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6351593268886505e-05,
+      "loss": 0.0414,
+      "step": 9448
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6348012889366274e-05,
+      "loss": 0.0652,
+      "step": 9449
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6344432509846044e-05,
+      "loss": 0.1623,
+      "step": 9450
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6340852130325814e-05,
+      "loss": 0.268,
+      "step": 9451
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6337271750805587e-05,
+      "loss": 0.0349,
+      "step": 9452
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6333691371285357e-05,
+      "loss": 0.0863,
+      "step": 9453
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6330110991765127e-05,
+      "loss": 0.1224,
+      "step": 9454
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6326530612244897e-05,
+      "loss": 0.11,
+      "step": 9455
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.632295023272467e-05,
+      "loss": 0.0566,
+      "step": 9456
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6319369853204443e-05,
+      "loss": 0.0786,
+      "step": 9457
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6315789473684213e-05,
+      "loss": 0.0401,
+      "step": 9458
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6312209094163983e-05,
+      "loss": 0.1735,
+      "step": 9459
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6308628714643752e-05,
+      "loss": 0.1592,
+      "step": 9460
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6305048335123522e-05,
+      "loss": 0.2011,
+      "step": 9461
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6301467955603295e-05,
+      "loss": 0.1812,
+      "step": 9462
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6297887576083065e-05,
+      "loss": 0.1137,
+      "step": 9463
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6294307196562835e-05,
+      "loss": 0.1087,
+      "step": 9464
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6290726817042605e-05,
+      "loss": 0.242,
+      "step": 9465
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6287146437522378e-05,
+      "loss": 0.0495,
+      "step": 9466
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6283566058002148e-05,
+      "loss": 0.1236,
+      "step": 9467
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.627998567848192e-05,
+      "loss": 0.1132,
+      "step": 9468
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.627640529896169e-05,
+      "loss": 0.1066,
+      "step": 9469
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.627282491944146e-05,
+      "loss": 0.1083,
+      "step": 9470
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6269244539921234e-05,
+      "loss": 0.1471,
+      "step": 9471
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6265664160401004e-05,
+      "loss": 0.0539,
+      "step": 9472
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6262083780880773e-05,
+      "loss": 0.1722,
+      "step": 9473
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6258503401360543e-05,
+      "loss": 0.1806,
+      "step": 9474
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6254923021840316e-05,
+      "loss": 0.0594,
+      "step": 9475
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6251342642320086e-05,
+      "loss": 0.0967,
+      "step": 9476
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6247762262799856e-05,
+      "loss": 0.1903,
+      "step": 9477
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6244181883279626e-05,
+      "loss": 0.106,
+      "step": 9478
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.62406015037594e-05,
+      "loss": 0.1614,
+      "step": 9479
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6237021124239172e-05,
+      "loss": 0.1991,
+      "step": 9480
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6233440744718942e-05,
+      "loss": 0.1582,
+      "step": 9481
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6229860365198712e-05,
+      "loss": 0.1456,
+      "step": 9482
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.622627998567848e-05,
+      "loss": 0.065,
+      "step": 9483
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6222699606158255e-05,
+      "loss": 0.0889,
+      "step": 9484
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6219119226638025e-05,
+      "loss": 0.1091,
+      "step": 9485
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6215538847117794e-05,
+      "loss": 0.0669,
+      "step": 9486
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6211958467597564e-05,
+      "loss": 0.1087,
+      "step": 9487
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6208378088077337e-05,
+      "loss": 0.1077,
+      "step": 9488
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6204797708557107e-05,
+      "loss": 0.1276,
+      "step": 9489
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6201217329036877e-05,
+      "loss": 0.1618,
+      "step": 9490
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.619763694951665e-05,
+      "loss": 0.0624,
+      "step": 9491
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.619405656999642e-05,
+      "loss": 0.3187,
+      "step": 9492
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.1898,
+      "step": 9493
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6186895810955963e-05,
+      "loss": 0.0636,
+      "step": 9494
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6183315431435733e-05,
+      "loss": 0.1141,
+      "step": 9495
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6179735051915502e-05,
+      "loss": 0.1546,
+      "step": 9496
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6176154672395276e-05,
+      "loss": 0.0853,
+      "step": 9497
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6172574292875045e-05,
+      "loss": 0.1458,
+      "step": 9498
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6168993913354815e-05,
+      "loss": 0.1767,
+      "step": 9499
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6165413533834585e-05,
+      "loss": 0.0419,
+      "step": 9500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6161833154314358e-05,
+      "loss": 0.2248,
+      "step": 9501
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6158252774794128e-05,
+      "loss": 0.1478,
+      "step": 9502
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.61546723952739e-05,
+      "loss": 0.1919,
+      "step": 9503
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.615109201575367e-05,
+      "loss": 0.0391,
+      "step": 9504
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.614751163623344e-05,
+      "loss": 0.3211,
+      "step": 9505
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6143931256713214e-05,
+      "loss": 0.1291,
+      "step": 9506
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.1677,
+      "step": 9507
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6136770497672754e-05,
+      "loss": 0.1338,
+      "step": 9508
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6133190118152523e-05,
+      "loss": 0.0823,
+      "step": 9509
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6129609738632297e-05,
+      "loss": 0.1986,
+      "step": 9510
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6126029359112066e-05,
+      "loss": 0.1587,
+      "step": 9511
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6122448979591836e-05,
+      "loss": 0.1202,
+      "step": 9512
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6118868600071606e-05,
+      "loss": 0.0565,
+      "step": 9513
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.611528822055138e-05,
+      "loss": 0.0536,
+      "step": 9514
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6111707841031152e-05,
+      "loss": 0.0604,
+      "step": 9515
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6108127461510922e-05,
+      "loss": 0.1051,
+      "step": 9516
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6104547081990692e-05,
+      "loss": 0.217,
+      "step": 9517
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6100966702470462e-05,
+      "loss": 0.2215,
+      "step": 9518
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6097386322950235e-05,
+      "loss": 0.1447,
+      "step": 9519
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6093805943430005e-05,
+      "loss": 0.1311,
+      "step": 9520
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6090225563909775e-05,
+      "loss": 0.074,
+      "step": 9521
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6086645184389544e-05,
+      "loss": 0.1247,
+      "step": 9522
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6083064804869318e-05,
+      "loss": 0.3435,
+      "step": 9523
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6079484425349087e-05,
+      "loss": 0.2152,
+      "step": 9524
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6075904045828857e-05,
+      "loss": 0.1943,
+      "step": 9525
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.607232366630863e-05,
+      "loss": 0.2039,
+      "step": 9526
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.60687432867884e-05,
+      "loss": 0.0692,
+      "step": 9527
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6065162907268173e-05,
+      "loss": 0.3272,
+      "step": 9528
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6061582527747943e-05,
+      "loss": 0.1363,
+      "step": 9529
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6058002148227713e-05,
+      "loss": 0.1188,
+      "step": 9530
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6054421768707483e-05,
+      "loss": 0.1305,
+      "step": 9531
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6050841389187256e-05,
+      "loss": 0.0901,
+      "step": 9532
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6047261009667026e-05,
+      "loss": 0.1959,
+      "step": 9533
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6043680630146796e-05,
+      "loss": 0.0995,
+      "step": 9534
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6040100250626565e-05,
+      "loss": 0.0593,
+      "step": 9535
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.603651987110634e-05,
+      "loss": 0.2189,
+      "step": 9536
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.603293949158611e-05,
+      "loss": 0.0951,
+      "step": 9537
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.602935911206588e-05,
+      "loss": 0.1693,
+      "step": 9538
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.602577873254565e-05,
+      "loss": 0.1288,
+      "step": 9539
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.602219835302542e-05,
+      "loss": 0.2011,
+      "step": 9540
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6018617973505194e-05,
+      "loss": 0.1404,
+      "step": 9541
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6015037593984964e-05,
+      "loss": 0.0839,
+      "step": 9542
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6011457214464734e-05,
+      "loss": 0.0709,
+      "step": 9543
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.6007876834944504e-05,
+      "loss": 0.097,
+      "step": 9544
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6004296455424277e-05,
+      "loss": 0.0969,
+      "step": 9545
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.6000716075904047e-05,
+      "loss": 0.1391,
+      "step": 9546
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5997135696383816e-05,
+      "loss": 0.1278,
+      "step": 9547
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5993555316863586e-05,
+      "loss": 0.1446,
+      "step": 9548
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.598997493734336e-05,
+      "loss": 0.1576,
+      "step": 9549
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5986394557823133e-05,
+      "loss": 0.072,
+      "step": 9550
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5982814178302902e-05,
+      "loss": 0.0601,
+      "step": 9551
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5979233798782672e-05,
+      "loss": 0.2445,
+      "step": 9552
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5975653419262442e-05,
+      "loss": 0.2365,
+      "step": 9553
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5972073039742215e-05,
+      "loss": 0.2326,
+      "step": 9554
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5968492660221985e-05,
+      "loss": 0.1455,
+      "step": 9555
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5964912280701755e-05,
+      "loss": 0.0613,
+      "step": 9556
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5961331901181525e-05,
+      "loss": 0.2006,
+      "step": 9557
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5957751521661294e-05,
+      "loss": 0.2652,
+      "step": 9558
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5954171142141068e-05,
+      "loss": 0.1677,
+      "step": 9559
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5950590762620837e-05,
+      "loss": 0.1121,
+      "step": 9560
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.594701038310061e-05,
+      "loss": 0.1161,
+      "step": 9561
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.594343000358038e-05,
+      "loss": 0.0438,
+      "step": 9562
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5939849624060154e-05,
+      "loss": 0.0983,
+      "step": 9563
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5936269244539923e-05,
+      "loss": 0.1169,
+      "step": 9564
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5932688865019693e-05,
+      "loss": 0.1369,
+      "step": 9565
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5929108485499463e-05,
+      "loss": 0.2237,
+      "step": 9566
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5925528105979233e-05,
+      "loss": 0.0928,
+      "step": 9567
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5921947726459006e-05,
+      "loss": 0.193,
+      "step": 9568
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 0.1386,
+      "step": 9569
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5914786967418546e-05,
+      "loss": 0.1383,
+      "step": 9570
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.5911206587898315e-05,
+      "loss": 0.19,
+      "step": 9571
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.590762620837809e-05,
+      "loss": 0.0876,
+      "step": 9572
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5904045828857862e-05,
+      "loss": 0.2493,
+      "step": 9573
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.590046544933763e-05,
+      "loss": 0.1916,
+      "step": 9574
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.58968850698174e-05,
+      "loss": 0.131,
+      "step": 9575
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.589330469029717e-05,
+      "loss": 0.1525,
+      "step": 9576
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5889724310776944e-05,
+      "loss": 0.1377,
+      "step": 9577
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5886143931256714e-05,
+      "loss": 0.068,
+      "step": 9578
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5882563551736484e-05,
+      "loss": 0.1641,
+      "step": 9579
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5878983172216254e-05,
+      "loss": 0.2589,
+      "step": 9580
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5875402792696027e-05,
+      "loss": 0.0893,
+      "step": 9581
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5871822413175797e-05,
+      "loss": 0.0379,
+      "step": 9582
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5868242033655567e-05,
+      "loss": 0.081,
+      "step": 9583
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.586466165413534e-05,
+      "loss": 0.0813,
+      "step": 9584
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.586108127461511e-05,
+      "loss": 0.1763,
+      "step": 9585
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5857500895094883e-05,
+      "loss": 0.1769,
+      "step": 9586
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5853920515574652e-05,
+      "loss": 0.0554,
+      "step": 9587
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5850340136054422e-05,
+      "loss": 0.0684,
+      "step": 9588
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5846759756534192e-05,
+      "loss": 0.0971,
+      "step": 9589
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5843179377013965e-05,
+      "loss": 0.1689,
+      "step": 9590
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5839598997493735e-05,
+      "loss": 0.2297,
+      "step": 9591
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5836018617973505e-05,
+      "loss": 0.1513,
+      "step": 9592
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5832438238453275e-05,
+      "loss": 0.2153,
+      "step": 9593
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5828857858933048e-05,
+      "loss": 0.08,
+      "step": 9594
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5825277479412818e-05,
+      "loss": 0.1367,
+      "step": 9595
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.582169709989259e-05,
+      "loss": 0.1569,
+      "step": 9596
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.581811672037236e-05,
+      "loss": 0.0945,
+      "step": 9597
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.581453634085213e-05,
+      "loss": 0.1473,
+      "step": 9598
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5810955961331904e-05,
+      "loss": 0.0789,
+      "step": 9599
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5807375581811673e-05,
+      "loss": 0.0988,
+      "step": 9600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5803795202291443e-05,
+      "loss": 0.1634,
+      "step": 9601
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5800214822771213e-05,
+      "loss": 0.0717,
+      "step": 9602
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5796634443250986e-05,
+      "loss": 0.0662,
+      "step": 9603
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5793054063730756e-05,
+      "loss": 0.1316,
+      "step": 9604
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 0.1459,
+      "step": 9605
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5785893304690296e-05,
+      "loss": 0.0513,
+      "step": 9606
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.578231292517007e-05,
+      "loss": 0.0639,
+      "step": 9607
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5778732545649842e-05,
+      "loss": 0.3338,
+      "step": 9608
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5775152166129612e-05,
+      "loss": 0.1254,
+      "step": 9609
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.577157178660938e-05,
+      "loss": 0.0951,
+      "step": 9610
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.576799140708915e-05,
+      "loss": 0.1001,
+      "step": 9611
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5764411027568925e-05,
+      "loss": 0.0941,
+      "step": 9612
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5760830648048694e-05,
+      "loss": 0.0996,
+      "step": 9613
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5757250268528464e-05,
+      "loss": 0.0619,
+      "step": 9614
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5753669889008234e-05,
+      "loss": 0.0769,
+      "step": 9615
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5750089509488007e-05,
+      "loss": 0.1125,
+      "step": 9616
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5746509129967777e-05,
+      "loss": 0.1666,
+      "step": 9617
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5742928750447547e-05,
+      "loss": 0.1989,
+      "step": 9618
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.573934837092732e-05,
+      "loss": 0.1356,
+      "step": 9619
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.573576799140709e-05,
+      "loss": 0.2108,
+      "step": 9620
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5732187611886863e-05,
+      "loss": 0.1607,
+      "step": 9621
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5728607232366633e-05,
+      "loss": 0.1433,
+      "step": 9622
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5725026852846403e-05,
+      "loss": 0.0276,
+      "step": 9623
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5721446473326172e-05,
+      "loss": 0.1283,
+      "step": 9624
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5717866093805946e-05,
+      "loss": 0.0626,
+      "step": 9625
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.2052,
+      "step": 9626
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5710705334765485e-05,
+      "loss": 0.0927,
+      "step": 9627
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5707124955245255e-05,
+      "loss": 0.1137,
+      "step": 9628
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5703544575725028e-05,
+      "loss": 0.2095,
+      "step": 9629
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5699964196204798e-05,
+      "loss": 0.2026,
+      "step": 9630
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.569638381668457e-05,
+      "loss": 0.2027,
+      "step": 9631
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.569280343716434e-05,
+      "loss": 0.1445,
+      "step": 9632
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.568922305764411e-05,
+      "loss": 0.0913,
+      "step": 9633
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5685642678123884e-05,
+      "loss": 0.1537,
+      "step": 9634
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5682062298603654e-05,
+      "loss": 0.1843,
+      "step": 9635
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5678481919083423e-05,
+      "loss": 0.1313,
+      "step": 9636
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5674901539563193e-05,
+      "loss": 0.127,
+      "step": 9637
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5671321160042966e-05,
+      "loss": 0.0963,
+      "step": 9638
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5667740780522736e-05,
+      "loss": 0.1635,
+      "step": 9639
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5664160401002506e-05,
+      "loss": 0.262,
+      "step": 9640
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5660580021482276e-05,
+      "loss": 0.0932,
+      "step": 9641
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.565699964196205e-05,
+      "loss": 0.0835,
+      "step": 9642
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5653419262441822e-05,
+      "loss": 0.1226,
+      "step": 9643
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5649838882921592e-05,
+      "loss": 0.0513,
+      "step": 9644
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5646258503401362e-05,
+      "loss": 0.2038,
+      "step": 9645
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.564267812388113e-05,
+      "loss": 0.0679,
+      "step": 9646
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5639097744360905e-05,
+      "loss": 0.1168,
+      "step": 9647
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5635517364840675e-05,
+      "loss": 0.2299,
+      "step": 9648
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5631936985320444e-05,
+      "loss": 0.1475,
+      "step": 9649
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5628356605800214e-05,
+      "loss": 0.215,
+      "step": 9650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5624776226279984e-05,
+      "loss": 0.0808,
+      "step": 9651
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5621195846759757e-05,
+      "loss": 0.1536,
+      "step": 9652
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5617615467239527e-05,
+      "loss": 0.1327,
+      "step": 9653
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.56140350877193e-05,
+      "loss": 0.058,
+      "step": 9654
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.561045470819907e-05,
+      "loss": 0.1167,
+      "step": 9655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5606874328678843e-05,
+      "loss": 0.1251,
+      "step": 9656
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5603293949158613e-05,
+      "loss": 0.0575,
+      "step": 9657
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5599713569638383e-05,
+      "loss": 0.0576,
+      "step": 9658
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5596133190118153e-05,
+      "loss": 0.1258,
+      "step": 9659
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5592552810597922e-05,
+      "loss": 0.2614,
+      "step": 9660
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5588972431077696e-05,
+      "loss": 0.0847,
+      "step": 9661
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5585392051557465e-05,
+      "loss": 0.1364,
+      "step": 9662
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5581811672037235e-05,
+      "loss": 0.1031,
+      "step": 9663
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5578231292517005e-05,
+      "loss": 0.1217,
+      "step": 9664
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5574650912996778e-05,
+      "loss": 0.1504,
+      "step": 9665
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.557107053347655e-05,
+      "loss": 0.1432,
+      "step": 9666
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.556749015395632e-05,
+      "loss": 0.0447,
+      "step": 9667
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.556390977443609e-05,
+      "loss": 0.1624,
+      "step": 9668
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.556032939491586e-05,
+      "loss": 0.1217,
+      "step": 9669
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5556749015395634e-05,
+      "loss": 0.1308,
+      "step": 9670
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5553168635875404e-05,
+      "loss": 0.1179,
+      "step": 9671
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5549588256355174e-05,
+      "loss": 0.143,
+      "step": 9672
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5546007876834943e-05,
+      "loss": 0.1338,
+      "step": 9673
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5542427497314717e-05,
+      "loss": 0.2563,
+      "step": 9674
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5538847117794486e-05,
+      "loss": 0.1672,
+      "step": 9675
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5535266738274256e-05,
+      "loss": 0.3755,
+      "step": 9676
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5531686358754026e-05,
+      "loss": 0.3542,
+      "step": 9677
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.55281059792338e-05,
+      "loss": 0.0926,
+      "step": 9678
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5524525599713572e-05,
+      "loss": 0.0931,
+      "step": 9679
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5520945220193342e-05,
+      "loss": 0.103,
+      "step": 9680
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5517364840673112e-05,
+      "loss": 0.3396,
+      "step": 9681
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5513784461152882e-05,
+      "loss": 0.1376,
+      "step": 9682
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 0.2395,
+      "step": 9683
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5506623702112425e-05,
+      "loss": 0.1419,
+      "step": 9684
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5503043322592194e-05,
+      "loss": 0.0933,
+      "step": 9685
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5499462943071964e-05,
+      "loss": 0.1943,
+      "step": 9686
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5495882563551737e-05,
+      "loss": 0.134,
+      "step": 9687
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5492302184031507e-05,
+      "loss": 0.0537,
+      "step": 9688
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.548872180451128e-05,
+      "loss": 0.1807,
+      "step": 9689
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.548514142499105e-05,
+      "loss": 0.2279,
+      "step": 9690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.548156104547082e-05,
+      "loss": 0.1858,
+      "step": 9691
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5477980665950593e-05,
+      "loss": 0.087,
+      "step": 9692
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5474400286430363e-05,
+      "loss": 0.1716,
+      "step": 9693
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5470819906910133e-05,
+      "loss": 0.1632,
+      "step": 9694
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5467239527389903e-05,
+      "loss": 0.2106,
+      "step": 9695
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5463659147869676e-05,
+      "loss": 0.1915,
+      "step": 9696
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5460078768349446e-05,
+      "loss": 0.1335,
+      "step": 9697
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5456498388829215e-05,
+      "loss": 0.1645,
+      "step": 9698
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5452918009308985e-05,
+      "loss": 0.2075,
+      "step": 9699
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.544933762978876e-05,
+      "loss": 0.1039,
+      "step": 9700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.544575725026853e-05,
+      "loss": 0.1626,
+      "step": 9701
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.54421768707483e-05,
+      "loss": 0.0995,
+      "step": 9702
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.1636,
+      "step": 9703
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.543501611170784e-05,
+      "loss": 0.0975,
+      "step": 9704
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5431435732187614e-05,
+      "loss": 0.1071,
+      "step": 9705
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5427855352667384e-05,
+      "loss": 0.1544,
+      "step": 9706
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5424274973147154e-05,
+      "loss": 0.0665,
+      "step": 9707
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5420694593626924e-05,
+      "loss": 0.1066,
+      "step": 9708
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5417114214106697e-05,
+      "loss": 0.1128,
+      "step": 9709
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5413533834586467e-05,
+      "loss": 0.109,
+      "step": 9710
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5409953455066236e-05,
+      "loss": 0.1385,
+      "step": 9711
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5406373075546006e-05,
+      "loss": 0.1034,
+      "step": 9712
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.540279269602578e-05,
+      "loss": 0.0626,
+      "step": 9713
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5399212316505553e-05,
+      "loss": 0.1605,
+      "step": 9714
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5395631936985322e-05,
+      "loss": 0.1502,
+      "step": 9715
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5392051557465092e-05,
+      "loss": 0.1193,
+      "step": 9716
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5388471177944862e-05,
+      "loss": 0.1174,
+      "step": 9717
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5384890798424635e-05,
+      "loss": 0.2258,
+      "step": 9718
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5381310418904405e-05,
+      "loss": 0.155,
+      "step": 9719
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5377730039384175e-05,
+      "loss": 0.1024,
+      "step": 9720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5374149659863945e-05,
+      "loss": 0.284,
+      "step": 9721
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5370569280343718e-05,
+      "loss": 0.1409,
+      "step": 9722
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5366988900823488e-05,
+      "loss": 0.1031,
+      "step": 9723
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.536340852130326e-05,
+      "loss": 0.1534,
+      "step": 9724
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.535982814178303e-05,
+      "loss": 0.1789,
+      "step": 9725
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.53562477622628e-05,
+      "loss": 0.0981,
+      "step": 9726
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5352667382742573e-05,
+      "loss": 0.1226,
+      "step": 9727
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5349087003222343e-05,
+      "loss": 0.0734,
+      "step": 9728
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5345506623702113e-05,
+      "loss": 0.0505,
+      "step": 9729
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5341926244181883e-05,
+      "loss": 0.1478,
+      "step": 9730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5338345864661656e-05,
+      "loss": 0.0952,
+      "step": 9731
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5334765485141426e-05,
+      "loss": 0.0803,
+      "step": 9732
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5331185105621196e-05,
+      "loss": 0.1074,
+      "step": 9733
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5327604726100965e-05,
+      "loss": 0.1073,
+      "step": 9734
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.532402434658074e-05,
+      "loss": 0.1586,
+      "step": 9735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5320443967060512e-05,
+      "loss": 0.1471,
+      "step": 9736
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.531686358754028e-05,
+      "loss": 0.2553,
+      "step": 9737
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.531328320802005e-05,
+      "loss": 0.1329,
+      "step": 9738
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.530970282849982e-05,
+      "loss": 0.1826,
+      "step": 9739
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5306122448979594e-05,
+      "loss": 0.2481,
+      "step": 9740
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5302542069459364e-05,
+      "loss": 0.1292,
+      "step": 9741
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5298961689939134e-05,
+      "loss": 0.1655,
+      "step": 9742
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5295381310418904e-05,
+      "loss": 0.3045,
+      "step": 9743
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5291800930898677e-05,
+      "loss": 0.2477,
+      "step": 9744
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5288220551378447e-05,
+      "loss": 0.0578,
+      "step": 9745
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5284640171858217e-05,
+      "loss": 0.1575,
+      "step": 9746
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5281059792337986e-05,
+      "loss": 0.0735,
+      "step": 9747
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.527747941281776e-05,
+      "loss": 0.2742,
+      "step": 9748
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5273899033297533e-05,
+      "loss": 0.1388,
+      "step": 9749
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5270318653777303e-05,
+      "loss": 0.2729,
+      "step": 9750
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5266738274257072e-05,
+      "loss": 0.1191,
+      "step": 9751
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5263157894736842e-05,
+      "loss": 0.089,
+      "step": 9752
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5259577515216615e-05,
+      "loss": 0.1499,
+      "step": 9753
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5255997135696385e-05,
+      "loss": 0.1217,
+      "step": 9754
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5252416756176155e-05,
+      "loss": 0.0413,
+      "step": 9755
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5248836376655925e-05,
+      "loss": 0.2261,
+      "step": 9756
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5245255997135696e-05,
+      "loss": 0.1303,
+      "step": 9757
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5241675617615466e-05,
+      "loss": 0.2815,
+      "step": 9758
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5238095238095241e-05,
+      "loss": 0.0526,
+      "step": 9759
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.523451485857501e-05,
+      "loss": 0.0719,
+      "step": 9760
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5230934479054782e-05,
+      "loss": 0.0747,
+      "step": 9761
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5227354099534552e-05,
+      "loss": 0.1726,
+      "step": 9762
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5223773720014322e-05,
+      "loss": 0.1324,
+      "step": 9763
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5220193340494093e-05,
+      "loss": 0.0856,
+      "step": 9764
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5216612960973863e-05,
+      "loss": 0.12,
+      "step": 9765
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5213032581453635e-05,
+      "loss": 0.112,
+      "step": 9766
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5209452201933404e-05,
+      "loss": 0.0633,
+      "step": 9767
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5205871822413176e-05,
+      "loss": 0.1779,
+      "step": 9768
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5202291442892946e-05,
+      "loss": 0.1052,
+      "step": 9769
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5198711063372717e-05,
+      "loss": 0.1055,
+      "step": 9770
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.519513068385249e-05,
+      "loss": 0.0707,
+      "step": 9771
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.519155030433226e-05,
+      "loss": 0.1189,
+      "step": 9772
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5187969924812032e-05,
+      "loss": 0.0375,
+      "step": 9773
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5184389545291801e-05,
+      "loss": 0.3241,
+      "step": 9774
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5180809165771573e-05,
+      "loss": 0.121,
+      "step": 9775
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5177228786251343e-05,
+      "loss": 0.1415,
+      "step": 9776
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5173648406731114e-05,
+      "loss": 0.1431,
+      "step": 9777
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5170068027210884e-05,
+      "loss": 0.1761,
+      "step": 9778
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5166487647690656e-05,
+      "loss": 0.1861,
+      "step": 9779
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5162907268170425e-05,
+      "loss": 0.0709,
+      "step": 9780
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5159326888650197e-05,
+      "loss": 0.2627,
+      "step": 9781
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5155746509129967e-05,
+      "loss": 0.1087,
+      "step": 9782
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.515216612960974e-05,
+      "loss": 0.1806,
+      "step": 9783
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5148585750089511e-05,
+      "loss": 0.1787,
+      "step": 9784
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5145005370569281e-05,
+      "loss": 0.1333,
+      "step": 9785
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5141424991049053e-05,
+      "loss": 0.1175,
+      "step": 9786
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5137844611528822e-05,
+      "loss": 0.0806,
+      "step": 9787
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5134264232008594e-05,
+      "loss": 0.1095,
+      "step": 9788
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5130683852488364e-05,
+      "loss": 0.0782,
+      "step": 9789
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5127103472968135e-05,
+      "loss": 0.058,
+      "step": 9790
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5123523093447905e-05,
+      "loss": 0.0949,
+      "step": 9791
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5119942713927677e-05,
+      "loss": 0.1213,
+      "step": 9792
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5116362334407446e-05,
+      "loss": 0.1976,
+      "step": 9793
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.511278195488722e-05,
+      "loss": 0.1094,
+      "step": 9794
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5109201575366991e-05,
+      "loss": 0.07,
+      "step": 9795
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.510562119584676e-05,
+      "loss": 0.0909,
+      "step": 9796
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5102040816326532e-05,
+      "loss": 0.1125,
+      "step": 9797
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5098460436806302e-05,
+      "loss": 0.0852,
+      "step": 9798
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5094880057286074e-05,
+      "loss": 0.0988,
+      "step": 9799
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5091299677765843e-05,
+      "loss": 0.1372,
+      "step": 9800
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.0901,
+      "step": 9801
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5084138918725385e-05,
+      "loss": 0.1267,
+      "step": 9802
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5080558539205156e-05,
+      "loss": 0.0577,
+      "step": 9803
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5076978159684926e-05,
+      "loss": 0.2286,
+      "step": 9804
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5073397780164697e-05,
+      "loss": 0.0658,
+      "step": 9805
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.506981740064447e-05,
+      "loss": 0.2853,
+      "step": 9806
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.506623702112424e-05,
+      "loss": 0.2769,
+      "step": 9807
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5062656641604012e-05,
+      "loss": 0.1845,
+      "step": 9808
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5059076262083782e-05,
+      "loss": 0.1538,
+      "step": 9809
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5055495882563553e-05,
+      "loss": 0.2102,
+      "step": 9810
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5051915503043323e-05,
+      "loss": 0.0825,
+      "step": 9811
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5048335123523095e-05,
+      "loss": 0.1127,
+      "step": 9812
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5044754744002864e-05,
+      "loss": 0.2401,
+      "step": 9813
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5041174364482636e-05,
+      "loss": 0.0603,
+      "step": 9814
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5037593984962406e-05,
+      "loss": 0.2061,
+      "step": 9815
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5034013605442177e-05,
+      "loss": 0.191,
+      "step": 9816
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5030433225921947e-05,
+      "loss": 0.101,
+      "step": 9817
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.502685284640172e-05,
+      "loss": 0.1102,
+      "step": 9818
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5023272466881492e-05,
+      "loss": 0.053,
+      "step": 9819
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5019692087361261e-05,
+      "loss": 0.1835,
+      "step": 9820
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5016111707841033e-05,
+      "loss": 0.1399,
+      "step": 9821
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5012531328320803e-05,
+      "loss": 0.2155,
+      "step": 9822
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5008950948800574e-05,
+      "loss": 0.0407,
+      "step": 9823
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5005370569280344e-05,
+      "loss": 0.1591,
+      "step": 9824
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5001790189760115e-05,
+      "loss": 0.0917,
+      "step": 9825
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4998209810239885e-05,
+      "loss": 0.1484,
+      "step": 9826
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4994629430719657e-05,
+      "loss": 0.1247,
+      "step": 9827
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4991049051199427e-05,
+      "loss": 0.1606,
+      "step": 9828
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4987468671679198e-05,
+      "loss": 0.2122,
+      "step": 9829
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4983888292158971e-05,
+      "loss": 0.1263,
+      "step": 9830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4980307912638741e-05,
+      "loss": 0.1147,
+      "step": 9831
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4976727533118513e-05,
+      "loss": 0.2271,
+      "step": 9832
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4973147153598282e-05,
+      "loss": 0.2013,
+      "step": 9833
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4969566774078054e-05,
+      "loss": 0.1042,
+      "step": 9834
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4965986394557824e-05,
+      "loss": 0.1255,
+      "step": 9835
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4962406015037595e-05,
+      "loss": 0.1195,
+      "step": 9836
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4958825635517365e-05,
+      "loss": 0.0829,
+      "step": 9837
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4955245255997136e-05,
+      "loss": 0.2241,
+      "step": 9838
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4951664876476906e-05,
+      "loss": 0.0616,
+      "step": 9839
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4948084496956678e-05,
+      "loss": 0.1391,
+      "step": 9840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4944504117436451e-05,
+      "loss": 0.0564,
+      "step": 9841
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.494092373791622e-05,
+      "loss": 0.0656,
+      "step": 9842
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4937343358395992e-05,
+      "loss": 0.0496,
+      "step": 9843
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4933762978875762e-05,
+      "loss": 0.0924,
+      "step": 9844
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4930182599355533e-05,
+      "loss": 0.3091,
+      "step": 9845
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4926602219835303e-05,
+      "loss": 0.1271,
+      "step": 9846
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4923021840315075e-05,
+      "loss": 0.2254,
+      "step": 9847
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4919441460794845e-05,
+      "loss": 0.2215,
+      "step": 9848
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4915861081274616e-05,
+      "loss": 0.1587,
+      "step": 9849
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4912280701754386e-05,
+      "loss": 0.1753,
+      "step": 9850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4908700322234156e-05,
+      "loss": 0.0676,
+      "step": 9851
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4905119942713927e-05,
+      "loss": 0.1318,
+      "step": 9852
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.49015395631937e-05,
+      "loss": 0.0661,
+      "step": 9853
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4897959183673472e-05,
+      "loss": 0.1995,
+      "step": 9854
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4894378804153242e-05,
+      "loss": 0.1071,
+      "step": 9855
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4890798424633013e-05,
+      "loss": 0.0718,
+      "step": 9856
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4887218045112783e-05,
+      "loss": 0.0919,
+      "step": 9857
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4883637665592553e-05,
+      "loss": 0.1928,
+      "step": 9858
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4880057286072324e-05,
+      "loss": 0.092,
+      "step": 9859
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4876476906552094e-05,
+      "loss": 0.0973,
+      "step": 9860
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4872896527031866e-05,
+      "loss": 0.093,
+      "step": 9861
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4869316147511635e-05,
+      "loss": 0.1593,
+      "step": 9862
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4865735767991407e-05,
+      "loss": 0.14,
+      "step": 9863
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4862155388471177e-05,
+      "loss": 0.1082,
+      "step": 9864
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4858575008950952e-05,
+      "loss": 0.1777,
+      "step": 9865
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4854994629430721e-05,
+      "loss": 0.241,
+      "step": 9866
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4851414249910491e-05,
+      "loss": 0.2038,
+      "step": 9867
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4847833870390263e-05,
+      "loss": 0.1114,
+      "step": 9868
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4844253490870032e-05,
+      "loss": 0.2362,
+      "step": 9869
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4840673111349804e-05,
+      "loss": 0.2409,
+      "step": 9870
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4837092731829574e-05,
+      "loss": 0.1422,
+      "step": 9871
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4833512352309345e-05,
+      "loss": 0.1715,
+      "step": 9872
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4829931972789115e-05,
+      "loss": 0.1152,
+      "step": 9873
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4826351593268886e-05,
+      "loss": 0.2811,
+      "step": 9874
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4822771213748656e-05,
+      "loss": 0.1082,
+      "step": 9875
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.481919083422843e-05,
+      "loss": 0.0728,
+      "step": 9876
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4815610454708201e-05,
+      "loss": 0.2061,
+      "step": 9877
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.481203007518797e-05,
+      "loss": 0.1875,
+      "step": 9878
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4808449695667742e-05,
+      "loss": 0.0884,
+      "step": 9879
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4804869316147512e-05,
+      "loss": 0.2047,
+      "step": 9880
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4801288936627284e-05,
+      "loss": 0.103,
+      "step": 9881
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4797708557107053e-05,
+      "loss": 0.1041,
+      "step": 9882
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4794128177586825e-05,
+      "loss": 0.0652,
+      "step": 9883
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4790547798066595e-05,
+      "loss": 0.1694,
+      "step": 9884
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4786967418546366e-05,
+      "loss": 0.1301,
+      "step": 9885
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4783387039026136e-05,
+      "loss": 0.1505,
+      "step": 9886
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4779806659505907e-05,
+      "loss": 0.1158,
+      "step": 9887
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.477622627998568e-05,
+      "loss": 0.1435,
+      "step": 9888
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.477264590046545e-05,
+      "loss": 0.113,
+      "step": 9889
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4769065520945222e-05,
+      "loss": 0.1598,
+      "step": 9890
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4765485141424992e-05,
+      "loss": 0.111,
+      "step": 9891
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4761904761904763e-05,
+      "loss": 0.1161,
+      "step": 9892
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4758324382384533e-05,
+      "loss": 0.0825,
+      "step": 9893
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4754744002864304e-05,
+      "loss": 0.1366,
+      "step": 9894
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4751163623344074e-05,
+      "loss": 0.2186,
+      "step": 9895
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4747583243823846e-05,
+      "loss": 0.1925,
+      "step": 9896
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4744002864303616e-05,
+      "loss": 0.1125,
+      "step": 9897
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4740422484783387e-05,
+      "loss": 0.0332,
+      "step": 9898
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4736842105263157e-05,
+      "loss": 0.1015,
+      "step": 9899
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.473326172574293e-05,
+      "loss": 0.171,
+      "step": 9900
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4729681346222702e-05,
+      "loss": 0.2903,
+      "step": 9901
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4726100966702471e-05,
+      "loss": 0.1789,
+      "step": 9902
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4722520587182243e-05,
+      "loss": 0.1857,
+      "step": 9903
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4718940207662013e-05,
+      "loss": 0.1235,
+      "step": 9904
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4715359828141784e-05,
+      "loss": 0.2371,
+      "step": 9905
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4711779448621554e-05,
+      "loss": 0.1078,
+      "step": 9906
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4708199069101325e-05,
+      "loss": 0.1246,
+      "step": 9907
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4704618689581095e-05,
+      "loss": 0.148,
+      "step": 9908
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4701038310060867e-05,
+      "loss": 0.1036,
+      "step": 9909
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4697457930540637e-05,
+      "loss": 0.0791,
+      "step": 9910
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 0.1466,
+      "step": 9911
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4690297171500181e-05,
+      "loss": 0.0685,
+      "step": 9912
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4686716791979951e-05,
+      "loss": 0.2036,
+      "step": 9913
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4683136412459722e-05,
+      "loss": 0.134,
+      "step": 9914
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4679556032939492e-05,
+      "loss": 0.1451,
+      "step": 9915
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4675975653419264e-05,
+      "loss": 0.0957,
+      "step": 9916
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4672395273899034e-05,
+      "loss": 0.0828,
+      "step": 9917
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4668814894378805e-05,
+      "loss": 0.0637,
+      "step": 9918
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4665234514858575e-05,
+      "loss": 0.0942,
+      "step": 9919
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4661654135338346e-05,
+      "loss": 0.1648,
+      "step": 9920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4658073755818116e-05,
+      "loss": 0.2112,
+      "step": 9921
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4654493376297888e-05,
+      "loss": 0.1976,
+      "step": 9922
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4650912996777661e-05,
+      "loss": 0.2583,
+      "step": 9923
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.464733261725743e-05,
+      "loss": 0.0459,
+      "step": 9924
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4643752237737202e-05,
+      "loss": 0.1344,
+      "step": 9925
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4640171858216972e-05,
+      "loss": 0.1923,
+      "step": 9926
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4636591478696743e-05,
+      "loss": 0.0597,
+      "step": 9927
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4633011099176513e-05,
+      "loss": 0.0484,
+      "step": 9928
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4629430719656285e-05,
+      "loss": 0.0763,
+      "step": 9929
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4625850340136055e-05,
+      "loss": 0.17,
+      "step": 9930
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4622269960615826e-05,
+      "loss": 0.1871,
+      "step": 9931
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4618689581095596e-05,
+      "loss": 0.2606,
+      "step": 9932
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4615109201575367e-05,
+      "loss": 0.1565,
+      "step": 9933
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4611528822055137e-05,
+      "loss": 0.1477,
+      "step": 9934
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.460794844253491e-05,
+      "loss": 0.1204,
+      "step": 9935
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4604368063014682e-05,
+      "loss": 0.1975,
+      "step": 9936
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4600787683494452e-05,
+      "loss": 0.0863,
+      "step": 9937
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4597207303974223e-05,
+      "loss": 0.0952,
+      "step": 9938
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4593626924453993e-05,
+      "loss": 0.1442,
+      "step": 9939
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4590046544933764e-05,
+      "loss": 0.1891,
+      "step": 9940
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4586466165413534e-05,
+      "loss": 0.0977,
+      "step": 9941
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4582885785893306e-05,
+      "loss": 0.2857,
+      "step": 9942
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4579305406373075e-05,
+      "loss": 0.1035,
+      "step": 9943
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4575725026852847e-05,
+      "loss": 0.1897,
+      "step": 9944
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4572144647332617e-05,
+      "loss": 0.1072,
+      "step": 9945
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.456856426781239e-05,
+      "loss": 0.2337,
+      "step": 9946
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4564983888292161e-05,
+      "loss": 0.076,
+      "step": 9947
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4561403508771931e-05,
+      "loss": 0.1095,
+      "step": 9948
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4557823129251703e-05,
+      "loss": 0.1505,
+      "step": 9949
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4554242749731473e-05,
+      "loss": 0.1246,
+      "step": 9950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4550662370211244e-05,
+      "loss": 0.2673,
+      "step": 9951
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4547081990691014e-05,
+      "loss": 0.0806,
+      "step": 9952
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4543501611170784e-05,
+      "loss": 0.0958,
+      "step": 9953
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4539921231650555e-05,
+      "loss": 0.1213,
+      "step": 9954
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4536340852130325e-05,
+      "loss": 0.2283,
+      "step": 9955
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4532760472610096e-05,
+      "loss": 0.1073,
+      "step": 9956
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4529180093089866e-05,
+      "loss": 0.0487,
+      "step": 9957
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4525599713569641e-05,
+      "loss": 0.0824,
+      "step": 9958
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4522019334049411e-05,
+      "loss": 0.1299,
+      "step": 9959
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4518438954529182e-05,
+      "loss": 0.2141,
+      "step": 9960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4514858575008952e-05,
+      "loss": 0.0425,
+      "step": 9961
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4511278195488722e-05,
+      "loss": 0.1041,
+      "step": 9962
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4507697815968493e-05,
+      "loss": 0.1221,
+      "step": 9963
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4504117436448263e-05,
+      "loss": 0.0809,
+      "step": 9964
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4500537056928035e-05,
+      "loss": 0.0778,
+      "step": 9965
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4496956677407805e-05,
+      "loss": 0.077,
+      "step": 9966
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4493376297887576e-05,
+      "loss": 0.1054,
+      "step": 9967
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4489795918367346e-05,
+      "loss": 0.1024,
+      "step": 9968
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4486215538847117e-05,
+      "loss": 0.0971,
+      "step": 9969
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.448263515932689e-05,
+      "loss": 0.1435,
+      "step": 9970
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.447905477980666e-05,
+      "loss": 0.0609,
+      "step": 9971
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4475474400286432e-05,
+      "loss": 0.0642,
+      "step": 9972
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4471894020766202e-05,
+      "loss": 0.064,
+      "step": 9973
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4468313641245973e-05,
+      "loss": 0.1063,
+      "step": 9974
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4464733261725743e-05,
+      "loss": 0.0615,
+      "step": 9975
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4461152882205514e-05,
+      "loss": 0.1697,
+      "step": 9976
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4457572502685284e-05,
+      "loss": 0.1005,
+      "step": 9977
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4453992123165056e-05,
+      "loss": 0.1677,
+      "step": 9978
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4450411743644826e-05,
+      "loss": 0.1299,
+      "step": 9979
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4446831364124597e-05,
+      "loss": 0.1638,
+      "step": 9980
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.444325098460437e-05,
+      "loss": 0.0885,
+      "step": 9981
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.443967060508414e-05,
+      "loss": 0.1632,
+      "step": 9982
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4436090225563912e-05,
+      "loss": 0.1767,
+      "step": 9983
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4432509846043681e-05,
+      "loss": 0.1523,
+      "step": 9984
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4428929466523453e-05,
+      "loss": 0.1619,
+      "step": 9985
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4425349087003223e-05,
+      "loss": 0.0678,
+      "step": 9986
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4421768707482994e-05,
+      "loss": 0.2752,
+      "step": 9987
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4418188327962764e-05,
+      "loss": 0.1322,
+      "step": 9988
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4414607948442535e-05,
+      "loss": 0.1446,
+      "step": 9989
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4411027568922305e-05,
+      "loss": 0.2079,
+      "step": 9990
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4407447189402077e-05,
+      "loss": 0.1096,
+      "step": 9991
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4403866809881846e-05,
+      "loss": 0.1113,
+      "step": 9992
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.440028643036162e-05,
+      "loss": 0.0976,
+      "step": 9993
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4396706050841391e-05,
+      "loss": 0.1028,
+      "step": 9994
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4393125671321161e-05,
+      "loss": 0.0714,
+      "step": 9995
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4389545291800932e-05,
+      "loss": 0.1562,
+      "step": 9996
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4385964912280702e-05,
+      "loss": 0.1232,
+      "step": 9997
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4382384532760474e-05,
+      "loss": 0.4365,
+      "step": 9998
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4378804153240244e-05,
+      "loss": 0.1128,
+      "step": 9999
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4375223773720015e-05,
+      "loss": 0.1928,
+      "step": 10000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4371643394199785e-05,
+      "loss": 0.0663,
+      "step": 10001
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4368063014679556e-05,
+      "loss": 0.2075,
+      "step": 10002
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4364482635159326e-05,
+      "loss": 0.275,
+      "step": 10003
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4360902255639098e-05,
+      "loss": 0.1467,
+      "step": 10004
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.435732187611887e-05,
+      "loss": 0.2236,
+      "step": 10005
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.435374149659864e-05,
+      "loss": 0.1194,
+      "step": 10006
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4350161117078412e-05,
+      "loss": 0.1457,
+      "step": 10007
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4346580737558182e-05,
+      "loss": 0.2439,
+      "step": 10008
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4343000358037953e-05,
+      "loss": 0.2114,
+      "step": 10009
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4339419978517723e-05,
+      "loss": 0.1444,
+      "step": 10010
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4335839598997495e-05,
+      "loss": 0.1145,
+      "step": 10011
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4332259219477264e-05,
+      "loss": 0.0668,
+      "step": 10012
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4328678839957036e-05,
+      "loss": 0.1008,
+      "step": 10013
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4325098460436806e-05,
+      "loss": 0.083,
+      "step": 10014
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4321518080916577e-05,
+      "loss": 0.3096,
+      "step": 10015
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.431793770139635e-05,
+      "loss": 0.0421,
+      "step": 10016
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.431435732187612e-05,
+      "loss": 0.0849,
+      "step": 10017
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4310776942355892e-05,
+      "loss": 0.1608,
+      "step": 10018
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4307196562835662e-05,
+      "loss": 0.0736,
+      "step": 10019
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4303616183315433e-05,
+      "loss": 0.0718,
+      "step": 10020
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4300035803795203e-05,
+      "loss": 0.0874,
+      "step": 10021
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4296455424274974e-05,
+      "loss": 0.1345,
+      "step": 10022
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4292875044754744e-05,
+      "loss": 0.1388,
+      "step": 10023
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4289294665234516e-05,
+      "loss": 0.0673,
+      "step": 10024
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.1235,
+      "step": 10025
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4282133906194057e-05,
+      "loss": 0.1727,
+      "step": 10026
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4278553526673827e-05,
+      "loss": 0.2174,
+      "step": 10027
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.42749731471536e-05,
+      "loss": 0.3183,
+      "step": 10028
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4271392767633371e-05,
+      "loss": 0.1193,
+      "step": 10029
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4267812388113141e-05,
+      "loss": 0.1814,
+      "step": 10030
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4264232008592913e-05,
+      "loss": 0.262,
+      "step": 10031
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4260651629072682e-05,
+      "loss": 0.1094,
+      "step": 10032
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4257071249552454e-05,
+      "loss": 0.2678,
+      "step": 10033
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4253490870032224e-05,
+      "loss": 0.052,
+      "step": 10034
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4249910490511995e-05,
+      "loss": 0.0942,
+      "step": 10035
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4246330110991765e-05,
+      "loss": 0.092,
+      "step": 10036
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4242749731471537e-05,
+      "loss": 0.064,
+      "step": 10037
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4239169351951306e-05,
+      "loss": 0.0806,
+      "step": 10038
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4235588972431078e-05,
+      "loss": 0.2431,
+      "step": 10039
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4232008592910851e-05,
+      "loss": 0.1525,
+      "step": 10040
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4228428213390621e-05,
+      "loss": 0.121,
+      "step": 10041
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4224847833870392e-05,
+      "loss": 0.1214,
+      "step": 10042
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4221267454350162e-05,
+      "loss": 0.2492,
+      "step": 10043
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4217687074829934e-05,
+      "loss": 0.2269,
+      "step": 10044
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4214106695309703e-05,
+      "loss": 0.1087,
+      "step": 10045
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4210526315789475e-05,
+      "loss": 0.0739,
+      "step": 10046
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4206945936269245e-05,
+      "loss": 0.1608,
+      "step": 10047
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4203365556749015e-05,
+      "loss": 0.0617,
+      "step": 10048
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4199785177228786e-05,
+      "loss": 0.0462,
+      "step": 10049
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4196204797708556e-05,
+      "loss": 0.2092,
+      "step": 10050
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.419262441818833e-05,
+      "loss": 0.0831,
+      "step": 10051
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.41890440386681e-05,
+      "loss": 0.1401,
+      "step": 10052
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4185463659147872e-05,
+      "loss": 0.285,
+      "step": 10053
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4181883279627642e-05,
+      "loss": 0.0932,
+      "step": 10054
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4178302900107413e-05,
+      "loss": 0.1251,
+      "step": 10055
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4174722520587183e-05,
+      "loss": 0.0906,
+      "step": 10056
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4171142141066953e-05,
+      "loss": 0.1777,
+      "step": 10057
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4167561761546724e-05,
+      "loss": 0.1095,
+      "step": 10058
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4163981382026494e-05,
+      "loss": 0.0968,
+      "step": 10059
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4160401002506266e-05,
+      "loss": 0.0824,
+      "step": 10060
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4156820622986035e-05,
+      "loss": 0.1199,
+      "step": 10061
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4153240243465807e-05,
+      "loss": 0.1706,
+      "step": 10062
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.414965986394558e-05,
+      "loss": 0.2732,
+      "step": 10063
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.414607948442535e-05,
+      "loss": 0.0989,
+      "step": 10064
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4142499104905121e-05,
+      "loss": 0.1046,
+      "step": 10065
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4138918725384891e-05,
+      "loss": 0.1147,
+      "step": 10066
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4135338345864663e-05,
+      "loss": 0.1026,
+      "step": 10067
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4131757966344433e-05,
+      "loss": 0.1382,
+      "step": 10068
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4128177586824204e-05,
+      "loss": 0.1451,
+      "step": 10069
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4124597207303974e-05,
+      "loss": 0.1441,
+      "step": 10070
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4121016827783745e-05,
+      "loss": 0.1809,
+      "step": 10071
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4117436448263515e-05,
+      "loss": 0.203,
+      "step": 10072
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4113856068743287e-05,
+      "loss": 0.206,
+      "step": 10073
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4110275689223056e-05,
+      "loss": 0.1841,
+      "step": 10074
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.410669530970283e-05,
+      "loss": 0.3282,
+      "step": 10075
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4103114930182601e-05,
+      "loss": 0.0957,
+      "step": 10076
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4099534550662371e-05,
+      "loss": 0.1156,
+      "step": 10077
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4095954171142142e-05,
+      "loss": 0.1243,
+      "step": 10078
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4092373791621912e-05,
+      "loss": 0.2283,
+      "step": 10079
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4088793412101684e-05,
+      "loss": 0.0973,
+      "step": 10080
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4085213032581453e-05,
+      "loss": 0.0803,
+      "step": 10081
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 0.0661,
+      "step": 10082
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4078052273540995e-05,
+      "loss": 0.1107,
+      "step": 10083
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4074471894020766e-05,
+      "loss": 0.1639,
+      "step": 10084
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4070891514500536e-05,
+      "loss": 0.1546,
+      "step": 10085
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.406731113498031e-05,
+      "loss": 0.0874,
+      "step": 10086
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.406373075546008e-05,
+      "loss": 0.2625,
+      "step": 10087
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.406015037593985e-05,
+      "loss": 0.144,
+      "step": 10088
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4056569996419622e-05,
+      "loss": 0.1492,
+      "step": 10089
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4052989616899392e-05,
+      "loss": 0.2071,
+      "step": 10090
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4049409237379163e-05,
+      "loss": 0.1329,
+      "step": 10091
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4045828857858933e-05,
+      "loss": 0.1063,
+      "step": 10092
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4042248478338705e-05,
+      "loss": 0.0719,
+      "step": 10093
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4038668098818474e-05,
+      "loss": 0.1698,
+      "step": 10094
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.0755,
+      "step": 10095
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4031507339778016e-05,
+      "loss": 0.1528,
+      "step": 10096
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4027926960257787e-05,
+      "loss": 0.2103,
+      "step": 10097
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.402434658073756e-05,
+      "loss": 0.0942,
+      "step": 10098
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.402076620121733e-05,
+      "loss": 0.1145,
+      "step": 10099
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4017185821697102e-05,
+      "loss": 0.0951,
+      "step": 10100
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4013605442176872e-05,
+      "loss": 0.1583,
+      "step": 10101
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4010025062656643e-05,
+      "loss": 0.1361,
+      "step": 10102
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4006444683136413e-05,
+      "loss": 0.1495,
+      "step": 10103
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4002864303616184e-05,
+      "loss": 0.0985,
+      "step": 10104
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3999283924095954e-05,
+      "loss": 0.1833,
+      "step": 10105
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3995703544575726e-05,
+      "loss": 0.2252,
+      "step": 10106
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3992123165055495e-05,
+      "loss": 0.2401,
+      "step": 10107
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3988542785535267e-05,
+      "loss": 0.178,
+      "step": 10108
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3984962406015037e-05,
+      "loss": 0.0661,
+      "step": 10109
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.398138202649481e-05,
+      "loss": 0.1122,
+      "step": 10110
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3977801646974581e-05,
+      "loss": 0.2631,
+      "step": 10111
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3974221267454351e-05,
+      "loss": 0.1072,
+      "step": 10112
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3970640887934123e-05,
+      "loss": 0.1323,
+      "step": 10113
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3967060508413892e-05,
+      "loss": 0.1725,
+      "step": 10114
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3963480128893664e-05,
+      "loss": 0.0661,
+      "step": 10115
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3959899749373434e-05,
+      "loss": 0.1784,
+      "step": 10116
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3956319369853205e-05,
+      "loss": 0.034,
+      "step": 10117
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3952738990332975e-05,
+      "loss": 0.2823,
+      "step": 10118
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3949158610812747e-05,
+      "loss": 0.1773,
+      "step": 10119
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3945578231292516e-05,
+      "loss": 0.1972,
+      "step": 10120
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.394199785177229e-05,
+      "loss": 0.0822,
+      "step": 10121
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3938417472252061e-05,
+      "loss": 0.1881,
+      "step": 10122
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.393483709273183e-05,
+      "loss": 0.2194,
+      "step": 10123
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3931256713211602e-05,
+      "loss": 0.0779,
+      "step": 10124
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3927676333691372e-05,
+      "loss": 0.1527,
+      "step": 10125
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3924095954171144e-05,
+      "loss": 0.1684,
+      "step": 10126
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3920515574650913e-05,
+      "loss": 0.0891,
+      "step": 10127
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3916935195130685e-05,
+      "loss": 0.2362,
+      "step": 10128
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3913354815610455e-05,
+      "loss": 0.0903,
+      "step": 10129
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3909774436090226e-05,
+      "loss": 0.1179,
+      "step": 10130
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3906194056569996e-05,
+      "loss": 0.0684,
+      "step": 10131
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.3902613677049767e-05,
+      "loss": 0.1309,
+      "step": 10132
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.389903329752954e-05,
+      "loss": 0.1474,
+      "step": 10133
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.389545291800931e-05,
+      "loss": 0.3093,
+      "step": 10134
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3891872538489082e-05,
+      "loss": 0.202,
+      "step": 10135
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3888292158968852e-05,
+      "loss": 0.3507,
+      "step": 10136
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3884711779448623e-05,
+      "loss": 0.1239,
+      "step": 10137
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3881131399928393e-05,
+      "loss": 0.0937,
+      "step": 10138
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 0.1132,
+      "step": 10139
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3873970640887934e-05,
+      "loss": 0.116,
+      "step": 10140
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3870390261367706e-05,
+      "loss": 0.1751,
+      "step": 10141
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3866809881847476e-05,
+      "loss": 0.1857,
+      "step": 10142
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3863229502327245e-05,
+      "loss": 0.1705,
+      "step": 10143
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3859649122807017e-05,
+      "loss": 0.1325,
+      "step": 10144
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.385606874328679e-05,
+      "loss": 0.2003,
+      "step": 10145
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3852488363766562e-05,
+      "loss": 0.0944,
+      "step": 10146
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3848907984246331e-05,
+      "loss": 0.1318,
+      "step": 10147
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3845327604726103e-05,
+      "loss": 0.178,
+      "step": 10148
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3841747225205873e-05,
+      "loss": 0.0897,
+      "step": 10149
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3838166845685644e-05,
+      "loss": 0.1674,
+      "step": 10150
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3834586466165414e-05,
+      "loss": 0.1053,
+      "step": 10151
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3831006086645184e-05,
+      "loss": 0.1607,
+      "step": 10152
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3827425707124955e-05,
+      "loss": 0.1064,
+      "step": 10153
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3823845327604725e-05,
+      "loss": 0.134,
+      "step": 10154
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3820264948084497e-05,
+      "loss": 0.1837,
+      "step": 10155
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.381668456856427e-05,
+      "loss": 0.0992,
+      "step": 10156
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3813104189044041e-05,
+      "loss": 0.068,
+      "step": 10157
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3809523809523811e-05,
+      "loss": 0.0875,
+      "step": 10158
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3805943430003581e-05,
+      "loss": 0.2789,
+      "step": 10159
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.3802363050483352e-05,
+      "loss": 0.167,
+      "step": 10160
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3798782670963122e-05,
+      "loss": 0.1269,
+      "step": 10161
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3795202291442894e-05,
+      "loss": 0.1198,
+      "step": 10162
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3791621911922663e-05,
+      "loss": 0.1151,
+      "step": 10163
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3788041532402435e-05,
+      "loss": 0.136,
+      "step": 10164
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3784461152882205e-05,
+      "loss": 0.1712,
+      "step": 10165
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3780880773361976e-05,
+      "loss": 0.095,
+      "step": 10166
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3777300393841746e-05,
+      "loss": 0.1226,
+      "step": 10167
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.377372001432152e-05,
+      "loss": 0.137,
+      "step": 10168
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.377013963480129e-05,
+      "loss": 0.0911,
+      "step": 10169
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.376655925528106e-05,
+      "loss": 0.0654,
+      "step": 10170
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3762978875760832e-05,
+      "loss": 0.0449,
+      "step": 10171
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3759398496240602e-05,
+      "loss": 0.0843,
+      "step": 10172
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3755818116720373e-05,
+      "loss": 0.0895,
+      "step": 10173
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3752237737200143e-05,
+      "loss": 0.1564,
+      "step": 10174
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3748657357679915e-05,
+      "loss": 0.1933,
+      "step": 10175
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3745076978159684e-05,
+      "loss": 0.2202,
+      "step": 10176
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3741496598639456e-05,
+      "loss": 0.2755,
+      "step": 10177
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3737916219119226e-05,
+      "loss": 0.1896,
+      "step": 10178
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3734335839598997e-05,
+      "loss": 0.1807,
+      "step": 10179
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.373075546007877e-05,
+      "loss": 0.2009,
+      "step": 10180
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.372717508055854e-05,
+      "loss": 0.0972,
+      "step": 10181
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3723594701038312e-05,
+      "loss": 0.1526,
+      "step": 10182
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3720014321518081e-05,
+      "loss": 0.0916,
+      "step": 10183
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3716433941997853e-05,
+      "loss": 0.082,
+      "step": 10184
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3712853562477623e-05,
+      "loss": 0.096,
+      "step": 10185
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3709273182957394e-05,
+      "loss": 0.1749,
+      "step": 10186
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3705692803437164e-05,
+      "loss": 0.0499,
+      "step": 10187
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.3702112423916936e-05,
+      "loss": 0.0687,
+      "step": 10188
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3698532044396705e-05,
+      "loss": 0.151,
+      "step": 10189
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3694951664876477e-05,
+      "loss": 0.2348,
+      "step": 10190
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.369137128535625e-05,
+      "loss": 0.0749,
+      "step": 10191
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.368779090583602e-05,
+      "loss": 0.1203,
+      "step": 10192
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.1453,
+      "step": 10193
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3680630146795561e-05,
+      "loss": 0.1313,
+      "step": 10194
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3677049767275333e-05,
+      "loss": 0.0623,
+      "step": 10195
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3673469387755102e-05,
+      "loss": 0.221,
+      "step": 10196
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3669889008234874e-05,
+      "loss": 0.149,
+      "step": 10197
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3666308628714644e-05,
+      "loss": 0.1427,
+      "step": 10198
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3662728249194415e-05,
+      "loss": 0.1756,
+      "step": 10199
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3659147869674185e-05,
+      "loss": 0.1626,
+      "step": 10200
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3655567490153956e-05,
+      "loss": 0.1873,
+      "step": 10201
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3651987110633726e-05,
+      "loss": 0.1762,
+      "step": 10202
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.36484067311135e-05,
+      "loss": 0.1406,
+      "step": 10203
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3644826351593271e-05,
+      "loss": 0.1036,
+      "step": 10204
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.364124597207304e-05,
+      "loss": 0.0947,
+      "step": 10205
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3637665592552812e-05,
+      "loss": 0.2658,
+      "step": 10206
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3634085213032582e-05,
+      "loss": 0.1767,
+      "step": 10207
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3630504833512354e-05,
+      "loss": 0.2337,
+      "step": 10208
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3626924453992123e-05,
+      "loss": 0.1456,
+      "step": 10209
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3623344074471895e-05,
+      "loss": 0.1636,
+      "step": 10210
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3619763694951665e-05,
+      "loss": 0.1108,
+      "step": 10211
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3616183315431436e-05,
+      "loss": 0.128,
+      "step": 10212
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3612602935911206e-05,
+      "loss": 0.1524,
+      "step": 10213
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.3609022556390977e-05,
+      "loss": 0.1401,
+      "step": 10214
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.360544217687075e-05,
+      "loss": 0.1325,
+      "step": 10215
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.360186179735052e-05,
+      "loss": 0.1245,
+      "step": 10216
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3598281417830292e-05,
+      "loss": 0.1448,
+      "step": 10217
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3594701038310062e-05,
+      "loss": 0.24,
+      "step": 10218
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3591120658789833e-05,
+      "loss": 0.0801,
+      "step": 10219
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3587540279269603e-05,
+      "loss": 0.109,
+      "step": 10220
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3583959899749374e-05,
+      "loss": 0.2157,
+      "step": 10221
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3580379520229144e-05,
+      "loss": 0.0923,
+      "step": 10222
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3576799140708916e-05,
+      "loss": 0.3353,
+      "step": 10223
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3573218761188686e-05,
+      "loss": 0.1453,
+      "step": 10224
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3569638381668457e-05,
+      "loss": 0.0775,
+      "step": 10225
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3566058002148227e-05,
+      "loss": 0.0733,
+      "step": 10226
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3562477622628e-05,
+      "loss": 0.1677,
+      "step": 10227
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3558897243107772e-05,
+      "loss": 0.1123,
+      "step": 10228
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3555316863587541e-05,
+      "loss": 0.1639,
+      "step": 10229
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3551736484067313e-05,
+      "loss": 0.125,
+      "step": 10230
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3548156104547083e-05,
+      "loss": 0.0618,
+      "step": 10231
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3544575725026854e-05,
+      "loss": 0.0993,
+      "step": 10232
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3540995345506624e-05,
+      "loss": 0.2436,
+      "step": 10233
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3537414965986395e-05,
+      "loss": 0.1666,
+      "step": 10234
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3533834586466165e-05,
+      "loss": 0.2281,
+      "step": 10235
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3530254206945937e-05,
+      "loss": 0.0656,
+      "step": 10236
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3526673827425707e-05,
+      "loss": 0.1033,
+      "step": 10237
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.352309344790548e-05,
+      "loss": 0.1589,
+      "step": 10238
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3519513068385251e-05,
+      "loss": 0.1678,
+      "step": 10239
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3515932688865021e-05,
+      "loss": 0.1949,
+      "step": 10240
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3512352309344793e-05,
+      "loss": 0.1393,
+      "step": 10241
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3508771929824562e-05,
+      "loss": 0.0783,
+      "step": 10242
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3505191550304334e-05,
+      "loss": 0.158,
+      "step": 10243
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.3501611170784104e-05,
+      "loss": 0.0534,
+      "step": 10244
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3498030791263875e-05,
+      "loss": 0.1087,
+      "step": 10245
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3494450411743645e-05,
+      "loss": 0.1144,
+      "step": 10246
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3490870032223415e-05,
+      "loss": 0.2158,
+      "step": 10247
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3487289652703186e-05,
+      "loss": 0.1334,
+      "step": 10248
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3483709273182956e-05,
+      "loss": 0.1397,
+      "step": 10249
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3480128893662731e-05,
+      "loss": 0.0634,
+      "step": 10250
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.34765485141425e-05,
+      "loss": 0.1084,
+      "step": 10251
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3472968134622272e-05,
+      "loss": 0.1393,
+      "step": 10252
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 0.1241,
+      "step": 10253
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3465807375581813e-05,
+      "loss": 0.1101,
+      "step": 10254
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3462226996061583e-05,
+      "loss": 0.1592,
+      "step": 10255
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3458646616541353e-05,
+      "loss": 0.2484,
+      "step": 10256
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3455066237021125e-05,
+      "loss": 0.1384,
+      "step": 10257
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3451485857500894e-05,
+      "loss": 0.1336,
+      "step": 10258
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3447905477980666e-05,
+      "loss": 0.1698,
+      "step": 10259
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3444325098460436e-05,
+      "loss": 0.0612,
+      "step": 10260
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3440744718940207e-05,
+      "loss": 0.1732,
+      "step": 10261
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.343716433941998e-05,
+      "loss": 0.0345,
+      "step": 10262
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.343358395989975e-05,
+      "loss": 0.1071,
+      "step": 10263
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3430003580379522e-05,
+      "loss": 0.0878,
+      "step": 10264
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3426423200859291e-05,
+      "loss": 0.0685,
+      "step": 10265
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3422842821339063e-05,
+      "loss": 0.1133,
+      "step": 10266
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3419262441818833e-05,
+      "loss": 0.1196,
+      "step": 10267
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3415682062298604e-05,
+      "loss": 0.0804,
+      "step": 10268
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3412101682778374e-05,
+      "loss": 0.1434,
+      "step": 10269
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3408521303258145e-05,
+      "loss": 0.0879,
+      "step": 10270
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3404940923737915e-05,
+      "loss": 0.1945,
+      "step": 10271
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.3401360544217687e-05,
+      "loss": 0.2039,
+      "step": 10272
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.339778016469746e-05,
+      "loss": 0.1011,
+      "step": 10273
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.339419978517723e-05,
+      "loss": 0.2872,
+      "step": 10274
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3390619405657001e-05,
+      "loss": 0.1408,
+      "step": 10275
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3387039026136771e-05,
+      "loss": 0.1465,
+      "step": 10276
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3383458646616543e-05,
+      "loss": 0.1113,
+      "step": 10277
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3379878267096312e-05,
+      "loss": 0.1194,
+      "step": 10278
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3376297887576084e-05,
+      "loss": 0.0689,
+      "step": 10279
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3372717508055854e-05,
+      "loss": 0.1237,
+      "step": 10280
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3369137128535625e-05,
+      "loss": 0.1353,
+      "step": 10281
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3365556749015395e-05,
+      "loss": 0.132,
+      "step": 10282
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3361976369495166e-05,
+      "loss": 0.0781,
+      "step": 10283
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3358395989974936e-05,
+      "loss": 0.0617,
+      "step": 10284
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.335481561045471e-05,
+      "loss": 0.1125,
+      "step": 10285
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3351235230934481e-05,
+      "loss": 0.0515,
+      "step": 10286
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.334765485141425e-05,
+      "loss": 0.1885,
+      "step": 10287
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3344074471894022e-05,
+      "loss": 0.1251,
+      "step": 10288
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3340494092373792e-05,
+      "loss": 0.0933,
+      "step": 10289
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3336913712853563e-05,
+      "loss": 0.11,
+      "step": 10290
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.1973,
+      "step": 10291
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3329752953813105e-05,
+      "loss": 0.0767,
+      "step": 10292
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3326172574292875e-05,
+      "loss": 0.1374,
+      "step": 10293
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3322592194772646e-05,
+      "loss": 0.0899,
+      "step": 10294
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3319011815252416e-05,
+      "loss": 0.1123,
+      "step": 10295
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3315431435732187e-05,
+      "loss": 0.0787,
+      "step": 10296
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.331185105621196e-05,
+      "loss": 0.2476,
+      "step": 10297
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.330827067669173e-05,
+      "loss": 0.1906,
+      "step": 10298
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3304690297171502e-05,
+      "loss": 0.0517,
+      "step": 10299
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3301109917651272e-05,
+      "loss": 0.0842,
+      "step": 10300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.3297529538131043e-05,
+      "loss": 0.1499,
+      "step": 10301
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3293949158610813e-05,
+      "loss": 0.1407,
+      "step": 10302
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3290368779090584e-05,
+      "loss": 0.0676,
+      "step": 10303
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3286788399570354e-05,
+      "loss": 0.1044,
+      "step": 10304
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3283208020050126e-05,
+      "loss": 0.0424,
+      "step": 10305
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3279627640529896e-05,
+      "loss": 0.1396,
+      "step": 10306
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3276047261009667e-05,
+      "loss": 0.1274,
+      "step": 10307
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.327246688148944e-05,
+      "loss": 0.0537,
+      "step": 10308
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.326888650196921e-05,
+      "loss": 0.1045,
+      "step": 10309
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3265306122448982e-05,
+      "loss": 0.2062,
+      "step": 10310
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3261725742928751e-05,
+      "loss": 0.0392,
+      "step": 10311
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3258145363408523e-05,
+      "loss": 0.1406,
+      "step": 10312
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3254564983888293e-05,
+      "loss": 0.2218,
+      "step": 10313
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3250984604368064e-05,
+      "loss": 0.0877,
+      "step": 10314
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3247404224847834e-05,
+      "loss": 0.1802,
+      "step": 10315
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3243823845327605e-05,
+      "loss": 0.2465,
+      "step": 10316
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3240243465807375e-05,
+      "loss": 0.2107,
+      "step": 10317
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3236663086287147e-05,
+      "loss": 0.2403,
+      "step": 10318
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3233082706766916e-05,
+      "loss": 0.1006,
+      "step": 10319
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.322950232724669e-05,
+      "loss": 0.1492,
+      "step": 10320
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3225921947726461e-05,
+      "loss": 0.2207,
+      "step": 10321
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3222341568206231e-05,
+      "loss": 0.2284,
+      "step": 10322
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3218761188686002e-05,
+      "loss": 0.2179,
+      "step": 10323
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3215180809165772e-05,
+      "loss": 0.3131,
+      "step": 10324
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3211600429645544e-05,
+      "loss": 0.1419,
+      "step": 10325
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3208020050125314e-05,
+      "loss": 0.2663,
+      "step": 10326
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3204439670605085e-05,
+      "loss": 0.0942,
+      "step": 10327
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3200859291084855e-05,
+      "loss": 0.1517,
+      "step": 10328
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.3197278911564626e-05,
+      "loss": 0.1236,
+      "step": 10329
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3193698532044396e-05,
+      "loss": 0.2174,
+      "step": 10330
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3190118152524168e-05,
+      "loss": 0.0511,
+      "step": 10331
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.318653777300394e-05,
+      "loss": 0.0824,
+      "step": 10332
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.318295739348371e-05,
+      "loss": 0.192,
+      "step": 10333
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3179377013963482e-05,
+      "loss": 0.1746,
+      "step": 10334
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3175796634443252e-05,
+      "loss": 0.1721,
+      "step": 10335
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3172216254923023e-05,
+      "loss": 0.1887,
+      "step": 10336
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3168635875402793e-05,
+      "loss": 0.2325,
+      "step": 10337
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3165055495882565e-05,
+      "loss": 0.0905,
+      "step": 10338
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3161475116362334e-05,
+      "loss": 0.1296,
+      "step": 10339
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.1602,
+      "step": 10340
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3154314357321876e-05,
+      "loss": 0.086,
+      "step": 10341
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3150733977801646e-05,
+      "loss": 0.0606,
+      "step": 10342
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.314715359828142e-05,
+      "loss": 0.1896,
+      "step": 10343
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.314357321876119e-05,
+      "loss": 0.1436,
+      "step": 10344
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3139992839240962e-05,
+      "loss": 0.1219,
+      "step": 10345
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3136412459720732e-05,
+      "loss": 0.2131,
+      "step": 10346
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3132832080200503e-05,
+      "loss": 0.0812,
+      "step": 10347
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3129251700680273e-05,
+      "loss": 0.1372,
+      "step": 10348
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3125671321160044e-05,
+      "loss": 0.0975,
+      "step": 10349
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3122090941639814e-05,
+      "loss": 0.2059,
+      "step": 10350
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3118510562119584e-05,
+      "loss": 0.2463,
+      "step": 10351
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3114930182599355e-05,
+      "loss": 0.2157,
+      "step": 10352
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3111349803079125e-05,
+      "loss": 0.2352,
+      "step": 10353
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3107769423558897e-05,
+      "loss": 0.1137,
+      "step": 10354
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.310418904403867e-05,
+      "loss": 0.0844,
+      "step": 10355
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3100608664518441e-05,
+      "loss": 0.0775,
+      "step": 10356
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.3097028284998211e-05,
+      "loss": 0.1614,
+      "step": 10357
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3093447905477981e-05,
+      "loss": 0.1622,
+      "step": 10358
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3089867525957753e-05,
+      "loss": 0.1564,
+      "step": 10359
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3086287146437522e-05,
+      "loss": 0.1751,
+      "step": 10360
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3082706766917294e-05,
+      "loss": 0.0516,
+      "step": 10361
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3079126387397064e-05,
+      "loss": 0.0931,
+      "step": 10362
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3075546007876835e-05,
+      "loss": 0.0778,
+      "step": 10363
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3071965628356605e-05,
+      "loss": 0.1082,
+      "step": 10364
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3068385248836376e-05,
+      "loss": 0.1743,
+      "step": 10365
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3064804869316146e-05,
+      "loss": 0.1306,
+      "step": 10366
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.306122448979592e-05,
+      "loss": 0.0464,
+      "step": 10367
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3057644110275691e-05,
+      "loss": 0.1642,
+      "step": 10368
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.305406373075546e-05,
+      "loss": 0.1747,
+      "step": 10369
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3050483351235232e-05,
+      "loss": 0.1759,
+      "step": 10370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3046902971715002e-05,
+      "loss": 0.0495,
+      "step": 10371
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3043322592194773e-05,
+      "loss": 0.1518,
+      "step": 10372
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3039742212674543e-05,
+      "loss": 0.1657,
+      "step": 10373
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3036161833154315e-05,
+      "loss": 0.153,
+      "step": 10374
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3032581453634085e-05,
+      "loss": 0.0401,
+      "step": 10375
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3029001074113856e-05,
+      "loss": 0.2672,
+      "step": 10376
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3025420694593626e-05,
+      "loss": 0.0401,
+      "step": 10377
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3021840315073399e-05,
+      "loss": 0.1527,
+      "step": 10378
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.301825993555317e-05,
+      "loss": 0.1982,
+      "step": 10379
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.301467955603294e-05,
+      "loss": 0.0716,
+      "step": 10380
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3011099176512712e-05,
+      "loss": 0.124,
+      "step": 10381
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3007518796992482e-05,
+      "loss": 0.0937,
+      "step": 10382
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3003938417472253e-05,
+      "loss": 0.1646,
+      "step": 10383
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.3000358037952023e-05,
+      "loss": 0.156,
+      "step": 10384
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.2996777658431794e-05,
+      "loss": 0.1074,
+      "step": 10385
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2993197278911564e-05,
+      "loss": 0.0774,
+      "step": 10386
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2989616899391336e-05,
+      "loss": 0.1084,
+      "step": 10387
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2986036519871105e-05,
+      "loss": 0.2416,
+      "step": 10388
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2982456140350877e-05,
+      "loss": 0.0859,
+      "step": 10389
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.297887576083065e-05,
+      "loss": 0.1187,
+      "step": 10390
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.297529538131042e-05,
+      "loss": 0.1311,
+      "step": 10391
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2971715001790191e-05,
+      "loss": 0.1031,
+      "step": 10392
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2968134622269961e-05,
+      "loss": 0.2332,
+      "step": 10393
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2964554242749733e-05,
+      "loss": 0.1776,
+      "step": 10394
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2960973863229503e-05,
+      "loss": 0.15,
+      "step": 10395
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2957393483709274e-05,
+      "loss": 0.1424,
+      "step": 10396
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2953813104189044e-05,
+      "loss": 0.0341,
+      "step": 10397
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2950232724668815e-05,
+      "loss": 0.0264,
+      "step": 10398
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2946652345148585e-05,
+      "loss": 0.1313,
+      "step": 10399
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2943071965628357e-05,
+      "loss": 0.1546,
+      "step": 10400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2939491586108126e-05,
+      "loss": 0.075,
+      "step": 10401
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.29359112065879e-05,
+      "loss": 0.1867,
+      "step": 10402
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2932330827067671e-05,
+      "loss": 0.0947,
+      "step": 10403
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2928750447547441e-05,
+      "loss": 0.1737,
+      "step": 10404
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2925170068027212e-05,
+      "loss": 0.0784,
+      "step": 10405
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2921589688506982e-05,
+      "loss": 0.1173,
+      "step": 10406
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2918009308986754e-05,
+      "loss": 0.0836,
+      "step": 10407
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2914428929466523e-05,
+      "loss": 0.1345,
+      "step": 10408
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2910848549946295e-05,
+      "loss": 0.1047,
+      "step": 10409
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2907268170426065e-05,
+      "loss": 0.1959,
+      "step": 10410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2903687790905836e-05,
+      "loss": 0.1647,
+      "step": 10411
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.2900107411385606e-05,
+      "loss": 0.1066,
+      "step": 10412
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.289652703186538e-05,
+      "loss": 0.0811,
+      "step": 10413
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.289294665234515e-05,
+      "loss": 0.1531,
+      "step": 10414
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.288936627282492e-05,
+      "loss": 0.1118,
+      "step": 10415
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2885785893304692e-05,
+      "loss": 0.1212,
+      "step": 10416
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2882205513784462e-05,
+      "loss": 0.1396,
+      "step": 10417
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2878625134264233e-05,
+      "loss": 0.3426,
+      "step": 10418
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2875044754744003e-05,
+      "loss": 0.2379,
+      "step": 10419
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2871464375223775e-05,
+      "loss": 0.0967,
+      "step": 10420
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2867883995703544e-05,
+      "loss": 0.1697,
+      "step": 10421
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2864303616183316e-05,
+      "loss": 0.1605,
+      "step": 10422
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2860723236663086e-05,
+      "loss": 0.1059,
+      "step": 10423
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2857142857142857e-05,
+      "loss": 0.1509,
+      "step": 10424
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.285356247762263e-05,
+      "loss": 0.1167,
+      "step": 10425
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.28499820981024e-05,
+      "loss": 0.1499,
+      "step": 10426
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2846401718582172e-05,
+      "loss": 0.2956,
+      "step": 10427
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2842821339061942e-05,
+      "loss": 0.0488,
+      "step": 10428
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2839240959541713e-05,
+      "loss": 0.1097,
+      "step": 10429
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2835660580021483e-05,
+      "loss": 0.2898,
+      "step": 10430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2832080200501254e-05,
+      "loss": 0.1822,
+      "step": 10431
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2828499820981024e-05,
+      "loss": 0.2123,
+      "step": 10432
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2824919441460796e-05,
+      "loss": 0.0839,
+      "step": 10433
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2821339061940565e-05,
+      "loss": 0.0831,
+      "step": 10434
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2817758682420337e-05,
+      "loss": 0.0813,
+      "step": 10435
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2814178302900107e-05,
+      "loss": 0.1459,
+      "step": 10436
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.281059792337988e-05,
+      "loss": 0.0723,
+      "step": 10437
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2807017543859651e-05,
+      "loss": 0.0716,
+      "step": 10438
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2803437164339421e-05,
+      "loss": 0.1713,
+      "step": 10439
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2799856784819193e-05,
+      "loss": 0.1208,
+      "step": 10440
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.2796276405298962e-05,
+      "loss": 0.1124,
+      "step": 10441
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2792696025778734e-05,
+      "loss": 0.065,
+      "step": 10442
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2789115646258504e-05,
+      "loss": 0.1215,
+      "step": 10443
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2785535266738275e-05,
+      "loss": 0.2439,
+      "step": 10444
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2781954887218045e-05,
+      "loss": 0.1592,
+      "step": 10445
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2778374507697815e-05,
+      "loss": 0.1154,
+      "step": 10446
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2774794128177586e-05,
+      "loss": 0.1962,
+      "step": 10447
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.277121374865736e-05,
+      "loss": 0.2077,
+      "step": 10448
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2767633369137131e-05,
+      "loss": 0.1019,
+      "step": 10449
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.27640529896169e-05,
+      "loss": 0.1872,
+      "step": 10450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2760472610096672e-05,
+      "loss": 0.2508,
+      "step": 10451
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2756892230576442e-05,
+      "loss": 0.0952,
+      "step": 10452
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2753311851056212e-05,
+      "loss": 0.1195,
+      "step": 10453
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2749731471535983e-05,
+      "loss": 0.2464,
+      "step": 10454
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2746151092015753e-05,
+      "loss": 0.2358,
+      "step": 10455
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2742570712495525e-05,
+      "loss": 0.1398,
+      "step": 10456
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2738990332975294e-05,
+      "loss": 0.1855,
+      "step": 10457
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2735409953455066e-05,
+      "loss": 0.2063,
+      "step": 10458
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2731829573934836e-05,
+      "loss": 0.1138,
+      "step": 10459
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.272824919441461e-05,
+      "loss": 0.0822,
+      "step": 10460
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.272466881489438e-05,
+      "loss": 0.1666,
+      "step": 10461
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.272108843537415e-05,
+      "loss": 0.1023,
+      "step": 10462
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2717508055853922e-05,
+      "loss": 0.1403,
+      "step": 10463
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2713927676333692e-05,
+      "loss": 0.1437,
+      "step": 10464
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2710347296813463e-05,
+      "loss": 0.1329,
+      "step": 10465
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2706766917293233e-05,
+      "loss": 0.1884,
+      "step": 10466
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2703186537773004e-05,
+      "loss": 0.1381,
+      "step": 10467
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2699606158252774e-05,
+      "loss": 0.1329,
+      "step": 10468
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.2696025778732546e-05,
+      "loss": 0.2823,
+      "step": 10469
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2692445399212315e-05,
+      "loss": 0.1162,
+      "step": 10470
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2688865019692087e-05,
+      "loss": 0.1043,
+      "step": 10471
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.268528464017186e-05,
+      "loss": 0.0688,
+      "step": 10472
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.268170426065163e-05,
+      "loss": 0.0292,
+      "step": 10473
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2678123881131401e-05,
+      "loss": 0.0395,
+      "step": 10474
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2674543501611171e-05,
+      "loss": 0.0519,
+      "step": 10475
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2670963122090943e-05,
+      "loss": 0.093,
+      "step": 10476
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2667382742570713e-05,
+      "loss": 0.1775,
+      "step": 10477
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2663802363050484e-05,
+      "loss": 0.158,
+      "step": 10478
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2660221983530254e-05,
+      "loss": 0.148,
+      "step": 10479
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2656641604010025e-05,
+      "loss": 0.1835,
+      "step": 10480
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2653061224489795e-05,
+      "loss": 0.1119,
+      "step": 10481
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2649480844969567e-05,
+      "loss": 0.0627,
+      "step": 10482
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.264590046544934e-05,
+      "loss": 0.118,
+      "step": 10483
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.264232008592911e-05,
+      "loss": 0.062,
+      "step": 10484
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2638739706408881e-05,
+      "loss": 0.1665,
+      "step": 10485
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2635159326888651e-05,
+      "loss": 0.1245,
+      "step": 10486
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2631578947368422e-05,
+      "loss": 0.2127,
+      "step": 10487
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2627998567848192e-05,
+      "loss": 0.1143,
+      "step": 10488
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2624418188327964e-05,
+      "loss": 0.1564,
+      "step": 10489
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2620837808807733e-05,
+      "loss": 0.1527,
+      "step": 10490
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2617257429287505e-05,
+      "loss": 0.1037,
+      "step": 10491
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2613677049767275e-05,
+      "loss": 0.2524,
+      "step": 10492
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2610096670247046e-05,
+      "loss": 0.1943,
+      "step": 10493
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.2606516290726816e-05,
+      "loss": 0.2014,
+      "step": 10494
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.260293591120659e-05,
+      "loss": 0.1369,
+      "step": 10495
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.259935553168636e-05,
+      "loss": 0.0761,
+      "step": 10496
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.259577515216613e-05,
+      "loss": 0.2597,
+      "step": 10497
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2592194772645902e-05,
+      "loss": 0.1166,
+      "step": 10498
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2588614393125672e-05,
+      "loss": 0.161,
+      "step": 10499
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2585034013605443e-05,
+      "loss": 0.0397,
+      "step": 10500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2581453634085213e-05,
+      "loss": 0.1971,
+      "step": 10501
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2577873254564985e-05,
+      "loss": 0.1139,
+      "step": 10502
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2574292875044754e-05,
+      "loss": 0.1014,
+      "step": 10503
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2570712495524526e-05,
+      "loss": 0.092,
+      "step": 10504
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2567132116004296e-05,
+      "loss": 0.1208,
+      "step": 10505
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2563551736484067e-05,
+      "loss": 0.0965,
+      "step": 10506
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.255997135696384e-05,
+      "loss": 0.0765,
+      "step": 10507
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.255639097744361e-05,
+      "loss": 0.0988,
+      "step": 10508
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2552810597923382e-05,
+      "loss": 0.1895,
+      "step": 10509
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2549230218403151e-05,
+      "loss": 0.0955,
+      "step": 10510
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2545649838882923e-05,
+      "loss": 0.1142,
+      "step": 10511
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2542069459362693e-05,
+      "loss": 0.1272,
+      "step": 10512
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2538489079842464e-05,
+      "loss": 0.0547,
+      "step": 10513
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2534908700322234e-05,
+      "loss": 0.1464,
+      "step": 10514
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2531328320802006e-05,
+      "loss": 0.0432,
+      "step": 10515
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2527747941281775e-05,
+      "loss": 0.1737,
+      "step": 10516
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2524167561761547e-05,
+      "loss": 0.1428,
+      "step": 10517
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.252058718224132e-05,
+      "loss": 0.2997,
+      "step": 10518
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.251700680272109e-05,
+      "loss": 0.1551,
+      "step": 10519
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2513426423200861e-05,
+      "loss": 0.0274,
+      "step": 10520
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2509846043680631e-05,
+      "loss": 0.1353,
+      "step": 10521
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2506265664160403e-05,
+      "loss": 0.3116,
+      "step": 10522
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2502685284640172e-05,
+      "loss": 0.0693,
+      "step": 10523
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2499104905119944e-05,
+      "loss": 0.0445,
+      "step": 10524
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.2495524525599714e-05,
+      "loss": 0.1868,
+      "step": 10525
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2491944146079485e-05,
+      "loss": 0.0836,
+      "step": 10526
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2488363766559257e-05,
+      "loss": 0.0279,
+      "step": 10527
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2484783387039026e-05,
+      "loss": 0.1373,
+      "step": 10528
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2481203007518798e-05,
+      "loss": 0.1152,
+      "step": 10529
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2477622627998568e-05,
+      "loss": 0.1166,
+      "step": 10530
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.247404224847834e-05,
+      "loss": 0.1485,
+      "step": 10531
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2470461868958109e-05,
+      "loss": 0.0901,
+      "step": 10532
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2466881489437882e-05,
+      "loss": 0.0802,
+      "step": 10533
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2463301109917652e-05,
+      "loss": 0.0911,
+      "step": 10534
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2459720730397424e-05,
+      "loss": 0.0764,
+      "step": 10535
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2456140350877193e-05,
+      "loss": 0.0665,
+      "step": 10536
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2452559971356965e-05,
+      "loss": 0.0701,
+      "step": 10537
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2448979591836735e-05,
+      "loss": 0.1719,
+      "step": 10538
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2445399212316506e-05,
+      "loss": 0.1164,
+      "step": 10539
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2441818832796278e-05,
+      "loss": 0.1071,
+      "step": 10540
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2438238453276047e-05,
+      "loss": 0.1905,
+      "step": 10541
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2434658073755819e-05,
+      "loss": 0.1586,
+      "step": 10542
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2431077694235589e-05,
+      "loss": 0.065,
+      "step": 10543
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.242749731471536e-05,
+      "loss": 0.1944,
+      "step": 10544
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2423916935195132e-05,
+      "loss": 0.1411,
+      "step": 10545
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2420336555674903e-05,
+      "loss": 0.1187,
+      "step": 10546
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2416756176154673e-05,
+      "loss": 0.1764,
+      "step": 10547
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2413175796634443e-05,
+      "loss": 0.3071,
+      "step": 10548
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2409595417114214e-05,
+      "loss": 0.1523,
+      "step": 10549
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2406015037593984e-05,
+      "loss": 0.0639,
+      "step": 10550
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2402434658073757e-05,
+      "loss": 0.1498,
+      "step": 10551
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2398854278553527e-05,
+      "loss": 0.2299,
+      "step": 10552
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.2395273899033299e-05,
+      "loss": 0.083,
+      "step": 10553
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2391693519513068e-05,
+      "loss": 0.0877,
+      "step": 10554
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.238811313999284e-05,
+      "loss": 0.0911,
+      "step": 10555
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2384532760472611e-05,
+      "loss": 0.1461,
+      "step": 10556
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2380952380952381e-05,
+      "loss": 0.2264,
+      "step": 10557
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2377372001432153e-05,
+      "loss": 0.0927,
+      "step": 10558
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2373791621911922e-05,
+      "loss": 0.1476,
+      "step": 10559
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2370211242391694e-05,
+      "loss": 0.0913,
+      "step": 10560
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2366630862871464e-05,
+      "loss": 0.1566,
+      "step": 10561
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2363050483351237e-05,
+      "loss": 0.1524,
+      "step": 10562
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2359470103831007e-05,
+      "loss": 0.161,
+      "step": 10563
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2355889724310778e-05,
+      "loss": 0.1256,
+      "step": 10564
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2352309344790548e-05,
+      "loss": 0.1615,
+      "step": 10565
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.234872896527032e-05,
+      "loss": 0.1128,
+      "step": 10566
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.234514858575009e-05,
+      "loss": 0.1392,
+      "step": 10567
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.234156820622986e-05,
+      "loss": 0.0467,
+      "step": 10568
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2337987826709632e-05,
+      "loss": 0.1964,
+      "step": 10569
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2334407447189402e-05,
+      "loss": 0.161,
+      "step": 10570
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2330827067669174e-05,
+      "loss": 0.0656,
+      "step": 10571
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2327246688148943e-05,
+      "loss": 0.0903,
+      "step": 10572
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2323666308628715e-05,
+      "loss": 0.1923,
+      "step": 10573
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2320085929108486e-05,
+      "loss": 0.145,
+      "step": 10574
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2316505549588258e-05,
+      "loss": 0.0638,
+      "step": 10575
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2312925170068028e-05,
+      "loss": 0.0736,
+      "step": 10576
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.23093447905478e-05,
+      "loss": 0.1861,
+      "step": 10577
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2305764411027569e-05,
+      "loss": 0.1203,
+      "step": 10578
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.230218403150734e-05,
+      "loss": 0.2069,
+      "step": 10579
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2298603651987112e-05,
+      "loss": 0.0946,
+      "step": 10580
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.2295023272466882e-05,
+      "loss": 0.0433,
+      "step": 10581
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2291442892946653e-05,
+      "loss": 0.1421,
+      "step": 10582
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2287862513426423e-05,
+      "loss": 0.0523,
+      "step": 10583
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2284282133906195e-05,
+      "loss": 0.2832,
+      "step": 10584
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2280701754385964e-05,
+      "loss": 0.0552,
+      "step": 10585
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2277121374865738e-05,
+      "loss": 0.1279,
+      "step": 10586
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2273540995345507e-05,
+      "loss": 0.2159,
+      "step": 10587
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2269960615825279e-05,
+      "loss": 0.074,
+      "step": 10588
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2266380236305049e-05,
+      "loss": 0.1464,
+      "step": 10589
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.226279985678482e-05,
+      "loss": 0.1243,
+      "step": 10590
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2259219477264592e-05,
+      "loss": 0.161,
+      "step": 10591
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2255639097744361e-05,
+      "loss": 0.1355,
+      "step": 10592
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2252058718224133e-05,
+      "loss": 0.2223,
+      "step": 10593
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2248478338703903e-05,
+      "loss": 0.1205,
+      "step": 10594
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 0.1083,
+      "step": 10595
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2241317579663444e-05,
+      "loss": 0.1812,
+      "step": 10596
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2237737200143217e-05,
+      "loss": 0.0375,
+      "step": 10597
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2234156820622987e-05,
+      "loss": 0.0989,
+      "step": 10598
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2230576441102758e-05,
+      "loss": 0.2194,
+      "step": 10599
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2226996061582528e-05,
+      "loss": 0.1412,
+      "step": 10600
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2223415682062298e-05,
+      "loss": 0.1041,
+      "step": 10601
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.221983530254207e-05,
+      "loss": 0.1133,
+      "step": 10602
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2216254923021841e-05,
+      "loss": 0.2542,
+      "step": 10603
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2212674543501613e-05,
+      "loss": 0.1935,
+      "step": 10604
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2209094163981382e-05,
+      "loss": 0.2163,
+      "step": 10605
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2205513784461154e-05,
+      "loss": 0.1107,
+      "step": 10606
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2201933404940924e-05,
+      "loss": 0.1526,
+      "step": 10607
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2198353025420695e-05,
+      "loss": 0.2738,
+      "step": 10608
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.2194772645900467e-05,
+      "loss": 0.1123,
+      "step": 10609
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2191192266380236e-05,
+      "loss": 0.0382,
+      "step": 10610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2187611886860008e-05,
+      "loss": 0.1453,
+      "step": 10611
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2184031507339778e-05,
+      "loss": 0.2753,
+      "step": 10612
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.218045112781955e-05,
+      "loss": 0.1185,
+      "step": 10613
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2176870748299319e-05,
+      "loss": 0.1432,
+      "step": 10614
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2173290368779092e-05,
+      "loss": 0.202,
+      "step": 10615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2169709989258862e-05,
+      "loss": 0.1475,
+      "step": 10616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2166129609738633e-05,
+      "loss": 0.2683,
+      "step": 10617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2162549230218403e-05,
+      "loss": 0.1568,
+      "step": 10618
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2158968850698175e-05,
+      "loss": 0.1532,
+      "step": 10619
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2155388471177945e-05,
+      "loss": 0.1356,
+      "step": 10620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2151808091657716e-05,
+      "loss": 0.1652,
+      "step": 10621
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2148227712137488e-05,
+      "loss": 0.0997,
+      "step": 10622
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2144647332617257e-05,
+      "loss": 0.1762,
+      "step": 10623
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2141066953097029e-05,
+      "loss": 0.0713,
+      "step": 10624
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2137486573576799e-05,
+      "loss": 0.1484,
+      "step": 10625
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2133906194056572e-05,
+      "loss": 0.1851,
+      "step": 10626
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2130325814536342e-05,
+      "loss": 0.0663,
+      "step": 10627
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2126745435016113e-05,
+      "loss": 0.1064,
+      "step": 10628
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2123165055495883e-05,
+      "loss": 0.0998,
+      "step": 10629
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2119584675975654e-05,
+      "loss": 0.1144,
+      "step": 10630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2116004296455424e-05,
+      "loss": 0.1326,
+      "step": 10631
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2112423916935196e-05,
+      "loss": 0.1227,
+      "step": 10632
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2108843537414967e-05,
+      "loss": 0.1476,
+      "step": 10633
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2105263157894737e-05,
+      "loss": 0.1581,
+      "step": 10634
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2101682778374509e-05,
+      "loss": 0.1062,
+      "step": 10635
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.2098102398854278e-05,
+      "loss": 0.0386,
+      "step": 10636
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.209452201933405e-05,
+      "loss": 0.045,
+      "step": 10637
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2090941639813821e-05,
+      "loss": 0.26,
+      "step": 10638
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2087361260293593e-05,
+      "loss": 0.1639,
+      "step": 10639
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2083780880773363e-05,
+      "loss": 0.2309,
+      "step": 10640
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2080200501253134e-05,
+      "loss": 0.1797,
+      "step": 10641
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2076620121732904e-05,
+      "loss": 0.2797,
+      "step": 10642
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2073039742212674e-05,
+      "loss": 0.197,
+      "step": 10643
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2069459362692447e-05,
+      "loss": 0.1692,
+      "step": 10644
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2065878983172217e-05,
+      "loss": 0.1695,
+      "step": 10645
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2062298603651988e-05,
+      "loss": 0.0701,
+      "step": 10646
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2058718224131758e-05,
+      "loss": 0.168,
+      "step": 10647
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.205513784461153e-05,
+      "loss": 0.1245,
+      "step": 10648
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.20515574650913e-05,
+      "loss": 0.2385,
+      "step": 10649
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2047977085571072e-05,
+      "loss": 0.2014,
+      "step": 10650
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2044396706050842e-05,
+      "loss": 0.0384,
+      "step": 10651
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2040816326530612e-05,
+      "loss": 0.2544,
+      "step": 10652
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2037235947010384e-05,
+      "loss": 0.0805,
+      "step": 10653
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2033655567490153e-05,
+      "loss": 0.0676,
+      "step": 10654
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2030075187969925e-05,
+      "loss": 0.0596,
+      "step": 10655
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2026494808449696e-05,
+      "loss": 0.1857,
+      "step": 10656
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2022914428929468e-05,
+      "loss": 0.1611,
+      "step": 10657
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2019334049409238e-05,
+      "loss": 0.1007,
+      "step": 10658
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2015753669889009e-05,
+      "loss": 0.146,
+      "step": 10659
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2012173290368779e-05,
+      "loss": 0.1151,
+      "step": 10660
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.200859291084855e-05,
+      "loss": 0.0667,
+      "step": 10661
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2005012531328322e-05,
+      "loss": 0.114,
+      "step": 10662
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2001432151808092e-05,
+      "loss": 0.063,
+      "step": 10663
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1997851772287863e-05,
+      "loss": 0.0712,
+      "step": 10664
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.1994271392767633e-05,
+      "loss": 0.076,
+      "step": 10665
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1990691013247404e-05,
+      "loss": 0.0751,
+      "step": 10666
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1987110633727176e-05,
+      "loss": 0.16,
+      "step": 10667
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1983530254206947e-05,
+      "loss": 0.0423,
+      "step": 10668
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1979949874686717e-05,
+      "loss": 0.1573,
+      "step": 10669
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1976369495166489e-05,
+      "loss": 0.1367,
+      "step": 10670
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1972789115646259e-05,
+      "loss": 0.0675,
+      "step": 10671
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.196920873612603e-05,
+      "loss": 0.2272,
+      "step": 10672
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1965628356605802e-05,
+      "loss": 0.1134,
+      "step": 10673
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1962047977085571e-05,
+      "loss": 0.0556,
+      "step": 10674
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1958467597565343e-05,
+      "loss": 0.1116,
+      "step": 10675
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1954887218045113e-05,
+      "loss": 0.1208,
+      "step": 10676
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1951306838524884e-05,
+      "loss": 0.1663,
+      "step": 10677
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1947726459004654e-05,
+      "loss": 0.2234,
+      "step": 10678
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1944146079484427e-05,
+      "loss": 0.2375,
+      "step": 10679
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1940565699964197e-05,
+      "loss": 0.1553,
+      "step": 10680
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1936985320443968e-05,
+      "loss": 0.1328,
+      "step": 10681
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1933404940923738e-05,
+      "loss": 0.0734,
+      "step": 10682
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 0.2073,
+      "step": 10683
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.192624418188328e-05,
+      "loss": 0.1522,
+      "step": 10684
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1922663802363051e-05,
+      "loss": 0.0834,
+      "step": 10685
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1919083422842823e-05,
+      "loss": 0.1963,
+      "step": 10686
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1915503043322592e-05,
+      "loss": 0.1202,
+      "step": 10687
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1911922663802364e-05,
+      "loss": 0.0483,
+      "step": 10688
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1908342284282134e-05,
+      "loss": 0.2155,
+      "step": 10689
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.1381,
+      "step": 10690
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1901181525241677e-05,
+      "loss": 0.0624,
+      "step": 10691
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1897601145721448e-05,
+      "loss": 0.0874,
+      "step": 10692
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1894020766201218e-05,
+      "loss": 0.0804,
+      "step": 10693
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.189044038668099e-05,
+      "loss": 0.1852,
+      "step": 10694
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.188686000716076e-05,
+      "loss": 0.1859,
+      "step": 10695
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.188327962764053e-05,
+      "loss": 0.1449,
+      "step": 10696
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1879699248120302e-05,
+      "loss": 0.2357,
+      "step": 10697
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1876118868600072e-05,
+      "loss": 0.1089,
+      "step": 10698
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1872538489079843e-05,
+      "loss": 0.1495,
+      "step": 10699
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1868958109559613e-05,
+      "loss": 0.3387,
+      "step": 10700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1865377730039385e-05,
+      "loss": 0.0612,
+      "step": 10701
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1861797350519156e-05,
+      "loss": 0.2218,
+      "step": 10702
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1858216970998926e-05,
+      "loss": 0.0223,
+      "step": 10703
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1854636591478698e-05,
+      "loss": 0.0511,
+      "step": 10704
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1851056211958467e-05,
+      "loss": 0.2114,
+      "step": 10705
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1847475832438239e-05,
+      "loss": 0.1503,
+      "step": 10706
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1843895452918009e-05,
+      "loss": 0.0733,
+      "step": 10707
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1840315073397782e-05,
+      "loss": 0.0582,
+      "step": 10708
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 0.1933,
+      "step": 10709
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1833154314357323e-05,
+      "loss": 0.1881,
+      "step": 10710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1829573934837093e-05,
+      "loss": 0.0839,
+      "step": 10711
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1825993555316864e-05,
+      "loss": 0.1752,
+      "step": 10712
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1822413175796634e-05,
+      "loss": 0.1228,
+      "step": 10713
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1818832796276406e-05,
+      "loss": 0.2715,
+      "step": 10714
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1815252416756177e-05,
+      "loss": 0.0977,
+      "step": 10715
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1811672037235947e-05,
+      "loss": 0.0807,
+      "step": 10716
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1808091657715718e-05,
+      "loss": 0.1085,
+      "step": 10717
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1804511278195488e-05,
+      "loss": 0.0978,
+      "step": 10718
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.180093089867526e-05,
+      "loss": 0.103,
+      "step": 10719
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1797350519155031e-05,
+      "loss": 0.1346,
+      "step": 10720
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1793770139634803e-05,
+      "loss": 0.1537,
+      "step": 10721
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1790189760114573e-05,
+      "loss": 0.0862,
+      "step": 10722
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1786609380594344e-05,
+      "loss": 0.0815,
+      "step": 10723
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1783029001074114e-05,
+      "loss": 0.1277,
+      "step": 10724
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1779448621553885e-05,
+      "loss": 0.1242,
+      "step": 10725
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1775868242033657e-05,
+      "loss": 0.1615,
+      "step": 10726
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1772287862513427e-05,
+      "loss": 0.0834,
+      "step": 10727
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1768707482993198e-05,
+      "loss": 0.1347,
+      "step": 10728
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1765127103472968e-05,
+      "loss": 0.1065,
+      "step": 10729
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.176154672395274e-05,
+      "loss": 0.0829,
+      "step": 10730
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.175796634443251e-05,
+      "loss": 0.086,
+      "step": 10731
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1754385964912282e-05,
+      "loss": 0.0488,
+      "step": 10732
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1750805585392052e-05,
+      "loss": 0.1238,
+      "step": 10733
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1747225205871824e-05,
+      "loss": 0.0744,
+      "step": 10734
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1743644826351593e-05,
+      "loss": 0.0701,
+      "step": 10735
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1740064446831365e-05,
+      "loss": 0.1583,
+      "step": 10736
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1736484067311136e-05,
+      "loss": 0.2036,
+      "step": 10737
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1732903687790906e-05,
+      "loss": 0.2207,
+      "step": 10738
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1729323308270678e-05,
+      "loss": 0.2242,
+      "step": 10739
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1725742928750448e-05,
+      "loss": 0.0689,
+      "step": 10740
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1722162549230219e-05,
+      "loss": 0.1796,
+      "step": 10741
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1718582169709989e-05,
+      "loss": 0.0821,
+      "step": 10742
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1715001790189762e-05,
+      "loss": 0.1478,
+      "step": 10743
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1711421410669532e-05,
+      "loss": 0.1775,
+      "step": 10744
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1707841031149303e-05,
+      "loss": 0.168,
+      "step": 10745
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1704260651629073e-05,
+      "loss": 0.1069,
+      "step": 10746
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1700680272108843e-05,
+      "loss": 0.0976,
+      "step": 10747
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1697099892588614e-05,
+      "loss": 0.1626,
+      "step": 10748
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.1693519513068386e-05,
+      "loss": 0.1941,
+      "step": 10749
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1689939133548157e-05,
+      "loss": 0.0628,
+      "step": 10750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1686358754027927e-05,
+      "loss": 0.143,
+      "step": 10751
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1682778374507699e-05,
+      "loss": 0.12,
+      "step": 10752
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1679197994987469e-05,
+      "loss": 0.0802,
+      "step": 10753
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.167561761546724e-05,
+      "loss": 0.0886,
+      "step": 10754
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1672037235947012e-05,
+      "loss": 0.0836,
+      "step": 10755
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1668456856426781e-05,
+      "loss": 0.0934,
+      "step": 10756
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1664876476906553e-05,
+      "loss": 0.2337,
+      "step": 10757
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1661296097386323e-05,
+      "loss": 0.1414,
+      "step": 10758
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1657715717866094e-05,
+      "loss": 0.0906,
+      "step": 10759
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1654135338345864e-05,
+      "loss": 0.1133,
+      "step": 10760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1650554958825637e-05,
+      "loss": 0.0994,
+      "step": 10761
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1646974579305407e-05,
+      "loss": 0.0877,
+      "step": 10762
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1643394199785178e-05,
+      "loss": 0.0887,
+      "step": 10763
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1639813820264948e-05,
+      "loss": 0.2287,
+      "step": 10764
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.163623344074472e-05,
+      "loss": 0.0837,
+      "step": 10765
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.163265306122449e-05,
+      "loss": 0.0778,
+      "step": 10766
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1629072681704261e-05,
+      "loss": 0.2377,
+      "step": 10767
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1625492302184032e-05,
+      "loss": 0.1759,
+      "step": 10768
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1621911922663802e-05,
+      "loss": 0.0983,
+      "step": 10769
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1618331543143574e-05,
+      "loss": 0.167,
+      "step": 10770
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1614751163623344e-05,
+      "loss": 0.1695,
+      "step": 10771
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1611170784103117e-05,
+      "loss": 0.1058,
+      "step": 10772
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1607590404582887e-05,
+      "loss": 0.0583,
+      "step": 10773
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1604010025062658e-05,
+      "loss": 0.0782,
+      "step": 10774
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1600429645542428e-05,
+      "loss": 0.0724,
+      "step": 10775
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.15968492660222e-05,
+      "loss": 0.1749,
+      "step": 10776
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.1593268886501969e-05,
+      "loss": 0.0648,
+      "step": 10777
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.158968850698174e-05,
+      "loss": 0.066,
+      "step": 10778
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1586108127461512e-05,
+      "loss": 0.1237,
+      "step": 10779
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1582527747941282e-05,
+      "loss": 0.1626,
+      "step": 10780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.0583,
+      "step": 10781
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1575366988900823e-05,
+      "loss": 0.1173,
+      "step": 10782
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1571786609380595e-05,
+      "loss": 0.0815,
+      "step": 10783
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1568206229860366e-05,
+      "loss": 0.1816,
+      "step": 10784
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1564625850340138e-05,
+      "loss": 0.0417,
+      "step": 10785
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1561045470819907e-05,
+      "loss": 0.3272,
+      "step": 10786
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1557465091299679e-05,
+      "loss": 0.1848,
+      "step": 10787
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1553884711779449e-05,
+      "loss": 0.1968,
+      "step": 10788
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.155030433225922e-05,
+      "loss": 0.1175,
+      "step": 10789
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1546723952738992e-05,
+      "loss": 0.0865,
+      "step": 10790
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1543143573218762e-05,
+      "loss": 0.2311,
+      "step": 10791
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1539563193698533e-05,
+      "loss": 0.1212,
+      "step": 10792
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1535982814178303e-05,
+      "loss": 0.0132,
+      "step": 10793
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1532402434658074e-05,
+      "loss": 0.0647,
+      "step": 10794
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1528822055137844e-05,
+      "loss": 0.1111,
+      "step": 10795
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1525241675617617e-05,
+      "loss": 0.2656,
+      "step": 10796
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1521661296097387e-05,
+      "loss": 0.1864,
+      "step": 10797
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1518080916577157e-05,
+      "loss": 0.0711,
+      "step": 10798
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1514500537056928e-05,
+      "loss": 0.2074,
+      "step": 10799
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1510920157536698e-05,
+      "loss": 0.1574,
+      "step": 10800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.150733977801647e-05,
+      "loss": 0.0553,
+      "step": 10801
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1503759398496241e-05,
+      "loss": 0.184,
+      "step": 10802
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1500179018976013e-05,
+      "loss": 0.0885,
+      "step": 10803
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1496598639455783e-05,
+      "loss": 0.111,
+      "step": 10804
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.1493018259935554e-05,
+      "loss": 0.1155,
+      "step": 10805
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1489437880415324e-05,
+      "loss": 0.2143,
+      "step": 10806
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1485857500895095e-05,
+      "loss": 0.1644,
+      "step": 10807
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1482277121374867e-05,
+      "loss": 0.0696,
+      "step": 10808
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1478696741854637e-05,
+      "loss": 0.1478,
+      "step": 10809
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1475116362334408e-05,
+      "loss": 0.1765,
+      "step": 10810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1471535982814178e-05,
+      "loss": 0.1116,
+      "step": 10811
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.146795560329395e-05,
+      "loss": 0.1752,
+      "step": 10812
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1464375223773721e-05,
+      "loss": 0.1011,
+      "step": 10813
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1460794844253492e-05,
+      "loss": 0.0576,
+      "step": 10814
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1457214464733262e-05,
+      "loss": 0.1983,
+      "step": 10815
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1453634085213034e-05,
+      "loss": 0.1461,
+      "step": 10816
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1450053705692803e-05,
+      "loss": 0.0947,
+      "step": 10817
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1446473326172575e-05,
+      "loss": 0.1981,
+      "step": 10818
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1442892946652346e-05,
+      "loss": 0.1569,
+      "step": 10819
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1439312567132116e-05,
+      "loss": 0.0892,
+      "step": 10820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1435732187611888e-05,
+      "loss": 0.2448,
+      "step": 10821
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1432151808091658e-05,
+      "loss": 0.1669,
+      "step": 10822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.1149,
+      "step": 10823
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1424991049051199e-05,
+      "loss": 0.1524,
+      "step": 10824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1421410669530972e-05,
+      "loss": 0.0819,
+      "step": 10825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1417830290010742e-05,
+      "loss": 0.225,
+      "step": 10826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1414249910490513e-05,
+      "loss": 0.1094,
+      "step": 10827
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1410669530970283e-05,
+      "loss": 0.0844,
+      "step": 10828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1407089151450055e-05,
+      "loss": 0.1379,
+      "step": 10829
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1403508771929824e-05,
+      "loss": 0.0499,
+      "step": 10830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1399928392409596e-05,
+      "loss": 0.2005,
+      "step": 10831
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1396348012889367e-05,
+      "loss": 0.0549,
+      "step": 10832
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.1392767633369137e-05,
+      "loss": 0.2166,
+      "step": 10833
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1389187253848909e-05,
+      "loss": 0.0967,
+      "step": 10834
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1385606874328678e-05,
+      "loss": 0.2808,
+      "step": 10835
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.138202649480845e-05,
+      "loss": 0.041,
+      "step": 10836
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1378446115288221e-05,
+      "loss": 0.0754,
+      "step": 10837
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1374865735767993e-05,
+      "loss": 0.1291,
+      "step": 10838
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1371285356247763e-05,
+      "loss": 0.1446,
+      "step": 10839
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1367704976727534e-05,
+      "loss": 0.1556,
+      "step": 10840
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1364124597207304e-05,
+      "loss": 0.0349,
+      "step": 10841
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1360544217687076e-05,
+      "loss": 0.1217,
+      "step": 10842
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1356963838166847e-05,
+      "loss": 0.1595,
+      "step": 10843
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1353383458646617e-05,
+      "loss": 0.197,
+      "step": 10844
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1349803079126388e-05,
+      "loss": 0.1097,
+      "step": 10845
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1346222699606158e-05,
+      "loss": 0.1602,
+      "step": 10846
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.134264232008593e-05,
+      "loss": 0.108,
+      "step": 10847
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1339061940565701e-05,
+      "loss": 0.1507,
+      "step": 10848
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1335481561045473e-05,
+      "loss": 0.1598,
+      "step": 10849
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1331901181525242e-05,
+      "loss": 0.2009,
+      "step": 10850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1328320802005012e-05,
+      "loss": 0.1626,
+      "step": 10851
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1324740422484784e-05,
+      "loss": 0.0787,
+      "step": 10852
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1321160042964553e-05,
+      "loss": 0.1989,
+      "step": 10853
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1317579663444327e-05,
+      "loss": 0.0691,
+      "step": 10854
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1313999283924096e-05,
+      "loss": 0.2022,
+      "step": 10855
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1310418904403868e-05,
+      "loss": 0.1651,
+      "step": 10856
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1306838524883638e-05,
+      "loss": 0.0463,
+      "step": 10857
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.130325814536341e-05,
+      "loss": 0.2009,
+      "step": 10858
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1299677765843179e-05,
+      "loss": 0.0972,
+      "step": 10859
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.129609738632295e-05,
+      "loss": 0.1882,
+      "step": 10860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.1292517006802722e-05,
+      "loss": 0.0993,
+      "step": 10861
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1288936627282492e-05,
+      "loss": 0.1521,
+      "step": 10862
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1285356247762263e-05,
+      "loss": 0.0189,
+      "step": 10863
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1281775868242033e-05,
+      "loss": 0.0353,
+      "step": 10864
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1278195488721805e-05,
+      "loss": 0.2585,
+      "step": 10865
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1274615109201576e-05,
+      "loss": 0.1008,
+      "step": 10866
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1271034729681348e-05,
+      "loss": 0.1372,
+      "step": 10867
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1267454350161117e-05,
+      "loss": 0.2151,
+      "step": 10868
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1263873970640889e-05,
+      "loss": 0.0453,
+      "step": 10869
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1260293591120659e-05,
+      "loss": 0.2056,
+      "step": 10870
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.125671321160043e-05,
+      "loss": 0.1374,
+      "step": 10871
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1253132832080202e-05,
+      "loss": 0.0935,
+      "step": 10872
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1249552452559972e-05,
+      "loss": 0.0367,
+      "step": 10873
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1245972073039743e-05,
+      "loss": 0.1428,
+      "step": 10874
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1242391693519513e-05,
+      "loss": 0.0849,
+      "step": 10875
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1238811313999284e-05,
+      "loss": 0.1227,
+      "step": 10876
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1235230934479056e-05,
+      "loss": 0.1121,
+      "step": 10877
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1231650554958827e-05,
+      "loss": 0.1672,
+      "step": 10878
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 0.1561,
+      "step": 10879
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1224489795918369e-05,
+      "loss": 0.1646,
+      "step": 10880
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1220909416398138e-05,
+      "loss": 0.1614,
+      "step": 10881
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.121732903687791e-05,
+      "loss": 0.138,
+      "step": 10882
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1213748657357681e-05,
+      "loss": 0.0876,
+      "step": 10883
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1210168277837451e-05,
+      "loss": 0.1281,
+      "step": 10884
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1206587898317223e-05,
+      "loss": 0.1751,
+      "step": 10885
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1203007518796992e-05,
+      "loss": 0.0756,
+      "step": 10886
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1199427139276764e-05,
+      "loss": 0.0716,
+      "step": 10887
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1195846759756534e-05,
+      "loss": 0.1396,
+      "step": 10888
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.1192266380236307e-05,
+      "loss": 0.1685,
+      "step": 10889
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1188686000716077e-05,
+      "loss": 0.0876,
+      "step": 10890
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1185105621195848e-05,
+      "loss": 0.1851,
+      "step": 10891
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1181525241675618e-05,
+      "loss": 0.1305,
+      "step": 10892
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1177944862155388e-05,
+      "loss": 0.0661,
+      "step": 10893
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.117436448263516e-05,
+      "loss": 0.1031,
+      "step": 10894
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.117078410311493e-05,
+      "loss": 0.2143,
+      "step": 10895
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1167203723594702e-05,
+      "loss": 0.1913,
+      "step": 10896
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1163623344074472e-05,
+      "loss": 0.1438,
+      "step": 10897
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1160042964554244e-05,
+      "loss": 0.0439,
+      "step": 10898
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1156462585034013e-05,
+      "loss": 0.1399,
+      "step": 10899
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1152882205513785e-05,
+      "loss": 0.1342,
+      "step": 10900
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1149301825993556e-05,
+      "loss": 0.0295,
+      "step": 10901
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1145721446473326e-05,
+      "loss": 0.1712,
+      "step": 10902
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1142141066953098e-05,
+      "loss": 0.1667,
+      "step": 10903
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1138560687432867e-05,
+      "loss": 0.2179,
+      "step": 10904
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1134980307912639e-05,
+      "loss": 0.1133,
+      "step": 10905
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1131399928392409e-05,
+      "loss": 0.0832,
+      "step": 10906
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1127819548872182e-05,
+      "loss": 0.1579,
+      "step": 10907
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1124239169351952e-05,
+      "loss": 0.1111,
+      "step": 10908
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1120658789831723e-05,
+      "loss": 0.2123,
+      "step": 10909
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1117078410311493e-05,
+      "loss": 0.1009,
+      "step": 10910
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1113498030791265e-05,
+      "loss": 0.2038,
+      "step": 10911
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1109917651271036e-05,
+      "loss": 0.1625,
+      "step": 10912
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1106337271750806e-05,
+      "loss": 0.1821,
+      "step": 10913
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1102756892230577e-05,
+      "loss": 0.116,
+      "step": 10914
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1099176512710347e-05,
+      "loss": 0.0754,
+      "step": 10915
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1095596133190119e-05,
+      "loss": 0.1039,
+      "step": 10916
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.1092015753669888e-05,
+      "loss": 0.2146,
+      "step": 10917
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1088435374149662e-05,
+      "loss": 0.1784,
+      "step": 10918
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1084854994629431e-05,
+      "loss": 0.1676,
+      "step": 10919
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1081274615109203e-05,
+      "loss": 0.1102,
+      "step": 10920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1077694235588973e-05,
+      "loss": 0.2371,
+      "step": 10921
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1074113856068744e-05,
+      "loss": 0.1495,
+      "step": 10922
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1070533476548514e-05,
+      "loss": 0.1051,
+      "step": 10923
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1066953097028285e-05,
+      "loss": 0.0852,
+      "step": 10924
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1063372717508057e-05,
+      "loss": 0.1382,
+      "step": 10925
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1059792337987827e-05,
+      "loss": 0.1212,
+      "step": 10926
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1056211958467598e-05,
+      "loss": 0.1069,
+      "step": 10927
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1052631578947368e-05,
+      "loss": 0.0576,
+      "step": 10928
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.104905119942714e-05,
+      "loss": 0.249,
+      "step": 10929
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1045470819906911e-05,
+      "loss": 0.1305,
+      "step": 10930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1041890440386683e-05,
+      "loss": 0.1667,
+      "step": 10931
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1038310060866452e-05,
+      "loss": 0.2162,
+      "step": 10932
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1034729681346224e-05,
+      "loss": 0.0477,
+      "step": 10933
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1031149301825994e-05,
+      "loss": 0.1617,
+      "step": 10934
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1027568922305765e-05,
+      "loss": 0.2582,
+      "step": 10935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1023988542785537e-05,
+      "loss": 0.1453,
+      "step": 10936
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 0.0609,
+      "step": 10937
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1016827783745078e-05,
+      "loss": 0.0692,
+      "step": 10938
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1013247404224848e-05,
+      "loss": 0.0885,
+      "step": 10939
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.100966702470462e-05,
+      "loss": 0.2956,
+      "step": 10940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1006086645184389e-05,
+      "loss": 0.0927,
+      "step": 10941
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.1002506265664162e-05,
+      "loss": 0.1462,
+      "step": 10942
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0998925886143932e-05,
+      "loss": 0.1648,
+      "step": 10943
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0995345506623704e-05,
+      "loss": 0.1001,
+      "step": 10944
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.0991765127103473e-05,
+      "loss": 0.1291,
+      "step": 10945
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0988184747583243e-05,
+      "loss": 0.0405,
+      "step": 10946
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0984604368063015e-05,
+      "loss": 0.1246,
+      "step": 10947
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0981023988542786e-05,
+      "loss": 0.0985,
+      "step": 10948
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0977443609022558e-05,
+      "loss": 0.2385,
+      "step": 10949
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0973863229502327e-05,
+      "loss": 0.1718,
+      "step": 10950
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0970282849982099e-05,
+      "loss": 0.0831,
+      "step": 10951
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0966702470461869e-05,
+      "loss": 0.1325,
+      "step": 10952
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.096312209094164e-05,
+      "loss": 0.0777,
+      "step": 10953
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0959541711421412e-05,
+      "loss": 0.0269,
+      "step": 10954
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0955961331901181e-05,
+      "loss": 0.2658,
+      "step": 10955
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0952380952380953e-05,
+      "loss": 0.067,
+      "step": 10956
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0948800572860723e-05,
+      "loss": 0.0727,
+      "step": 10957
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0945220193340494e-05,
+      "loss": 0.1131,
+      "step": 10958
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0941639813820266e-05,
+      "loss": 0.1722,
+      "step": 10959
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0938059434300037e-05,
+      "loss": 0.1221,
+      "step": 10960
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0934479054779807e-05,
+      "loss": 0.2225,
+      "step": 10961
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0930898675259579e-05,
+      "loss": 0.2301,
+      "step": 10962
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0927318295739348e-05,
+      "loss": 0.0563,
+      "step": 10963
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.092373791621912e-05,
+      "loss": 0.0973,
+      "step": 10964
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0920157536698891e-05,
+      "loss": 0.1126,
+      "step": 10965
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0916577157178661e-05,
+      "loss": 0.1401,
+      "step": 10966
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0912996777658433e-05,
+      "loss": 0.138,
+      "step": 10967
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0909416398138202e-05,
+      "loss": 0.1502,
+      "step": 10968
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0905836018617974e-05,
+      "loss": 0.137,
+      "step": 10969
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0902255639097744e-05,
+      "loss": 0.0791,
+      "step": 10970
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0898675259577517e-05,
+      "loss": 0.0939,
+      "step": 10971
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0895094880057287e-05,
+      "loss": 0.1344,
+      "step": 10972
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.0891514500537058e-05,
+      "loss": 0.0819,
+      "step": 10973
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0887934121016828e-05,
+      "loss": 0.1034,
+      "step": 10974
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.08843537414966e-05,
+      "loss": 0.1245,
+      "step": 10975
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.088077336197637e-05,
+      "loss": 0.0425,
+      "step": 10976
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.087719298245614e-05,
+      "loss": 0.0317,
+      "step": 10977
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0873612602935912e-05,
+      "loss": 0.1933,
+      "step": 10978
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0870032223415682e-05,
+      "loss": 0.1512,
+      "step": 10979
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0866451843895454e-05,
+      "loss": 0.2127,
+      "step": 10980
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0862871464375223e-05,
+      "loss": 0.1854,
+      "step": 10981
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0859291084854995e-05,
+      "loss": 0.1444,
+      "step": 10982
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0855710705334766e-05,
+      "loss": 0.0963,
+      "step": 10983
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0852130325814538e-05,
+      "loss": 0.1313,
+      "step": 10984
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0848549946294308e-05,
+      "loss": 0.1254,
+      "step": 10985
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0844969566774079e-05,
+      "loss": 0.0497,
+      "step": 10986
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0841389187253849e-05,
+      "loss": 0.1294,
+      "step": 10987
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.083780880773362e-05,
+      "loss": 0.0594,
+      "step": 10988
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0834228428213392e-05,
+      "loss": 0.1297,
+      "step": 10989
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0830648048693162e-05,
+      "loss": 0.259,
+      "step": 10990
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0827067669172933e-05,
+      "loss": 0.0525,
+      "step": 10991
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0823487289652703e-05,
+      "loss": 0.135,
+      "step": 10992
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0819906910132474e-05,
+      "loss": 0.1035,
+      "step": 10993
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0816326530612246e-05,
+      "loss": 0.15,
+      "step": 10994
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0812746151092017e-05,
+      "loss": 0.1506,
+      "step": 10995
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0809165771571787e-05,
+      "loss": 0.129,
+      "step": 10996
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0805585392051557e-05,
+      "loss": 0.0374,
+      "step": 10997
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0802005012531329e-05,
+      "loss": 0.0632,
+      "step": 10998
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0798424633011098e-05,
+      "loss": 0.0966,
+      "step": 10999
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0794844253490872e-05,
+      "loss": 0.0849,
+      "step": 11000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.0791263873970641e-05,
+      "loss": 0.1425,
+      "step": 11001
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0787683494450413e-05,
+      "loss": 0.1712,
+      "step": 11002
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0784103114930183e-05,
+      "loss": 0.0578,
+      "step": 11003
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0780522735409954e-05,
+      "loss": 0.114,
+      "step": 11004
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0776942355889724e-05,
+      "loss": 0.0845,
+      "step": 11005
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0773361976369495e-05,
+      "loss": 0.0946,
+      "step": 11006
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0769781596849267e-05,
+      "loss": 0.1154,
+      "step": 11007
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0766201217329037e-05,
+      "loss": 0.1238,
+      "step": 11008
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0762620837808808e-05,
+      "loss": 0.1548,
+      "step": 11009
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0759040458288578e-05,
+      "loss": 0.0754,
+      "step": 11010
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.075546007876835e-05,
+      "loss": 0.1876,
+      "step": 11011
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0751879699248121e-05,
+      "loss": 0.1433,
+      "step": 11012
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0748299319727893e-05,
+      "loss": 0.3242,
+      "step": 11013
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0744718940207662e-05,
+      "loss": 0.0894,
+      "step": 11014
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0741138560687434e-05,
+      "loss": 0.0677,
+      "step": 11015
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0737558181167204e-05,
+      "loss": 0.1304,
+      "step": 11016
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0733977801646975e-05,
+      "loss": 0.1908,
+      "step": 11017
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0730397422126747e-05,
+      "loss": 0.1467,
+      "step": 11018
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0726817042606516e-05,
+      "loss": 0.0938,
+      "step": 11019
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0723236663086288e-05,
+      "loss": 0.169,
+      "step": 11020
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0719656283566058e-05,
+      "loss": 0.2025,
+      "step": 11021
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.071607590404583e-05,
+      "loss": 0.1438,
+      "step": 11022
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.07124955245256e-05,
+      "loss": 0.1163,
+      "step": 11023
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0708915145005372e-05,
+      "loss": 0.2462,
+      "step": 11024
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0705334765485142e-05,
+      "loss": 0.0946,
+      "step": 11025
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0701754385964913e-05,
+      "loss": 0.2333,
+      "step": 11026
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0698174006444683e-05,
+      "loss": 0.1733,
+      "step": 11027
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0694593626924455e-05,
+      "loss": 0.1509,
+      "step": 11028
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.0691013247404226e-05,
+      "loss": 0.0567,
+      "step": 11029
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0687432867883996e-05,
+      "loss": 0.2422,
+      "step": 11030
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0683852488363768e-05,
+      "loss": 0.234,
+      "step": 11031
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0680272108843537e-05,
+      "loss": 0.1637,
+      "step": 11032
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0676691729323309e-05,
+      "loss": 0.1156,
+      "step": 11033
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0673111349803079e-05,
+      "loss": 0.1264,
+      "step": 11034
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0669530970282852e-05,
+      "loss": 0.0483,
+      "step": 11035
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0665950590762622e-05,
+      "loss": 0.0693,
+      "step": 11036
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0662370211242393e-05,
+      "loss": 0.154,
+      "step": 11037
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0658789831722163e-05,
+      "loss": 0.0853,
+      "step": 11038
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0655209452201934e-05,
+      "loss": 0.0785,
+      "step": 11039
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0651629072681704e-05,
+      "loss": 0.3243,
+      "step": 11040
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0648048693161476e-05,
+      "loss": 0.0941,
+      "step": 11041
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0644468313641247e-05,
+      "loss": 0.1049,
+      "step": 11042
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0640887934121017e-05,
+      "loss": 0.1344,
+      "step": 11043
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0637307554600788e-05,
+      "loss": 0.2271,
+      "step": 11044
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0633727175080558e-05,
+      "loss": 0.035,
+      "step": 11045
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.063014679556033e-05,
+      "loss": 0.1308,
+      "step": 11046
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0626566416040101e-05,
+      "loss": 0.1091,
+      "step": 11047
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0622986036519871e-05,
+      "loss": 0.1374,
+      "step": 11048
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0619405656999643e-05,
+      "loss": 0.229,
+      "step": 11049
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0615825277479412e-05,
+      "loss": 0.1262,
+      "step": 11050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0612244897959184e-05,
+      "loss": 0.1049,
+      "step": 11051
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0608664518438954e-05,
+      "loss": 0.1064,
+      "step": 11052
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0605084138918727e-05,
+      "loss": 0.1653,
+      "step": 11053
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0601503759398497e-05,
+      "loss": 0.1411,
+      "step": 11054
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0597923379878268e-05,
+      "loss": 0.1359,
+      "step": 11055
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0594343000358038e-05,
+      "loss": 0.1037,
+      "step": 11056
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.059076262083781e-05,
+      "loss": 0.0717,
+      "step": 11057
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0587182241317581e-05,
+      "loss": 0.0571,
+      "step": 11058
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.058360186179735e-05,
+      "loss": 0.0848,
+      "step": 11059
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0580021482277122e-05,
+      "loss": 0.0391,
+      "step": 11060
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0576441102756892e-05,
+      "loss": 0.1219,
+      "step": 11061
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0572860723236664e-05,
+      "loss": 0.1837,
+      "step": 11062
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0569280343716433e-05,
+      "loss": 0.0775,
+      "step": 11063
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0565699964196206e-05,
+      "loss": 0.097,
+      "step": 11064
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0562119584675976e-05,
+      "loss": 0.0876,
+      "step": 11065
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0558539205155748e-05,
+      "loss": 0.0723,
+      "step": 11066
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0554958825635518e-05,
+      "loss": 0.1367,
+      "step": 11067
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0551378446115289e-05,
+      "loss": 0.1057,
+      "step": 11068
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0547798066595059e-05,
+      "loss": 0.1807,
+      "step": 11069
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.054421768707483e-05,
+      "loss": 0.0468,
+      "step": 11070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0540637307554602e-05,
+      "loss": 0.1638,
+      "step": 11071
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0537056928034372e-05,
+      "loss": 0.0822,
+      "step": 11072
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0533476548514143e-05,
+      "loss": 0.2564,
+      "step": 11073
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0529896168993913e-05,
+      "loss": 0.1592,
+      "step": 11074
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.1109,
+      "step": 11075
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0522735409953456e-05,
+      "loss": 0.1285,
+      "step": 11076
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0519155030433227e-05,
+      "loss": 0.1778,
+      "step": 11077
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0515574650912997e-05,
+      "loss": 0.2428,
+      "step": 11078
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0511994271392769e-05,
+      "loss": 0.1625,
+      "step": 11079
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0508413891872539e-05,
+      "loss": 0.1123,
+      "step": 11080
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.050483351235231e-05,
+      "loss": 0.1852,
+      "step": 11081
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0501253132832082e-05,
+      "loss": 0.1155,
+      "step": 11082
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0497672753311851e-05,
+      "loss": 0.2285,
+      "step": 11083
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0494092373791623e-05,
+      "loss": 0.1008,
+      "step": 11084
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.0490511994271393e-05,
+      "loss": 0.102,
+      "step": 11085
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0486931614751164e-05,
+      "loss": 0.1086,
+      "step": 11086
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0483351235230934e-05,
+      "loss": 0.2441,
+      "step": 11087
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0479770855710707e-05,
+      "loss": 0.2192,
+      "step": 11088
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.0856,
+      "step": 11089
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0472610096670248e-05,
+      "loss": 0.0953,
+      "step": 11090
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0469029717150018e-05,
+      "loss": 0.161,
+      "step": 11091
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0465449337629788e-05,
+      "loss": 0.1072,
+      "step": 11092
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0461868958109561e-05,
+      "loss": 0.0978,
+      "step": 11093
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0458288578589331e-05,
+      "loss": 0.161,
+      "step": 11094
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0454708199069102e-05,
+      "loss": 0.2258,
+      "step": 11095
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0451127819548872e-05,
+      "loss": 0.0997,
+      "step": 11096
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0447547440028644e-05,
+      "loss": 0.0816,
+      "step": 11097
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0443967060508414e-05,
+      "loss": 0.2451,
+      "step": 11098
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0440386680988187e-05,
+      "loss": 0.1661,
+      "step": 11099
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0436806301467957e-05,
+      "loss": 0.1051,
+      "step": 11100
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0433225921947726e-05,
+      "loss": 0.1764,
+      "step": 11101
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0429645542427498e-05,
+      "loss": 0.2032,
+      "step": 11102
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0426065162907268e-05,
+      "loss": 0.0473,
+      "step": 11103
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0422484783387039e-05,
+      "loss": 0.213,
+      "step": 11104
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.041890440386681e-05,
+      "loss": 0.2438,
+      "step": 11105
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0415324024346582e-05,
+      "loss": 0.0816,
+      "step": 11106
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0411743644826352e-05,
+      "loss": 0.1011,
+      "step": 11107
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 0.1218,
+      "step": 11108
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0404582885785893e-05,
+      "loss": 0.051,
+      "step": 11109
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0401002506265665e-05,
+      "loss": 0.1198,
+      "step": 11110
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0397422126745436e-05,
+      "loss": 0.157,
+      "step": 11111
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0393841747225206e-05,
+      "loss": 0.1873,
+      "step": 11112
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0390261367704977e-05,
+      "loss": 0.0575,
+      "step": 11113
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0386680988184747e-05,
+      "loss": 0.1225,
+      "step": 11114
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0383100608664519e-05,
+      "loss": 0.2039,
+      "step": 11115
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0379520229144289e-05,
+      "loss": 0.2115,
+      "step": 11116
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0375939849624062e-05,
+      "loss": 0.1269,
+      "step": 11117
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0372359470103832e-05,
+      "loss": 0.1533,
+      "step": 11118
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0368779090583603e-05,
+      "loss": 0.0408,
+      "step": 11119
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0365198711063373e-05,
+      "loss": 0.075,
+      "step": 11120
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0361618331543144e-05,
+      "loss": 0.1859,
+      "step": 11121
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0358037952022914e-05,
+      "loss": 0.0865,
+      "step": 11122
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0354457572502686e-05,
+      "loss": 0.0977,
+      "step": 11123
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0350877192982457e-05,
+      "loss": 0.0353,
+      "step": 11124
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0347296813462227e-05,
+      "loss": 0.0601,
+      "step": 11125
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0343716433941998e-05,
+      "loss": 0.1301,
+      "step": 11126
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0340136054421768e-05,
+      "loss": 0.0613,
+      "step": 11127
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.033655567490154e-05,
+      "loss": 0.0785,
+      "step": 11128
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0332975295381311e-05,
+      "loss": 0.1258,
+      "step": 11129
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0329394915861083e-05,
+      "loss": 0.0269,
+      "step": 11130
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0325814536340853e-05,
+      "loss": 0.0898,
+      "step": 11131
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0322234156820624e-05,
+      "loss": 0.1004,
+      "step": 11132
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0318653777300394e-05,
+      "loss": 0.0498,
+      "step": 11133
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0315073397780165e-05,
+      "loss": 0.1675,
+      "step": 11134
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0311493018259937e-05,
+      "loss": 0.1072,
+      "step": 11135
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0307912638739707e-05,
+      "loss": 0.098,
+      "step": 11136
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0304332259219478e-05,
+      "loss": 0.0607,
+      "step": 11137
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0300751879699248e-05,
+      "loss": 0.1438,
+      "step": 11138
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.029717150017902e-05,
+      "loss": 0.345,
+      "step": 11139
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0293591120658791e-05,
+      "loss": 0.1714,
+      "step": 11140
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0290010741138562e-05,
+      "loss": 0.1904,
+      "step": 11141
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0286430361618332e-05,
+      "loss": 0.0975,
+      "step": 11142
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0282849982098102e-05,
+      "loss": 0.1168,
+      "step": 11143
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0279269602577873e-05,
+      "loss": 0.0515,
+      "step": 11144
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0275689223057643e-05,
+      "loss": 0.0903,
+      "step": 11145
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0272108843537416e-05,
+      "loss": 0.0684,
+      "step": 11146
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0268528464017186e-05,
+      "loss": 0.2624,
+      "step": 11147
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0264948084496958e-05,
+      "loss": 0.2139,
+      "step": 11148
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0261367704976728e-05,
+      "loss": 0.0831,
+      "step": 11149
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0257787325456499e-05,
+      "loss": 0.0718,
+      "step": 11150
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0254206945936269e-05,
+      "loss": 0.2784,
+      "step": 11151
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.025062656641604e-05,
+      "loss": 0.1762,
+      "step": 11152
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0247046186895812e-05,
+      "loss": 0.2004,
+      "step": 11153
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0243465807375582e-05,
+      "loss": 0.2536,
+      "step": 11154
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0239885427855353e-05,
+      "loss": 0.2249,
+      "step": 11155
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0236305048335123e-05,
+      "loss": 0.0774,
+      "step": 11156
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0232724668814894e-05,
+      "loss": 0.0505,
+      "step": 11157
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0229144289294666e-05,
+      "loss": 0.1335,
+      "step": 11158
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0225563909774437e-05,
+      "loss": 0.1675,
+      "step": 11159
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0221983530254207e-05,
+      "loss": 0.086,
+      "step": 11160
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0218403150733979e-05,
+      "loss": 0.1611,
+      "step": 11161
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0214822771213748e-05,
+      "loss": 0.0671,
+      "step": 11162
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.021124239169352e-05,
+      "loss": 0.0711,
+      "step": 11163
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0207662012173291e-05,
+      "loss": 0.175,
+      "step": 11164
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0204081632653061e-05,
+      "loss": 0.1553,
+      "step": 11165
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0200501253132833e-05,
+      "loss": 0.2588,
+      "step": 11166
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0196920873612603e-05,
+      "loss": 0.1101,
+      "step": 11167
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0193340494092374e-05,
+      "loss": 0.2285,
+      "step": 11168
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0189760114572146e-05,
+      "loss": 0.2283,
+      "step": 11169
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0186179735051917e-05,
+      "loss": 0.0934,
+      "step": 11170
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0182599355531687e-05,
+      "loss": 0.1262,
+      "step": 11171
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0179018976011458e-05,
+      "loss": 0.1151,
+      "step": 11172
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0175438596491228e-05,
+      "loss": 0.0932,
+      "step": 11173
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0171858216971e-05,
+      "loss": 0.1549,
+      "step": 11174
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0168277837450771e-05,
+      "loss": 0.0703,
+      "step": 11175
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0164697457930541e-05,
+      "loss": 0.2168,
+      "step": 11176
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0161117078410312e-05,
+      "loss": 0.0733,
+      "step": 11177
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0157536698890082e-05,
+      "loss": 0.0792,
+      "step": 11178
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0153956319369854e-05,
+      "loss": 0.0523,
+      "step": 11179
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0150375939849624e-05,
+      "loss": 0.2016,
+      "step": 11180
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0146795560329397e-05,
+      "loss": 0.0671,
+      "step": 11181
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0143215180809166e-05,
+      "loss": 0.1304,
+      "step": 11182
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0139634801288938e-05,
+      "loss": 0.0511,
+      "step": 11183
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0136054421768708e-05,
+      "loss": 0.0855,
+      "step": 11184
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.013247404224848e-05,
+      "loss": 0.138,
+      "step": 11185
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0128893662728249e-05,
+      "loss": 0.2063,
+      "step": 11186
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.012531328320802e-05,
+      "loss": 0.0781,
+      "step": 11187
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0121732903687792e-05,
+      "loss": 0.0696,
+      "step": 11188
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0118152524167562e-05,
+      "loss": 0.0895,
+      "step": 11189
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0114572144647333e-05,
+      "loss": 0.1675,
+      "step": 11190
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0110991765127103e-05,
+      "loss": 0.1974,
+      "step": 11191
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0107411385606875e-05,
+      "loss": 0.0345,
+      "step": 11192
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0103831006086646e-05,
+      "loss": 0.146,
+      "step": 11193
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0100250626566418e-05,
+      "loss": 0.1987,
+      "step": 11194
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0096670247046187e-05,
+      "loss": 0.135,
+      "step": 11195
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0093089867525957e-05,
+      "loss": 0.1666,
+      "step": 11196
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0089509488005729e-05,
+      "loss": 0.0452,
+      "step": 11197
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0085929108485499e-05,
+      "loss": 0.1037,
+      "step": 11198
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0082348728965272e-05,
+      "loss": 0.1172,
+      "step": 11199
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0078768349445042e-05,
+      "loss": 0.0896,
+      "step": 11200
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0075187969924813e-05,
+      "loss": 0.1171,
+      "step": 11201
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0071607590404583e-05,
+      "loss": 0.2304,
+      "step": 11202
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0068027210884354e-05,
+      "loss": 0.0713,
+      "step": 11203
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0064446831364126e-05,
+      "loss": 0.1335,
+      "step": 11204
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0060866451843896e-05,
+      "loss": 0.0641,
+      "step": 11205
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0057286072323667e-05,
+      "loss": 0.1218,
+      "step": 11206
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0053705692803437e-05,
+      "loss": 0.1312,
+      "step": 11207
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0050125313283208e-05,
+      "loss": 0.1206,
+      "step": 11208
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0046544933762978e-05,
+      "loss": 0.0747,
+      "step": 11209
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0042964554242751e-05,
+      "loss": 0.088,
+      "step": 11210
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0039384174722521e-05,
+      "loss": 0.1857,
+      "step": 11211
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0035803795202293e-05,
+      "loss": 0.203,
+      "step": 11212
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.8820604303712789,
+      "eval_loss": 0.3807993233203888,
+      "eval_runtime": 24.9955,
+      "eval_samples_per_second": 1794.121,
+      "eval_steps_per_second": 28.045,
+      "step": 11212
+    }
+  ],
+  "max_steps": 14015,
+  "num_train_epochs": 5,
+  "total_flos": 5957316866772156.0,
+  "trial_name": null,
+  "trial_params": null
+}