diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100755--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,74188 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9996359665089187,
+  "eval_steps": 500,
+  "global_step": 12360,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.8414,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3432167232451018e-06,
+      "loss": 0.7451,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7139106374061905e-06,
+      "loss": 0.7412,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6864334464902035e-06,
+      "loss": 0.7562,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4407807421127065e-06,
+      "loss": 0.6829,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.057127360651292e-06,
+      "loss": 0.6537,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.578241001449828e-06,
+      "loss": 0.6615,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.029650169735306e-06,
+      "loss": 0.6438,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.427821274812381e-06,
+      "loss": 0.6301,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.78399746535781e-06,
+      "loss": 0.6362,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.106198021713788e-06,
+      "loss": 0.5952,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400344083896394e-06,
+      "loss": 0.6229,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.6709322309086e-06,
+      "loss": 0.6266,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.92145772469493e-06,
+      "loss": 0.6579,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.154691379518897e-06,
+      "loss": 0.6326,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.372866892980407e-06,
+      "loss": 0.6084,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.577811285260736e-06,
+      "loss": 0.6463,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.771037998057481e-06,
+      "loss": 0.6696,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.953814788633992e-06,
+      "loss": 0.5939,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0127214188602913e-05,
+      "loss": 0.6241,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0292151638856019e-05,
+      "loss": 0.6015,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.044941474495889e-05,
+      "loss": 0.6641,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0599686024068119e-05,
+      "loss": 0.6207,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0743560807141497e-05,
+      "loss": 0.6345,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0881561484225413e-05,
+      "loss": 0.6026,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.10141489541537e-05,
+      "loss": 0.618,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1141731912218572e-05,
+      "loss": 0.6262,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1264674447940031e-05,
+      "loss": 0.5963,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.138330230898981e-05,
+      "loss": 0.638,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1497908102764e-05,
+      "loss": 0.6021,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1608755644737708e-05,
+      "loss": 0.6135,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1716083616225512e-05,
+      "loss": 0.6204,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1820108659119978e-05,
+      "loss": 0.6007,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1921028008505837e-05,
+      "loss": 0.552,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2019021743562535e-05,
+      "loss": 0.5865,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2114254721302584e-05,
+      "loss": 0.6043,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2206878245307236e-05,
+      "loss": 0.6225,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2297031511879094e-05,
+      "loss": 0.5845,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2384842868314788e-05,
+      "loss": 0.5972,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2470430911848012e-05,
+      "loss": 0.5559,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2553905452876413e-05,
+      "loss": 0.6112,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2635368362101122e-05,
+      "loss": 0.5858,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2714914317973437e-05,
+      "loss": 0.5851,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.279263146820399e-05,
+      "loss": 0.5784,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2868602016925087e-05,
+      "loss": 0.594,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.294290274731322e-05,
+      "loss": 0.6134,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3015605488001847e-05,
+      "loss": 0.6602,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3086777530386599e-05,
+      "loss": 0.6202,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3156482002899656e-05,
+      "loss": 0.6427,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3224778207470516e-05,
+      "loss": 0.5556,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3291721922666924e-05,
+      "loss": 0.5921,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3357365677398804e-05,
+      "loss": 0.5781,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3421758998550176e-05,
+      "loss": 0.6216,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3484948635463673e-05,
+      "loss": 0.613,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3546978763826494e-05,
+      "loss": 0.6003,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3607891171185134e-05,
+      "loss": 0.56,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3667725426040184e-05,
+      "loss": 0.6,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3726519032234911e-05,
+      "loss": 0.6589,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3784307570146229e-05,
+      "loss": 0.5975,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3841124826009101e-05,
+      "loss": 0.6217,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3897002910551487e-05,
+      "loss": 0.6171,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3951972367982811e-05,
+      "loss": 0.5797,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4006062276262209e-05,
+      "loss": 0.5538,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4059300339470611e-05,
+      "loss": 0.6327,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4111712973021306e-05,
+      "loss": 0.5889,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.416332538236508e-05,
+      "loss": 0.5791,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4214161635776994e-05,
+      "loss": 0.5414,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.426424473175094e-05,
+      "loss": 0.5784,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4313596661474309e-05,
+      "loss": 0.5952,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4362238466807638e-05,
+      "loss": 0.5573,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4410190294151767e-05,
+      "loss": 0.6117,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4457471444547686e-05,
+      "loss": 0.5796,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4504100420320892e-05,
+      "loss": 0.583,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.455009496855234e-05,
+      "loss": 0.6215,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4595472121631605e-05,
+      "loss": 0.5543,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4640248235124197e-05,
+      "loss": 0.6009,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4684439023163616e-05,
+      "loss": 0.5823,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.472805959155989e-05,
+      "loss": 0.5576,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.477112446879911e-05,
+      "loss": 0.5773,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4813647635093114e-05,
+      "loss": 0.5699,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4855642549624762e-05,
+      "loss": 0.6023,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4897122176121516e-05,
+      "loss": 0.5659,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.493809900687906e-05,
+      "loss": 0.5538,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4978585085346223e-05,
+      "loss": 0.6092,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5018592027373444e-05,
+      "loss": 0.5642,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5058131041218538e-05,
+      "loss": 0.5953,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5097212946396e-05,
+      "loss": 0.5832,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135848191449093e-05,
+      "loss": 0.597,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5174046870717845e-05,
+      "loss": 0.6531,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.521181874017019e-05,
+      "loss": 0.5952,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5249173232358424e-05,
+      "loss": 0.6041,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5286119470558323e-05,
+      "loss": 0.5693,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.53226662821439e-05,
+      "loss": 0.5443,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535882221124695e-05,
+      "loss": 0.581,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5394595530746697e-05,
+      "loss": 0.5915,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54299942536317e-05,
+      "loss": 0.5923,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546502614377299e-05,
+      "loss": 0.5987,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.549969872614476e-05,
+      "loss": 0.6187,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.553401929652617e-05,
+      "loss": 0.5335,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.556799493071562e-05,
+      "loss": 0.5946,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5601632493286575e-05,
+      "loss": 0.5981,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5634938645912027e-05,
+      "loss": 0.5245,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.566791985528291e-05,
+      "loss": 0.5516,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5700582400643905e-05,
+      "loss": 0.5979,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5732932380968725e-05,
+      "loss": 0.5767,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5764975721795278e-05,
+      "loss": 0.56,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5796718181739928e-05,
+      "loss": 0.5427,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5828165358708776e-05,
+      "loss": 0.5634,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.585932269582265e-05,
+      "loss": 0.5697,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5890195487071597e-05,
+      "loss": 0.5939,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5920788882713426e-05,
+      "loss": 0.559,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5951107894430234e-05,
+      "loss": 0.5526,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5981157400255696e-05,
+      "loss": 0.5556,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6010942149285286e-05,
+      "loss": 0.5927,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6040466766180827e-05,
+      "loss": 0.5383,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6069735755480014e-05,
+      "loss": 0.5548,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.609875350572098e-05,
+      "loss": 0.597,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6127524293391332e-05,
+      "loss": 0.5906,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6156052286710564e-05,
+      "loss": 0.5571,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6184341549254204e-05,
+      "loss": 0.5475,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6212396043427576e-05,
+      "loss": 0.5808,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.624021963379659e-05,
+      "loss": 0.549,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6267816090282603e-05,
+      "loss": 0.5945,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.629518909122791e-05,
+      "loss": 0.5548,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6322342226338123e-05,
+      "loss": 0.567,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.634927899950731e-05,
+      "loss": 0.5936,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6376002831531498e-05,
+      "loss": 0.5732,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6402517062715713e-05,
+      "loss": 0.6085,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.642882495537963e-05,
+      "loss": 0.6088,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.645492969626641e-05,
+      "loss": 0.5686,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6480834398859314e-05,
+      "loss": 0.5777,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6506542105610182e-05,
+      "loss": 0.5969,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.653205579008382e-05,
+      "loss": 0.5051,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6557378359022096e-05,
+      "loss": 0.5806,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.658251265433128e-05,
+      "loss": 0.6127,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6607461454996043e-05,
+      "loss": 0.5597,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6632227478923372e-05,
+      "loss": 0.5773,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6656813384719412e-05,
+      "loss": 0.5303,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6681221773402148e-05,
+      "loss": 0.5295,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6705455190052738e-05,
+      "loss": 0.6011,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.672951612540804e-05,
+      "loss": 0.5772,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6753407017396868e-05,
+      "loss": 0.5698,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6777130252622384e-05,
+      "loss": 0.5608,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.680068816779279e-05,
+      "loss": 0.568,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6824083051102515e-05,
+      "loss": 0.5571,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6847317143565996e-05,
+      "loss": 0.6113,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6870392640305847e-05,
+      "loss": 0.4984,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.689331169179744e-05,
+      "loss": 0.5374,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6916076405071538e-05,
+      "loss": 0.5532,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6938688844876708e-05,
+      "loss": 0.6116,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6961151034803126e-05,
+      "loss": 0.5574,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.6983464958369298e-05,
+      "loss": 0.5967,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7005632560073116e-05,
+      "loss": 0.5537,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7027655746408716e-05,
+      "loss": 0.5687,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7049536386850415e-05,
+      "loss": 0.543,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7071276314804994e-05,
+      "loss": 0.5839,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.709287732853357e-05,
+      "loss": 0.5232,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.711434119204421e-05,
+      "loss": 0.5781,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7135669635956363e-05,
+      "loss": 0.5618,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7156864358338217e-05,
+      "loss": 0.6007,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7177927025517946e-05,
+      "loss": 0.5501,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.719885927286986e-05,
+      "loss": 0.583,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.721966270557637e-05,
+      "loss": 0.548,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.724033889936662e-05,
+      "loss": 0.5566,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7260889401232683e-05,
+      "loss": 0.5771,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7281315730124163e-05,
+      "loss": 0.5545,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7301619377621842e-05,
+      "loss": 0.5506,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7321801808591323e-05,
+      "loss": 0.612,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.73418644618172e-05,
+      "loss": 0.5585,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7361808750618543e-05,
+      "loss": 0.584,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7381636063446374e-05,
+      "loss": 0.5086,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.740134776446364e-05,
+      "loss": 0.5877,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7420945194108467e-05,
+      "loss": 0.5904,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.74404296696411e-05,
+      "loss": 0.5642,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7459802485675245e-05,
+      "loss": 0.5679,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7479064914694195e-05,
+      "loss": 0.5766,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7498218207552417e-05,
+      "loss": 0.5392,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7517263593962946e-05,
+      "loss": 0.5852,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7536202282971145e-05,
+      "loss": 0.6286,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7555035463415293e-05,
+      "loss": 0.5561,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7573764304374367e-05,
+      "loss": 0.6223,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7592389955603527e-05,
+      "loss": 0.6039,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.761091354795768e-05,
+      "loss": 0.605,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7629336193803425e-05,
+      "loss": 0.5739,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.7647658987419943e-05,
+      "loss": 0.5732,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7665883005389003e-05,
+      "loss": 0.5553,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.768400930697452e-05,
+      "loss": 0.5516,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.770203893449205e-05,
+      "loss": 0.5757,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.77199729136684e-05,
+      "loss": 0.5343,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7737812253991802e-05,
+      "loss": 0.5644,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.77555579490529e-05,
+      "loss": 0.5285,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.77732109768768e-05,
+      "loss": 0.5605,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7790772300246525e-05,
+      "loss": 0.5677,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.780824286701809e-05,
+      "loss": 0.523,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7825623610427497e-05,
+      "loss": 0.5512,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.784291544938986e-05,
+      "loss": 0.5671,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7860119288790937e-05,
+      "loss": 0.5686,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7877236019771268e-05,
+      "loss": 0.5559,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7894266520003158e-05,
+      "loss": 0.5472,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.791121165396072e-05,
+      "loss": 0.5735,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7928072273183184e-05,
+      "loss": 0.5904,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7944849216531676e-05,
+      "loss": 0.5642,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7961543310439637e-05,
+      "loss": 0.5689,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7978155369157128e-05,
+      "loss": 0.6335,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7994686194989123e-05,
+      "loss": 0.6007,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.801113657852801e-05,
+      "loss": 0.542,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8027507298880497e-05,
+      "loss": 0.6384,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8043799123889006e-05,
+      "loss": 0.5737,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.806001281034778e-05,
+      "loss": 0.5997,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8076149104213827e-05,
+      "loss": 0.5704,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8092208740812865e-05,
+      "loss": 0.5621,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.810819244504038e-05,
+      "loss": 0.5574,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8124100931557957e-05,
+      "loss": 0.5671,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.813993490498503e-05,
+      "loss": 0.5724,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.815569506008614e-05,
+      "loss": 0.5259,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8171382081953878e-05,
+      "loss": 0.5812,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8186996646187537e-05,
+      "loss": 0.5173,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8202539419067756e-05,
+      "loss": 0.5657,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8218011057727082e-05,
+      "loss": 0.6091,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8233412210316702e-05,
+      "loss": 0.5828,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8248743516169333e-05,
+      "loss": 0.5411,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8264005605958528e-05,
+      "loss": 0.587,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8279199101854302e-05,
+      "loss": 0.5638,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8294324617675335e-05,
+      "loss": 0.5708,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8309382759037793e-05,
+      "loss": 0.6416,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8324374123500798e-05,
+      "loss": 0.5877,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8339299300708727e-05,
+      "loss": 0.5954,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8354158872530387e-05,
+      "loss": 0.573,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8368953413195114e-05,
+      "loss": 0.5999,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.838368348942593e-05,
+      "loss": 0.5623,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8398349660569805e-05,
+      "loss": 0.5978,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8412952478725116e-05,
+      "loss": 0.5737,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8427492488866333e-05,
+      "loss": 0.5768,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8441970228966082e-05,
+      "loss": 0.6096,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8456386230114556e-05,
+      "loss": 0.5546,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8470741016636433e-05,
+      "loss": 0.5494,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.84850351062053e-05,
+      "loss": 0.5798,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8499269009955666e-05,
+      "loss": 0.556,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.851344323259264e-05,
+      "loss": 0.5847,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8527558272499305e-05,
+      "loss": 0.5228,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.854161462184188e-05,
+      "loss": 0.5567,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8555612766672677e-05,
+      "loss": 0.5643,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.856955318703095e-05,
+      "loss": 0.5736,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.858343635704169e-05,
+      "loss": 0.5359,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8597262745012367e-05,
+      "loss": 0.561,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8611032813527705e-05,
+      "loss": 0.5491,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.862474701954259e-05,
+      "loss": 0.5491,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8638405814473016e-05,
+      "loss": 0.5798,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.865200964428525e-05,
+      "loss": 0.5298,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.866555894958322e-05,
+      "loss": 0.5378,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8679054165694142e-05,
+      "loss": 0.5641,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.869249572275241e-05,
+      "loss": 0.5823,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8705884045781907e-05,
+      "loss": 0.5753,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.87192195547766e-05,
+      "loss": 0.5198,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8732502664779632e-05,
+      "loss": 0.5744,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8745733785960814e-05,
+      "loss": 0.5648,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8758913323692628e-05,
+      "loss": 0.5913,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.877204167862473e-05,
+      "loss": 0.5317,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.878511924675706e-05,
+      "loss": 0.5298,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.879814641951151e-05,
+      "loss": 0.6114,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8811123583802192e-05,
+      "loss": 0.5602,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8824051122104415e-05,
+      "loss": 0.5679,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8836929412522337e-05,
+      "loss": 0.5667,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8849758828855284e-05,
+      "loss": 0.5435,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8862539740662883e-05,
+      "loss": 0.6069,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.887527251332892e-05,
+      "loss": 0.5441,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8887957508124035e-05,
+      "loss": 0.5603,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.89005950822672e-05,
+      "loss": 0.5616,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8913185588986104e-05,
+      "loss": 0.5752,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.892572937757638e-05,
+      "loss": 0.5379,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8938226793459762e-05,
+      "loss": 0.5463,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8950678178241144e-05,
+      "loss": 0.5557,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.896308386976462e-05,
+      "loss": 0.5675,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8975444202168474e-05,
+      "loss": 0.5839,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8987759505939203e-05,
+      "loss": 0.5336,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.900003010796451e-05,
+      "loss": 0.6026,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9012256331585393e-05,
+      "loss": 0.5879,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.902443849664725e-05,
+      "loss": 0.5672,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.903657691955009e-05,
+      "loss": 0.5699,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.904867191329784e-05,
+      "loss": 0.5966,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.906072378754679e-05,
+      "loss": 0.584,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.907273284865314e-05,
+      "loss": 0.5332,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9084699399719735e-05,
+      "loss": 0.5581,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.909662374064197e-05,
+      "loss": 0.5622,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.910850616815289e-05,
+      "loss": 0.5471,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.912034697586749e-05,
+      "loss": 0.5948,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9132146454326245e-05,
+      "loss": 0.5251,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9143904891037894e-05,
+      "loss": 0.5905,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.915562257052147e-05,
+      "loss": 0.5845,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.916729977434762e-05,
+      "loss": 0.5278,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.917893678117918e-05,
+      "loss": 0.5801,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9190533866811098e-05,
+      "loss": 0.524,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.920209130420964e-05,
+      "loss": 0.5737,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9213609363550945e-05,
+      "loss": 0.5278,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9225088312258937e-05,
+      "loss": 0.5623,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.923652841504254e-05,
+      "loss": 0.5476,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.924792993393236e-05,
+      "loss": 0.5624,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.925929312831664e-05,
+      "loss": 0.543,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9270618254976718e-05,
+      "loss": 0.5549,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.928190556812181e-05,
+      "loss": 0.5813,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9293155319423264e-05,
+      "loss": 0.5391,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.930436775804823e-05,
+      "loss": 0.5486,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.931554313069276e-05,
+      "loss": 0.5605,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.93266816816144e-05,
+      "loss": 0.5868,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9337783652664195e-05,
+      "loss": 0.5738,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9348849283318217e-05,
+      "loss": 0.5669,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.935987881070856e-05,
+      "loss": 0.5405,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9370872469653817e-05,
+      "loss": 0.5491,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.93818304926891e-05,
+      "loss": 0.5397,
+      "step": 309
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.939275311009552e-05,
+      "loss": 0.5982,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.940364054992925e-05,
+      "loss": 0.593,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9414493038050095e-05,
+      "loss": 0.5639,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9425310798149587e-05,
+      "loss": 0.5585,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9436094051778672e-05,
+      "loss": 0.57,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9446843018374915e-05,
+      "loss": 0.5815,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9457557915289312e-05,
+      "loss": 0.5673,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.946823895781265e-05,
+      "loss": 0.5773,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9478886359201468e-05,
+      "loss": 0.5503,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9489500330703598e-05,
+      "loss": 0.5348,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.950008108158332e-05,
+      "loss": 0.5778,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9510628819146118e-05,
+      "loss": 0.5604,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.952114374876305e-05,
+      "loss": 0.5645,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9531626073894726e-05,
+      "loss": 0.5604,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9542075996114963e-05,
+      "loss": 0.5431,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9552493715134014e-05,
+      "loss": 0.5282,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9562879428821476e-05,
+      "loss": 0.5349,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9573233333228844e-05,
+      "loss": 0.5323,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.958355562261172e-05,
+      "loss": 0.5594,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9593846489451674e-05,
+      "loss": 0.5522,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9604106124477787e-05,
+      "loss": 0.5724,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9614334716687847e-05,
+      "loss": 0.5927,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.962453245336926e-05,
+      "loss": 0.5608,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9634699520119616e-05,
+      "loss": 0.5584,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9644836100866947e-05,
+      "loss": 0.5306,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9654942377889704e-05,
+      "loss": 0.4725,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9665018531836427e-05,
+      "loss": 0.5687,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967506474174511e-05,
+      "loss": 0.5341,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.96850811850623e-05,
+      "loss": 0.5755,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9695068037661883e-05,
+      "loss": 0.5168,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9705025473863648e-05,
+      "loss": 0.5639,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9714953666451494e-05,
+      "loss": 0.5652,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9724852786691472e-05,
+      "loss": 0.5931,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9734723004349483e-05,
+      "loss": 0.5978,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9744564487708742e-05,
+      "loss": 0.5107,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9754377403587017e-05,
+      "loss": 0.5534,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.976416191735357e-05,
+      "loss": 0.536,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.97739181929459e-05,
+      "loss": 0.5595,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9783646392886204e-05,
+      "loss": 0.5353,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9793346678297643e-05,
+      "loss": 0.5391,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9803019208920343e-05,
+      "loss": 0.548,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812664143127168e-05,
+      "loss": 0.5516,
+      "step": 351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822281637939296e-05,
+      "loss": 0.549,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983187184904155e-05,
+      "loss": 0.5275,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841434930797522e-05,
+      "loss": 0.5358,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9850971036264473e-05,
+      "loss": 0.5744,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9860480317208048e-05,
+      "loss": 0.5702,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.986996292411675e-05,
+      "loss": 0.5781,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.987941900621625e-05,
+      "loss": 0.5549,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9888848711483447e-05,
+      "loss": 0.5415,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9898252186660394e-05,
+      "loss": 0.5808,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9907629577267985e-05,
+      "loss": 0.5753,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9916981027619465e-05,
+      "loss": 0.553,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9926306680833762e-05,
+      "loss": 0.547,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9935606678848632e-05,
+      "loss": 0.5085,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99448811624336e-05,
+      "loss": 0.5361,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.995413027120278e-05,
+      "loss": 0.5061,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9963354143627454e-05,
+      "loss": 0.5604,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9972552917048526e-05,
+      "loss": 0.5587,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9981726727688793e-05,
+      "loss": 0.536,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9990875710665044e-05,
+      "loss": 0.5076,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5544,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.5741,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999833180415381e-05,
+      "loss": 0.5938,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9996663608307617e-05,
+      "loss": 0.5492,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9994995412461424e-05,
+      "loss": 0.5425,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9993327216615233e-05,
+      "loss": 0.558,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999165902076904e-05,
+      "loss": 0.5221,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9989990824922846e-05,
+      "loss": 0.508,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9988322629076655e-05,
+      "loss": 0.5506,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9986654433230465e-05,
+      "loss": 0.5707,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.998498623738427e-05,
+      "loss": 0.5443,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9983318041538077e-05,
+      "loss": 0.5108,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9981649845691887e-05,
+      "loss": 0.569,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9979981649845693e-05,
+      "loss": 0.5311,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99783134539995e-05,
+      "loss": 0.5515,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997664525815331e-05,
+      "loss": 0.5851,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9974977062307118e-05,
+      "loss": 0.5376,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9973308866460924e-05,
+      "loss": 0.5519,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.997164067061473e-05,
+      "loss": 0.5244,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.996997247476854e-05,
+      "loss": 0.5556,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9968304278922346e-05,
+      "loss": 0.5056,
+      "step": 391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9966636083076153e-05,
+      "loss": 0.562,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9964967887229962e-05,
+      "loss": 0.5626,
+      "step": 393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9963299691383772e-05,
+      "loss": 0.5499,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9961631495537578e-05,
+      "loss": 0.5655,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9959963299691384e-05,
+      "loss": 0.5664,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9958295103845194e-05,
+      "loss": 0.5445,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9956626907999e-05,
+      "loss": 0.5358,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9954958712152806e-05,
+      "loss": 0.5686,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9953290516306616e-05,
+      "loss": 0.5362,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9951622320460425e-05,
+      "loss": 0.5563,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994995412461423e-05,
+      "loss": 0.5617,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9948285928768038e-05,
+      "loss": 0.5215,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9946617732921847e-05,
+      "loss": 0.5732,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9944949537075657e-05,
+      "loss": 0.5554,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994328134122946e-05,
+      "loss": 0.5394,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994161314538327e-05,
+      "loss": 0.5454,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.993994494953708e-05,
+      "loss": 0.5673,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9938276753690885e-05,
+      "loss": 0.5218,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.993660855784469e-05,
+      "loss": 0.5539,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99349403619985e-05,
+      "loss": 0.5276,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.993327216615231e-05,
+      "loss": 0.5346,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9931603970306113e-05,
+      "loss": 0.5492,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9929935774459923e-05,
+      "loss": 0.5454,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9928267578613732e-05,
+      "loss": 0.5638,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.992659938276754e-05,
+      "loss": 0.5055,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9924931186921345e-05,
+      "loss": 0.5079,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9923262991075154e-05,
+      "loss": 0.5455,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9921594795228964e-05,
+      "loss": 0.5364,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.991992659938277e-05,
+      "loss": 0.5673,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9918258403536576e-05,
+      "loss": 0.5604,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9916590207690386e-05,
+      "loss": 0.5475,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9914922011844192e-05,
+      "loss": 0.591,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9913253815997998e-05,
+      "loss": 0.5757,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9911585620151808e-05,
+      "loss": 0.4984,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9909917424305617e-05,
+      "loss": 0.5599,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9908249228459424e-05,
+      "loss": 0.5788,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.990658103261323e-05,
+      "loss": 0.5222,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.990491283676704e-05,
+      "loss": 0.5568,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9903244640920846e-05,
+      "loss": 0.55,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9901576445074652e-05,
+      "loss": 0.5273,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.989990824922846e-05,
+      "loss": 0.5166,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.989824005338227e-05,
+      "loss": 0.5769,
+      "step": 433
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9896571857536077e-05,
+      "loss": 0.5589,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9894903661689883e-05,
+      "loss": 0.5609,
+      "step": 435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9893235465843693e-05,
+      "loss": 0.5234,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.98915672699975e-05,
+      "loss": 0.5413,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9889899074151305e-05,
+      "loss": 0.5316,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9888230878305115e-05,
+      "loss": 0.54,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9886562682458924e-05,
+      "loss": 0.5566,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.988489448661273e-05,
+      "loss": 0.5263,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9883226290766537e-05,
+      "loss": 0.539,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9881558094920346e-05,
+      "loss": 0.5703,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9879889899074153e-05,
+      "loss": 0.5653,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987822170322796e-05,
+      "loss": 0.546,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987655350738177e-05,
+      "loss": 0.5334,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9874885311535578e-05,
+      "loss": 0.5464,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9873217115689384e-05,
+      "loss": 0.5141,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.987154891984319e-05,
+      "loss": 0.5488,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9869880723997e-05,
+      "loss": 0.5482,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9868212528150806e-05,
+      "loss": 0.5688,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9866544332304612e-05,
+      "loss": 0.5577,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9864876136458422e-05,
+      "loss": 0.5083,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.986320794061223e-05,
+      "loss": 0.5643,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9861539744766038e-05,
+      "loss": 0.5504,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9859871548919844e-05,
+      "loss": 0.5416,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9858203353073653e-05,
+      "loss": 0.5522,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.985653515722746e-05,
+      "loss": 0.5126,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9854866961381266e-05,
+      "loss": 0.5403,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9853198765535075e-05,
+      "loss": 0.5796,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9851530569688885e-05,
+      "loss": 0.5199,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984986237384269e-05,
+      "loss": 0.5664,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9848194177996497e-05,
+      "loss": 0.5927,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9846525982150307e-05,
+      "loss": 0.541,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9844857786304113e-05,
+      "loss": 0.5868,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984318959045792e-05,
+      "loss": 0.578,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.984152139461173e-05,
+      "loss": 0.5423,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983985319876554e-05,
+      "loss": 0.5413,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9838185002919345e-05,
+      "loss": 0.5543,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983651680707315e-05,
+      "loss": 0.5641,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983484861122696e-05,
+      "loss": 0.5822,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.983318041538077e-05,
+      "loss": 0.5545,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9831512219534573e-05,
+      "loss": 0.5806,
+      "step": 473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9829844023688382e-05,
+      "loss": 0.5324,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9828175827842192e-05,
+      "loss": 0.5407,
+      "step": 475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9826507631995998e-05,
+      "loss": 0.5484,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9824839436149804e-05,
+      "loss": 0.5385,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9823171240303614e-05,
+      "loss": 0.5186,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9821503044457424e-05,
+      "loss": 0.575,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9819834848611226e-05,
+      "loss": 0.5669,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9818166652765036e-05,
+      "loss": 0.5412,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9816498456918845e-05,
+      "loss": 0.5137,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.981483026107265e-05,
+      "loss": 0.5419,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9813162065226458e-05,
+      "loss": 0.5481,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9811493869380267e-05,
+      "loss": 0.5299,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9809825673534077e-05,
+      "loss": 0.5151,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9808157477687883e-05,
+      "loss": 0.5204,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.980648928184169e-05,
+      "loss": 0.5562,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.98048210859955e-05,
+      "loss": 0.5542,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9803152890149305e-05,
+      "loss": 0.6002,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.980148469430311e-05,
+      "loss": 0.5558,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.979981649845692e-05,
+      "loss": 0.5153,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.979814830261073e-05,
+      "loss": 0.5465,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9796480106764537e-05,
+      "loss": 0.5487,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9794811910918343e-05,
+      "loss": 0.5781,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9793143715072153e-05,
+      "loss": 0.5377,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.979147551922596e-05,
+      "loss": 0.5542,
+      "step": 497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9789807323379765e-05,
+      "loss": 0.5907,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9788139127533574e-05,
+      "loss": 0.5288,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9786470931687384e-05,
+      "loss": 0.5191,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.978480273584119e-05,
+      "loss": 0.5699,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9783134539994996e-05,
+      "loss": 0.5262,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9781466344148806e-05,
+      "loss": 0.5491,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9779798148302612e-05,
+      "loss": 0.5554,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.977812995245642e-05,
+      "loss": 0.6116,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9776461756610228e-05,
+      "loss": 0.5522,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9774793560764038e-05,
+      "loss": 0.5537,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9773125364917844e-05,
+      "loss": 0.5142,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.977145716907165e-05,
+      "loss": 0.551,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976978897322546e-05,
+      "loss": 0.54,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9768120777379266e-05,
+      "loss": 0.5633,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9766452581533072e-05,
+      "loss": 0.4933,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976478438568688e-05,
+      "loss": 0.5126,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.976311618984069e-05,
+      "loss": 0.5203,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9761447993994497e-05,
+      "loss": 0.5396,
+      "step": 515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9759779798148303e-05,
+      "loss": 0.5694,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9758111602302113e-05,
+      "loss": 0.5146,
+      "step": 517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.975644340645592e-05,
+      "loss": 0.5557,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9754775210609725e-05,
+      "loss": 0.5735,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9753107014763535e-05,
+      "loss": 0.5784,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9751438818917345e-05,
+      "loss": 0.5622,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.974977062307115e-05,
+      "loss": 0.5302,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9748102427224957e-05,
+      "loss": 0.5108,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9746434231378767e-05,
+      "loss": 0.5156,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9744766035532573e-05,
+      "loss": 0.5495,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.974309783968638e-05,
+      "loss": 0.5188,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.974142964384019e-05,
+      "loss": 0.5314,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9739761447993998e-05,
+      "loss": 0.5226,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9738093252147804e-05,
+      "loss": 0.4997,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973642505630161e-05,
+      "loss": 0.6006,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.973475686045542e-05,
+      "loss": 0.5546,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9733088664609226e-05,
+      "loss": 0.5487,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9731420468763032e-05,
+      "loss": 0.5609,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9729752272916842e-05,
+      "loss": 0.5617,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.972808407707065e-05,
+      "loss": 0.5377,
+      "step": 535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9726415881224458e-05,
+      "loss": 0.5366,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9724747685378264e-05,
+      "loss": 0.5231,
+      "step": 537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9723079489532074e-05,
+      "loss": 0.5713,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.972141129368588e-05,
+      "loss": 0.5066,
+      "step": 539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9719743097839686e-05,
+      "loss": 0.5087,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9718074901993496e-05,
+      "loss": 0.5247,
+      "step": 541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9716406706147305e-05,
+      "loss": 0.5617,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.971473851030111e-05,
+      "loss": 0.562,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9713070314454918e-05,
+      "loss": 0.5341,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9711402118608727e-05,
+      "loss": 0.5128,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9709733922762533e-05,
+      "loss": 0.5619,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970806572691634e-05,
+      "loss": 0.5255,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970639753107015e-05,
+      "loss": 0.5362,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970472933522396e-05,
+      "loss": 0.5195,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9703061139377765e-05,
+      "loss": 0.5526,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.970139294353157e-05,
+      "loss": 0.5722,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969972474768538e-05,
+      "loss": 0.5427,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.969805655183919e-05,
+      "loss": 0.5104,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9696388355992993e-05,
+      "loss": 0.5616,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9694720160146803e-05,
+      "loss": 0.4999,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9693051964300612e-05,
+      "loss": 0.5409,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.969138376845442e-05,
+      "loss": 0.5068,
+      "step": 557
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9689715572608225e-05,
+      "loss": 0.5324,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9688047376762034e-05,
+      "loss": 0.5387,
+      "step": 559
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9686379180915844e-05,
+      "loss": 0.5964,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.968471098506965e-05,
+      "loss": 0.5343,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9683042789223456e-05,
+      "loss": 0.5444,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9681374593377266e-05,
+      "loss": 0.5111,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9679706397531072e-05,
+      "loss": 0.5417,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9678038201684878e-05,
+      "loss": 0.5036,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9676370005838688e-05,
+      "loss": 0.5357,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9674701809992497e-05,
+      "loss": 0.5844,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9673033614146303e-05,
+      "loss": 0.5424,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.967136541830011e-05,
+      "loss": 0.536,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.966969722245392e-05,
+      "loss": 0.5613,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9668029026607725e-05,
+      "loss": 0.5464,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.966636083076153e-05,
+      "loss": 0.522,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.966469263491534e-05,
+      "loss": 0.5151,
+      "step": 573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.966302443906915e-05,
+      "loss": 0.5578,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9661356243222957e-05,
+      "loss": 0.4971,
+      "step": 575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9659688047376763e-05,
+      "loss": 0.6183,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9658019851530573e-05,
+      "loss": 0.5518,
+      "step": 577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.965635165568438e-05,
+      "loss": 0.5255,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9654683459838185e-05,
+      "loss": 0.5693,
+      "step": 579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9653015263991995e-05,
+      "loss": 0.5469,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9651347068145804e-05,
+      "loss": 0.5586,
+      "step": 581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.964967887229961e-05,
+      "loss": 0.5478,
+      "step": 582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9648010676453417e-05,
+      "loss": 0.52,
+      "step": 583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9646342480607226e-05,
+      "loss": 0.5332,
+      "step": 584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9644674284761032e-05,
+      "loss": 0.5114,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.964300608891484e-05,
+      "loss": 0.5419,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9641337893068648e-05,
+      "loss": 0.5266,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9639669697222458e-05,
+      "loss": 0.5485,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9638001501376264e-05,
+      "loss": 0.5714,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.963633330553007e-05,
+      "loss": 0.5654,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.963466510968388e-05,
+      "loss": 0.4931,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9632996913837686e-05,
+      "loss": 0.5089,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9631328717991492e-05,
+      "loss": 0.5392,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9629660522145302e-05,
+      "loss": 0.5517,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.962799232629911e-05,
+      "loss": 0.5332,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9626324130452918e-05,
+      "loss": 0.5359,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9624655934606724e-05,
+      "loss": 0.5292,
+      "step": 597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9622987738760533e-05,
+      "loss": 0.4969,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.962131954291434e-05,
+      "loss": 0.527,
+      "step": 599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9619651347068146e-05,
+      "loss": 0.5566,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9617983151221955e-05,
+      "loss": 0.5824,
+      "step": 601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9616314955375765e-05,
+      "loss": 0.531,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.961464675952957e-05,
+      "loss": 0.5019,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9612978563683377e-05,
+      "loss": 0.5982,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9611310367837187e-05,
+      "loss": 0.523,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9609642171990993e-05,
+      "loss": 0.5167,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.96079739761448e-05,
+      "loss": 0.5395,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960630578029861e-05,
+      "loss": 0.5364,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960463758445242e-05,
+      "loss": 0.56,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9602969388606225e-05,
+      "loss": 0.5159,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.960130119276003e-05,
+      "loss": 0.5802,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.959963299691384e-05,
+      "loss": 0.5725,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9597964801067647e-05,
+      "loss": 0.6056,
+      "step": 613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9596296605221453e-05,
+      "loss": 0.528,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9594628409375262e-05,
+      "loss": 0.545,
+      "step": 615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9592960213529072e-05,
+      "loss": 0.529,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9591292017682878e-05,
+      "loss": 0.5391,
+      "step": 617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9589623821836684e-05,
+      "loss": 0.5233,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9587955625990494e-05,
+      "loss": 0.5618,
+      "step": 619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.95862874301443e-05,
+      "loss": 0.5525,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9584619234298106e-05,
+      "loss": 0.5214,
+      "step": 621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9582951038451916e-05,
+      "loss": 0.5451,
+      "step": 622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9581282842605725e-05,
+      "loss": 0.5367,
+      "step": 623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957961464675953e-05,
+      "loss": 0.5238,
+      "step": 624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9577946450913338e-05,
+      "loss": 0.5177,
+      "step": 625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9576278255067147e-05,
+      "loss": 0.5254,
+      "step": 626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9574610059220954e-05,
+      "loss": 0.5444,
+      "step": 627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957294186337476e-05,
+      "loss": 0.5517,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.957127366752857e-05,
+      "loss": 0.5511,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956960547168238e-05,
+      "loss": 0.4936,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9567937275836185e-05,
+      "loss": 0.5549,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.956626907998999e-05,
+      "loss": 0.5083,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.95646008841438e-05,
+      "loss": 0.5579,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9562932688297607e-05,
+      "loss": 0.5698,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9561264492451417e-05,
+      "loss": 0.5291,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9559596296605223e-05,
+      "loss": 0.5065,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9557928100759032e-05,
+      "loss": 0.561,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.955625990491284e-05,
+      "loss": 0.5451,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9554591709066645e-05,
+      "loss": 0.5362,
+      "step": 639
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9552923513220454e-05,
+      "loss": 0.5484,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.955125531737426e-05,
+      "loss": 0.5182,
+      "step": 641
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.954958712152807e-05,
+      "loss": 0.5611,
+      "step": 642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9547918925681876e-05,
+      "loss": 0.5419,
+      "step": 643
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9546250729835686e-05,
+      "loss": 0.559,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9544582533989492e-05,
+      "loss": 0.5278,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.95429143381433e-05,
+      "loss": 0.5121,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9541246142297108e-05,
+      "loss": 0.5268,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9539577946450914e-05,
+      "loss": 0.5543,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9537909750604724e-05,
+      "loss": 0.5341,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.953624155475853e-05,
+      "loss": 0.5501,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.953457335891234e-05,
+      "loss": 0.521,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9532905163066146e-05,
+      "loss": 0.5173,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9531236967219952e-05,
+      "loss": 0.5015,
+      "step": 653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.952956877137376e-05,
+      "loss": 0.5512,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9527900575527568e-05,
+      "loss": 0.5624,
+      "step": 655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9526232379681377e-05,
+      "loss": 0.5397,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9524564183835183e-05,
+      "loss": 0.5595,
+      "step": 657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9522895987988993e-05,
+      "loss": 0.5233,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.95212277921428e-05,
+      "loss": 0.4981,
+      "step": 659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9519559596296605e-05,
+      "loss": 0.5152,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9517891400450415e-05,
+      "loss": 0.532,
+      "step": 661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.951622320460422e-05,
+      "loss": 0.5403,
+      "step": 662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.951455500875803e-05,
+      "loss": 0.5161,
+      "step": 663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9512886812911837e-05,
+      "loss": 0.5137,
+      "step": 664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9511218617065646e-05,
+      "loss": 0.554,
+      "step": 665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9509550421219453e-05,
+      "loss": 0.5545,
+      "step": 666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.950788222537326e-05,
+      "loss": 0.4913,
+      "step": 667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.950621402952707e-05,
+      "loss": 0.5219,
+      "step": 668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9504545833680878e-05,
+      "loss": 0.4969,
+      "step": 669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9502877637834684e-05,
+      "loss": 0.5463,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.950120944198849e-05,
+      "loss": 0.5228,
+      "step": 671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.94995412461423e-05,
+      "loss": 0.4767,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9497873050296106e-05,
+      "loss": 0.5549,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9496204854449912e-05,
+      "loss": 0.5322,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9494536658603722e-05,
+      "loss": 0.5319,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.949286846275753e-05,
+      "loss": 0.5823,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9491200266911338e-05,
+      "loss": 0.5296,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9489532071065144e-05,
+      "loss": 0.5715,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9487863875218954e-05,
+      "loss": 0.5075,
+      "step": 679
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.948619567937276e-05,
+      "loss": 0.567,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9484527483526566e-05,
+      "loss": 0.5385,
+      "step": 681
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9482859287680375e-05,
+      "loss": 0.5,
+      "step": 682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9481191091834185e-05,
+      "loss": 0.5444,
+      "step": 683
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947952289598799e-05,
+      "loss": 0.5277,
+      "step": 684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9477854700141797e-05,
+      "loss": 0.5336,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9476186504295607e-05,
+      "loss": 0.526,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9474518308449413e-05,
+      "loss": 0.5074,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947285011260322e-05,
+      "loss": 0.57,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.947118191675703e-05,
+      "loss": 0.5384,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946951372091084e-05,
+      "loss": 0.5319,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9467845525064645e-05,
+      "loss": 0.5062,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946617732921845e-05,
+      "loss": 0.5116,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.946450913337226e-05,
+      "loss": 0.5169,
+      "step": 693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9462840937526067e-05,
+      "loss": 0.5124,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9461172741679873e-05,
+      "loss": 0.5213,
+      "step": 695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9459504545833683e-05,
+      "loss": 0.5535,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9457836349987492e-05,
+      "loss": 0.5229,
+      "step": 697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9456168154141298e-05,
+      "loss": 0.5802,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9454499958295104e-05,
+      "loss": 0.5252,
+      "step": 699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9452831762448914e-05,
+      "loss": 0.5166,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.945116356660272e-05,
+      "loss": 0.5341,
+      "step": 701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.944949537075653e-05,
+      "loss": 0.5207,
+      "step": 702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9447827174910336e-05,
+      "loss": 0.5096,
+      "step": 703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9446158979064146e-05,
+      "loss": 0.5426,
+      "step": 704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9444490783217952e-05,
+      "loss": 0.5435,
+      "step": 705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9442822587371758e-05,
+      "loss": 0.5467,
+      "step": 706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9441154391525568e-05,
+      "loss": 0.5182,
+      "step": 707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9439486195679374e-05,
+      "loss": 0.5739,
+      "step": 708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9437817999833183e-05,
+      "loss": 0.5289,
+      "step": 709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.943614980398699e-05,
+      "loss": 0.5546,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.94344816081408e-05,
+      "loss": 0.5285,
+      "step": 711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9432813412294605e-05,
+      "loss": 0.5465,
+      "step": 712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.943114521644841e-05,
+      "loss": 0.5531,
+      "step": 713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.942947702060222e-05,
+      "loss": 0.5369,
+      "step": 714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9427808824756027e-05,
+      "loss": 0.5668,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9426140628909837e-05,
+      "loss": 0.5458,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9424472433063643e-05,
+      "loss": 0.5317,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9422804237217453e-05,
+      "loss": 0.5582,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.942113604137126e-05,
+      "loss": 0.5242,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9419467845525065e-05,
+      "loss": 0.5481,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9417799649678875e-05,
+      "loss": 0.5924,
+      "step": 721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.941613145383268e-05,
+      "loss": 0.5095,
+      "step": 722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.941446325798649e-05,
+      "loss": 0.5177,
+      "step": 723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9412795062140297e-05,
+      "loss": 0.4752,
+      "step": 724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9411126866294106e-05,
+      "loss": 0.5383,
+      "step": 725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9409458670447912e-05,
+      "loss": 0.5067,
+      "step": 726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.940779047460172e-05,
+      "loss": 0.5559,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9406122278755528e-05,
+      "loss": 0.5275,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9404454082909334e-05,
+      "loss": 0.5295,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9402785887063144e-05,
+      "loss": 0.4904,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.940111769121695e-05,
+      "loss": 0.5397,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.939944949537076e-05,
+      "loss": 0.5615,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9397781299524566e-05,
+      "loss": 0.5609,
+      "step": 733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9396113103678372e-05,
+      "loss": 0.5332,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.939444490783218e-05,
+      "loss": 0.5313,
+      "step": 735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9392776711985988e-05,
+      "loss": 0.5772,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9391108516139797e-05,
+      "loss": 0.5176,
+      "step": 737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9389440320293604e-05,
+      "loss": 0.5512,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9387772124447413e-05,
+      "loss": 0.5568,
+      "step": 739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.938610392860122e-05,
+      "loss": 0.5478,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9384435732755026e-05,
+      "loss": 0.5088,
+      "step": 741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9382767536908835e-05,
+      "loss": 0.5529,
+      "step": 742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.938109934106264e-05,
+      "loss": 0.4868,
+      "step": 743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.937943114521645e-05,
+      "loss": 0.506,
+      "step": 744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9377762949370257e-05,
+      "loss": 0.5403,
+      "step": 745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9376094753524067e-05,
+      "loss": 0.5394,
+      "step": 746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9374426557677873e-05,
+      "loss": 0.5511,
+      "step": 747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.937275836183168e-05,
+      "loss": 0.5394,
+      "step": 748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.937109016598549e-05,
+      "loss": 0.5574,
+      "step": 749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9369421970139295e-05,
+      "loss": 0.5235,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9367753774293104e-05,
+      "loss": 0.5407,
+      "step": 751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.936608557844691e-05,
+      "loss": 0.5148,
+      "step": 752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.936441738260072e-05,
+      "loss": 0.5092,
+      "step": 753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9362749186754526e-05,
+      "loss": 0.5459,
+      "step": 754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9361080990908333e-05,
+      "loss": 0.5209,
+      "step": 755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9359412795062142e-05,
+      "loss": 0.5153,
+      "step": 756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.935774459921595e-05,
+      "loss": 0.5311,
+      "step": 757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9356076403369758e-05,
+      "loss": 0.5185,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9354408207523564e-05,
+      "loss": 0.5221,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9352740011677374e-05,
+      "loss": 0.5312,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.935107181583118e-05,
+      "loss": 0.5196,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9349403619984986e-05,
+      "loss": 0.5489,
+      "step": 762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9347735424138796e-05,
+      "loss": 0.5288,
+      "step": 763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9346067228292602e-05,
+      "loss": 0.5358,
+      "step": 764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.934439903244641e-05,
+      "loss": 0.5203,
+      "step": 765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9342730836600218e-05,
+      "loss": 0.5796,
+      "step": 766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9341062640754027e-05,
+      "loss": 0.5547,
+      "step": 767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9339394444907833e-05,
+      "loss": 0.5266,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.933772624906164e-05,
+      "loss": 0.5229,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.933605805321545e-05,
+      "loss": 0.5081,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9334389857369255e-05,
+      "loss": 0.5248,
+      "step": 771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9332721661523065e-05,
+      "loss": 0.5313,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.933105346567687e-05,
+      "loss": 0.5467,
+      "step": 773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.932938526983068e-05,
+      "loss": 0.5053,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9327717073984487e-05,
+      "loss": 0.5245,
+      "step": 775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9326048878138297e-05,
+      "loss": 0.5251,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9324380682292103e-05,
+      "loss": 0.5553,
+      "step": 777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.932271248644591e-05,
+      "loss": 0.5245,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.932104429059972e-05,
+      "loss": 0.5003,
+      "step": 779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9319376094753525e-05,
+      "loss": 0.5558,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9317707898907334e-05,
+      "loss": 0.5633,
+      "step": 781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.931603970306114e-05,
+      "loss": 0.53,
+      "step": 782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.931437150721495e-05,
+      "loss": 0.5061,
+      "step": 783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9312703311368756e-05,
+      "loss": 0.5325,
+      "step": 784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9311035115522562e-05,
+      "loss": 0.5147,
+      "step": 785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9309366919676372e-05,
+      "loss": 0.5322,
+      "step": 786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9307698723830178e-05,
+      "loss": 0.5707,
+      "step": 787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9306030527983988e-05,
+      "loss": 0.5439,
+      "step": 788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9304362332137794e-05,
+      "loss": 0.5463,
+      "step": 789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9302694136291604e-05,
+      "loss": 0.551,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.930102594044541e-05,
+      "loss": 0.5423,
+      "step": 791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.929935774459922e-05,
+      "loss": 0.5277,
+      "step": 792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9297689548753026e-05,
+      "loss": 0.5544,
+      "step": 793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9296021352906832e-05,
+      "loss": 0.5564,
+      "step": 794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.929435315706064e-05,
+      "loss": 0.5533,
+      "step": 795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9292684961214447e-05,
+      "loss": 0.5564,
+      "step": 796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9291016765368257e-05,
+      "loss": 0.5193,
+      "step": 797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9289348569522063e-05,
+      "loss": 0.5255,
+      "step": 798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9287680373675873e-05,
+      "loss": 0.5288,
+      "step": 799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.928601217782968e-05,
+      "loss": 0.532,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9284343981983485e-05,
+      "loss": 0.5319,
+      "step": 801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9282675786137295e-05,
+      "loss": 0.5071,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.92810075902911e-05,
+      "loss": 0.5081,
+      "step": 803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.927933939444491e-05,
+      "loss": 0.5748,
+      "step": 804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9277671198598717e-05,
+      "loss": 0.5167,
+      "step": 805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9276003002752526e-05,
+      "loss": 0.5059,
+      "step": 806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9274334806906333e-05,
+      "loss": 0.5127,
+      "step": 807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.927266661106014e-05,
+      "loss": 0.5143,
+      "step": 808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.927099841521395e-05,
+      "loss": 0.546,
+      "step": 809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9269330219367755e-05,
+      "loss": 0.5422,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9267662023521564e-05,
+      "loss": 0.5434,
+      "step": 811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.926599382767537e-05,
+      "loss": 0.5727,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.926432563182918e-05,
+      "loss": 0.5233,
+      "step": 813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9262657435982986e-05,
+      "loss": 0.5191,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9260989240136792e-05,
+      "loss": 0.5146,
+      "step": 815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9259321044290602e-05,
+      "loss": 0.5375,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9257652848444408e-05,
+      "loss": 0.5834,
+      "step": 817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9255984652598218e-05,
+      "loss": 0.5113,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9254316456752024e-05,
+      "loss": 0.5496,
+      "step": 819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9252648260905833e-05,
+      "loss": 0.5318,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.925098006505964e-05,
+      "loss": 0.4976,
+      "step": 821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9249311869213446e-05,
+      "loss": 0.5435,
+      "step": 822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9247643673367255e-05,
+      "loss": 0.5121,
+      "step": 823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.924597547752106e-05,
+      "loss": 0.5154,
+      "step": 824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.924430728167487e-05,
+      "loss": 0.521,
+      "step": 825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9242639085828677e-05,
+      "loss": 0.5398,
+      "step": 826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9240970889982487e-05,
+      "loss": 0.5468,
+      "step": 827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9239302694136293e-05,
+      "loss": 0.5091,
+      "step": 828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.92376344982901e-05,
+      "loss": 0.5106,
+      "step": 829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.923596630244391e-05,
+      "loss": 0.5118,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9234298106597715e-05,
+      "loss": 0.515,
+      "step": 831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9232629910751525e-05,
+      "loss": 0.5021,
+      "step": 832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.923096171490533e-05,
+      "loss": 0.5385,
+      "step": 833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.922929351905914e-05,
+      "loss": 0.4837,
+      "step": 834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9227625323212947e-05,
+      "loss": 0.5559,
+      "step": 835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9225957127366753e-05,
+      "loss": 0.5406,
+      "step": 836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9224288931520562e-05,
+      "loss": 0.5172,
+      "step": 837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.922262073567437e-05,
+      "loss": 0.5168,
+      "step": 838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9220952539828178e-05,
+      "loss": 0.5313,
+      "step": 839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9219284343981984e-05,
+      "loss": 0.5489,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9217616148135794e-05,
+      "loss": 0.574,
+      "step": 841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.92159479522896e-05,
+      "loss": 0.529,
+      "step": 842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.921427975644341e-05,
+      "loss": 0.5316,
+      "step": 843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9212611560597216e-05,
+      "loss": 0.5297,
+      "step": 844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9210943364751022e-05,
+      "loss": 0.5475,
+      "step": 845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.920927516890483e-05,
+      "loss": 0.5164,
+      "step": 846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9207606973058638e-05,
+      "loss": 0.5326,
+      "step": 847
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9205938777212447e-05,
+      "loss": 0.5218,
+      "step": 848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9204270581366254e-05,
+      "loss": 0.5412,
+      "step": 849
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9202602385520063e-05,
+      "loss": 0.5395,
+      "step": 850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.920093418967387e-05,
+      "loss": 0.508,
+      "step": 851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9199265993827676e-05,
+      "loss": 0.4539,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9197597797981485e-05,
+      "loss": 0.4993,
+      "step": 853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.919592960213529e-05,
+      "loss": 0.4972,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.91942614062891e-05,
+      "loss": 0.511,
+      "step": 855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9192593210442907e-05,
+      "loss": 0.5112,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9190925014596717e-05,
+      "loss": 0.5573,
+      "step": 857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9189256818750523e-05,
+      "loss": 0.5899,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.918758862290433e-05,
+      "loss": 0.495,
+      "step": 859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.918592042705814e-05,
+      "loss": 0.4954,
+      "step": 860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9184252231211945e-05,
+      "loss": 0.5245,
+      "step": 861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9182584035365754e-05,
+      "loss": 0.5216,
+      "step": 862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.918091583951956e-05,
+      "loss": 0.5686,
+      "step": 863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.917924764367337e-05,
+      "loss": 0.5087,
+      "step": 864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9177579447827176e-05,
+      "loss": 0.5343,
+      "step": 865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9175911251980983e-05,
+      "loss": 0.5332,
+      "step": 866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9174243056134792e-05,
+      "loss": 0.5378,
+      "step": 867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.91725748602886e-05,
+      "loss": 0.4889,
+      "step": 868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9170906664442408e-05,
+      "loss": 0.5413,
+      "step": 869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9169238468596214e-05,
+      "loss": 0.5111,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9167570272750024e-05,
+      "loss": 0.5303,
+      "step": 871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916590207690383e-05,
+      "loss": 0.527,
+      "step": 872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9164233881057636e-05,
+      "loss": 0.4699,
+      "step": 873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9162565685211446e-05,
+      "loss": 0.5165,
+      "step": 874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9160897489365252e-05,
+      "loss": 0.4861,
+      "step": 875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.915922929351906e-05,
+      "loss": 0.5292,
+      "step": 876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9157561097672868e-05,
+      "loss": 0.5454,
+      "step": 877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9155892901826677e-05,
+      "loss": 0.5353,
+      "step": 878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9154224705980483e-05,
+      "loss": 0.5263,
+      "step": 879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.915255651013429e-05,
+      "loss": 0.531,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.91508883142881e-05,
+      "loss": 0.5371,
+      "step": 881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9149220118441905e-05,
+      "loss": 0.5419,
+      "step": 882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9147551922595715e-05,
+      "loss": 0.5052,
+      "step": 883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914588372674952e-05,
+      "loss": 0.5391,
+      "step": 884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914421553090333e-05,
+      "loss": 0.507,
+      "step": 885
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9142547335057137e-05,
+      "loss": 0.5038,
+      "step": 886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9140879139210943e-05,
+      "loss": 0.5342,
+      "step": 887
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9139210943364753e-05,
+      "loss": 0.4991,
+      "step": 888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913754274751856e-05,
+      "loss": 0.5255,
+      "step": 889
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913587455167237e-05,
+      "loss": 0.5407,
+      "step": 890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9134206355826175e-05,
+      "loss": 0.5381,
+      "step": 891
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9132538159979984e-05,
+      "loss": 0.5106,
+      "step": 892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913086996413379e-05,
+      "loss": 0.5249,
+      "step": 893
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9129201768287597e-05,
+      "loss": 0.5235,
+      "step": 894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9127533572441406e-05,
+      "loss": 0.4852,
+      "step": 895
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9125865376595212e-05,
+      "loss": 0.5279,
+      "step": 896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9124197180749022e-05,
+      "loss": 0.5357,
+      "step": 897
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9122528984902828e-05,
+      "loss": 0.4934,
+      "step": 898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9120860789056638e-05,
+      "loss": 0.5215,
+      "step": 899
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9119192593210444e-05,
+      "loss": 0.5373,
+      "step": 900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.911752439736425e-05,
+      "loss": 0.5349,
+      "step": 901
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.911585620151806e-05,
+      "loss": 0.5156,
+      "step": 902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9114188005671866e-05,
+      "loss": 0.5272,
+      "step": 903
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9112519809825676e-05,
+      "loss": 0.4778,
+      "step": 904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9110851613979482e-05,
+      "loss": 0.513,
+      "step": 905
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910918341813329e-05,
+      "loss": 0.5003,
+      "step": 906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9107515222287098e-05,
+      "loss": 0.5362,
+      "step": 907
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9105847026440907e-05,
+      "loss": 0.5551,
+      "step": 908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9104178830594713e-05,
+      "loss": 0.5061,
+      "step": 909
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910251063474852e-05,
+      "loss": 0.5735,
+      "step": 910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910084243890233e-05,
+      "loss": 0.5092,
+      "step": 911
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9099174243056135e-05,
+      "loss": 0.5413,
+      "step": 912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9097506047209945e-05,
+      "loss": 0.4973,
+      "step": 913
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.909583785136375e-05,
+      "loss": 0.4988,
+      "step": 914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.909416965551756e-05,
+      "loss": 0.5121,
+      "step": 915
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9092501459671367e-05,
+      "loss": 0.533,
+      "step": 916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9090833263825176e-05,
+      "loss": 0.5494,
+      "step": 917
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9089165067978983e-05,
+      "loss": 0.5429,
+      "step": 918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.908749687213279e-05,
+      "loss": 0.5475,
+      "step": 919
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.90858286762866e-05,
+      "loss": 0.5749,
+      "step": 920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9084160480440405e-05,
+      "loss": 0.553,
+      "step": 921
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9082492284594214e-05,
+      "loss": 0.4876,
+      "step": 922
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.908082408874802e-05,
+      "loss": 0.5526,
+      "step": 923
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.907915589290183e-05,
+      "loss": 0.5378,
+      "step": 924
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9077487697055636e-05,
+      "loss": 0.5465,
+      "step": 925
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9075819501209442e-05,
+      "loss": 0.5268,
+      "step": 926
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9074151305363252e-05,
+      "loss": 0.49,
+      "step": 927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9072483109517058e-05,
+      "loss": 0.5294,
+      "step": 928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9070814913670868e-05,
+      "loss": 0.5585,
+      "step": 929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9069146717824674e-05,
+      "loss": 0.5343,
+      "step": 930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9067478521978483e-05,
+      "loss": 0.498,
+      "step": 931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.906581032613229e-05,
+      "loss": 0.5507,
+      "step": 932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9064142130286096e-05,
+      "loss": 0.5081,
+      "step": 933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9062473934439905e-05,
+      "loss": 0.5056,
+      "step": 934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.906080573859371e-05,
+      "loss": 0.5118,
+      "step": 935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.905913754274752e-05,
+      "loss": 0.5088,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9057469346901327e-05,
+      "loss": 0.5588,
+      "step": 937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9055801151055137e-05,
+      "loss": 0.5423,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9054132955208943e-05,
+      "loss": 0.5133,
+      "step": 939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.905246475936275e-05,
+      "loss": 0.5322,
+      "step": 940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.905079656351656e-05,
+      "loss": 0.4869,
+      "step": 941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9049128367670365e-05,
+      "loss": 0.568,
+      "step": 942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9047460171824175e-05,
+      "loss": 0.5232,
+      "step": 943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.904579197597798e-05,
+      "loss": 0.5166,
+      "step": 944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.904412378013179e-05,
+      "loss": 0.5546,
+      "step": 945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9042455584285597e-05,
+      "loss": 0.5333,
+      "step": 946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9040787388439403e-05,
+      "loss": 0.5423,
+      "step": 947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9039119192593212e-05,
+      "loss": 0.5117,
+      "step": 948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.903745099674702e-05,
+      "loss": 0.5353,
+      "step": 949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9035782800900828e-05,
+      "loss": 0.5492,
+      "step": 950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9034114605054634e-05,
+      "loss": 0.5174,
+      "step": 951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9032446409208444e-05,
+      "loss": 0.5429,
+      "step": 952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.903077821336225e-05,
+      "loss": 0.5251,
+      "step": 953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9029110017516056e-05,
+      "loss": 0.5325,
+      "step": 954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9027441821669866e-05,
+      "loss": 0.5364,
+      "step": 955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9025773625823672e-05,
+      "loss": 0.5404,
+      "step": 956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9024105429977482e-05,
+      "loss": 0.5326,
+      "step": 957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9022437234131288e-05,
+      "loss": 0.5311,
+      "step": 958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9020769038285098e-05,
+      "loss": 0.5316,
+      "step": 959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9019100842438904e-05,
+      "loss": 0.4824,
+      "step": 960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.901743264659271e-05,
+      "loss": 0.5371,
+      "step": 961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.901576445074652e-05,
+      "loss": 0.5726,
+      "step": 962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9014096254900326e-05,
+      "loss": 0.5294,
+      "step": 963
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9012428059054135e-05,
+      "loss": 0.513,
+      "step": 964
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.901075986320794e-05,
+      "loss": 0.5146,
+      "step": 965
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.900909166736175e-05,
+      "loss": 0.4843,
+      "step": 966
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9007423471515557e-05,
+      "loss": 0.55,
+      "step": 967
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9005755275669363e-05,
+      "loss": 0.5013,
+      "step": 968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9004087079823173e-05,
+      "loss": 0.5218,
+      "step": 969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.900241888397698e-05,
+      "loss": 0.5268,
+      "step": 970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.900075068813079e-05,
+      "loss": 0.4796,
+      "step": 971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8999082492284595e-05,
+      "loss": 0.5529,
+      "step": 972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8997414296438405e-05,
+      "loss": 0.504,
+      "step": 973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.899574610059221e-05,
+      "loss": 0.5239,
+      "step": 974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8994077904746017e-05,
+      "loss": 0.522,
+      "step": 975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8992409708899827e-05,
+      "loss": 0.5225,
+      "step": 976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8990741513053633e-05,
+      "loss": 0.545,
+      "step": 977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8989073317207442e-05,
+      "loss": 0.5048,
+      "step": 978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.898740512136125e-05,
+      "loss": 0.4975,
+      "step": 979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8985736925515058e-05,
+      "loss": 0.5482,
+      "step": 980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8984068729668864e-05,
+      "loss": 0.527,
+      "step": 981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.898240053382267e-05,
+      "loss": 0.5069,
+      "step": 982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.898073233797648e-05,
+      "loss": 0.5204,
+      "step": 983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.897906414213029e-05,
+      "loss": 0.5162,
+      "step": 984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8977395946284096e-05,
+      "loss": 0.5083,
+      "step": 985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8975727750437902e-05,
+      "loss": 0.5603,
+      "step": 986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.897405955459171e-05,
+      "loss": 0.5569,
+      "step": 987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8972391358745518e-05,
+      "loss": 0.5019,
+      "step": 988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8970723162899324e-05,
+      "loss": 0.4989,
+      "step": 989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8969054967053134e-05,
+      "loss": 0.5591,
+      "step": 990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8967386771206943e-05,
+      "loss": 0.5617,
+      "step": 991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.896571857536075e-05,
+      "loss": 0.4935,
+      "step": 992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8964050379514556e-05,
+      "loss": 0.5366,
+      "step": 993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8962382183668365e-05,
+      "loss": 0.5203,
+      "step": 994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.896071398782217e-05,
+      "loss": 0.5109,
+      "step": 995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8959045791975977e-05,
+      "loss": 0.5512,
+      "step": 996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8957377596129787e-05,
+      "loss": 0.5057,
+      "step": 997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8955709400283597e-05,
+      "loss": 0.5373,
+      "step": 998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8954041204437403e-05,
+      "loss": 0.5104,
+      "step": 999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.895237300859121e-05,
+      "loss": 0.5255,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.895070481274502e-05,
+      "loss": 0.5268,
+      "step": 1001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8949036616898825e-05,
+      "loss": 0.5236,
+      "step": 1002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.4974,
+      "step": 1003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894570022520644e-05,
+      "loss": 0.5551,
+      "step": 1004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.894403202936025e-05,
+      "loss": 0.5325,
+      "step": 1005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8942363833514056e-05,
+      "loss": 0.5217,
+      "step": 1006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8940695637667863e-05,
+      "loss": 0.552,
+      "step": 1007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.8939027441821672e-05,
+      "loss": 0.5233,
+      "step": 1008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.893735924597548e-05,
+      "loss": 0.5299,
+      "step": 1009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8935691050129285e-05,
+      "loss": 0.5048,
+      "step": 1010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8934022854283094e-05,
+      "loss": 0.5369,
+      "step": 1011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8932354658436904e-05,
+      "loss": 0.5171,
+      "step": 1012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.893068646259071e-05,
+      "loss": 0.5402,
+      "step": 1013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8929018266744516e-05,
+      "loss": 0.4745,
+      "step": 1014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8927350070898326e-05,
+      "loss": 0.5332,
+      "step": 1015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8925681875052132e-05,
+      "loss": 0.5439,
+      "step": 1016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8924013679205938e-05,
+      "loss": 0.4946,
+      "step": 1017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8922345483359748e-05,
+      "loss": 0.5613,
+      "step": 1018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8920677287513557e-05,
+      "loss": 0.5207,
+      "step": 1019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8919009091667363e-05,
+      "loss": 0.5156,
+      "step": 1020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.891734089582117e-05,
+      "loss": 0.4977,
+      "step": 1021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.891567269997498e-05,
+      "loss": 0.4785,
+      "step": 1022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8914004504128785e-05,
+      "loss": 0.536,
+      "step": 1023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.891233630828259e-05,
+      "loss": 0.522,
+      "step": 1024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.89106681124364e-05,
+      "loss": 0.5128,
+      "step": 1025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.890899991659021e-05,
+      "loss": 0.5406,
+      "step": 1026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8907331720744017e-05,
+      "loss": 0.498,
+      "step": 1027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8905663524897823e-05,
+      "loss": 0.5202,
+      "step": 1028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8903995329051633e-05,
+      "loss": 0.5258,
+      "step": 1029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.890232713320544e-05,
+      "loss": 0.5232,
+      "step": 1030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.890065893735925e-05,
+      "loss": 0.5175,
+      "step": 1031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8898990741513055e-05,
+      "loss": 0.5463,
+      "step": 1032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8897322545666864e-05,
+      "loss": 0.5125,
+      "step": 1033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.889565434982067e-05,
+      "loss": 0.5472,
+      "step": 1034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8893986153974477e-05,
+      "loss": 0.5485,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8892317958128286e-05,
+      "loss": 0.5213,
+      "step": 1036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8890649762282092e-05,
+      "loss": 0.5147,
+      "step": 1037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8888981566435902e-05,
+      "loss": 0.5471,
+      "step": 1038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8887313370589708e-05,
+      "loss": 0.5013,
+      "step": 1039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8885645174743518e-05,
+      "loss": 0.5051,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8883976978897324e-05,
+      "loss": 0.5363,
+      "step": 1041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.888230878305113e-05,
+      "loss": 0.5232,
+      "step": 1042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.888064058720494e-05,
+      "loss": 0.5415,
+      "step": 1043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8878972391358746e-05,
+      "loss": 0.5401,
+      "step": 1044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8877304195512555e-05,
+      "loss": 0.5356,
+      "step": 1045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.887563599966636e-05,
+      "loss": 0.5124,
+      "step": 1046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.887396780382017e-05,
+      "loss": 0.516,
+      "step": 1047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8872299607973977e-05,
+      "loss": 0.5056,
+      "step": 1048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8870631412127784e-05,
+      "loss": 0.4753,
+      "step": 1049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8868963216281593e-05,
+      "loss": 0.5226,
+      "step": 1050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.88672950204354e-05,
+      "loss": 0.5271,
+      "step": 1051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.886562682458921e-05,
+      "loss": 0.529,
+      "step": 1052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8863958628743015e-05,
+      "loss": 0.5214,
+      "step": 1053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8862290432896825e-05,
+      "loss": 0.5578,
+      "step": 1054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.886062223705063e-05,
+      "loss": 0.5088,
+      "step": 1055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8858954041204437e-05,
+      "loss": 0.5241,
+      "step": 1056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8857285845358247e-05,
+      "loss": 0.5189,
+      "step": 1057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8855617649512056e-05,
+      "loss": 0.5272,
+      "step": 1058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8853949453665863e-05,
+      "loss": 0.5224,
+      "step": 1059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.885228125781967e-05,
+      "loss": 0.5007,
+      "step": 1060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8850613061973478e-05,
+      "loss": 0.5075,
+      "step": 1061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8848944866127284e-05,
+      "loss": 0.5161,
+      "step": 1062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.884727667028109e-05,
+      "loss": 0.5161,
+      "step": 1063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.88456084744349e-05,
+      "loss": 0.5075,
+      "step": 1064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.884394027858871e-05,
+      "loss": 0.5164,
+      "step": 1065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8842272082742516e-05,
+      "loss": 0.5072,
+      "step": 1066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8840603886896322e-05,
+      "loss": 0.4984,
+      "step": 1067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8838935691050132e-05,
+      "loss": 0.4971,
+      "step": 1068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8837267495203938e-05,
+      "loss": 0.5363,
+      "step": 1069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8835599299357744e-05,
+      "loss": 0.5297,
+      "step": 1070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8833931103511554e-05,
+      "loss": 0.5123,
+      "step": 1071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8832262907665363e-05,
+      "loss": 0.4967,
+      "step": 1072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.883059471181917e-05,
+      "loss": 0.4958,
+      "step": 1073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8828926515972976e-05,
+      "loss": 0.5391,
+      "step": 1074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8827258320126785e-05,
+      "loss": 0.5551,
+      "step": 1075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.882559012428059e-05,
+      "loss": 0.5253,
+      "step": 1076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8823921928434398e-05,
+      "loss": 0.5648,
+      "step": 1077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8822253732588207e-05,
+      "loss": 0.5254,
+      "step": 1078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8820585536742017e-05,
+      "loss": 0.5202,
+      "step": 1079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8818917340895823e-05,
+      "loss": 0.5308,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.881724914504963e-05,
+      "loss": 0.5425,
+      "step": 1081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.881558094920344e-05,
+      "loss": 0.5074,
+      "step": 1082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8813912753357245e-05,
+      "loss": 0.542,
+      "step": 1083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.881224455751105e-05,
+      "loss": 0.5182,
+      "step": 1084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.881057636166486e-05,
+      "loss": 0.5304,
+      "step": 1085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.880890816581867e-05,
+      "loss": 0.5552,
+      "step": 1086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8807239969972477e-05,
+      "loss": 0.5236,
+      "step": 1087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8805571774126283e-05,
+      "loss": 0.5393,
+      "step": 1088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8803903578280092e-05,
+      "loss": 0.5311,
+      "step": 1089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.88022353824339e-05,
+      "loss": 0.5649,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8800567186587705e-05,
+      "loss": 0.5136,
+      "step": 1091
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8798898990741514e-05,
+      "loss": 0.5519,
+      "step": 1092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8797230794895324e-05,
+      "loss": 0.5004,
+      "step": 1093
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.879556259904913e-05,
+      "loss": 0.5206,
+      "step": 1094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8793894403202936e-05,
+      "loss": 0.5073,
+      "step": 1095
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8792226207356746e-05,
+      "loss": 0.5156,
+      "step": 1096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8790558011510552e-05,
+      "loss": 0.5152,
+      "step": 1097
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8788889815664358e-05,
+      "loss": 0.5037,
+      "step": 1098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8787221619818168e-05,
+      "loss": 0.5747,
+      "step": 1099
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8785553423971977e-05,
+      "loss": 0.5135,
+      "step": 1100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8783885228125784e-05,
+      "loss": 0.5448,
+      "step": 1101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.878221703227959e-05,
+      "loss": 0.5013,
+      "step": 1102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.87805488364334e-05,
+      "loss": 0.557,
+      "step": 1103
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8778880640587206e-05,
+      "loss": 0.5315,
+      "step": 1104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8777212444741012e-05,
+      "loss": 0.5347,
+      "step": 1105
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.877554424889482e-05,
+      "loss": 0.5402,
+      "step": 1106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.877387605304863e-05,
+      "loss": 0.5359,
+      "step": 1107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8772207857202437e-05,
+      "loss": 0.5332,
+      "step": 1108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8770539661356243e-05,
+      "loss": 0.5693,
+      "step": 1109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8768871465510053e-05,
+      "loss": 0.5274,
+      "step": 1110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.876720326966386e-05,
+      "loss": 0.5496,
+      "step": 1111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8765535073817665e-05,
+      "loss": 0.502,
+      "step": 1112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8763866877971475e-05,
+      "loss": 0.5168,
+      "step": 1113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8762198682125284e-05,
+      "loss": 0.5178,
+      "step": 1114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.876053048627909e-05,
+      "loss": 0.5124,
+      "step": 1115
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8758862290432897e-05,
+      "loss": 0.5272,
+      "step": 1116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8757194094586706e-05,
+      "loss": 0.5311,
+      "step": 1117
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8755525898740513e-05,
+      "loss": 0.5073,
+      "step": 1118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.875385770289432e-05,
+      "loss": 0.5375,
+      "step": 1119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.875218950704813e-05,
+      "loss": 0.5005,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8750521311201938e-05,
+      "loss": 0.504,
+      "step": 1121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8748853115355744e-05,
+      "loss": 0.541,
+      "step": 1122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.874718491950955e-05,
+      "loss": 0.5092,
+      "step": 1123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.874551672366336e-05,
+      "loss": 0.5285,
+      "step": 1124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8743848527817166e-05,
+      "loss": 0.515,
+      "step": 1125
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8742180331970972e-05,
+      "loss": 0.5087,
+      "step": 1126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8740512136124782e-05,
+      "loss": 0.5309,
+      "step": 1127
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.873884394027859e-05,
+      "loss": 0.4875,
+      "step": 1128
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8737175744432398e-05,
+      "loss": 0.5769,
+      "step": 1129
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8735507548586204e-05,
+      "loss": 0.5344,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8733839352740013e-05,
+      "loss": 0.5325,
+      "step": 1131
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8732171156893823e-05,
+      "loss": 0.4903,
+      "step": 1132
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8730502961047626e-05,
+      "loss": 0.5214,
+      "step": 1133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8728834765201435e-05,
+      "loss": 0.5151,
+      "step": 1134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8727166569355245e-05,
+      "loss": 0.5437,
+      "step": 1135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.872549837350905e-05,
+      "loss": 0.4821,
+      "step": 1136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8723830177662857e-05,
+      "loss": 0.4938,
+      "step": 1137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8722161981816667e-05,
+      "loss": 0.5137,
+      "step": 1138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8720493785970477e-05,
+      "loss": 0.5136,
+      "step": 1139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.871882559012428e-05,
+      "loss": 0.5206,
+      "step": 1140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.871715739427809e-05,
+      "loss": 0.5348,
+      "step": 1141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.87154891984319e-05,
+      "loss": 0.5578,
+      "step": 1142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8713821002585705e-05,
+      "loss": 0.4835,
+      "step": 1143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.871215280673951e-05,
+      "loss": 0.5433,
+      "step": 1144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.871048461089332e-05,
+      "loss": 0.4843,
+      "step": 1145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.870881641504713e-05,
+      "loss": 0.4888,
+      "step": 1146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8707148219200936e-05,
+      "loss": 0.4909,
+      "step": 1147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8705480023354742e-05,
+      "loss": 0.5284,
+      "step": 1148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8703811827508552e-05,
+      "loss": 0.5225,
+      "step": 1149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8702143631662358e-05,
+      "loss": 0.5375,
+      "step": 1150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8700475435816164e-05,
+      "loss": 0.5372,
+      "step": 1151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8698807239969974e-05,
+      "loss": 0.4868,
+      "step": 1152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8697139044123784e-05,
+      "loss": 0.5112,
+      "step": 1153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.869547084827759e-05,
+      "loss": 0.5385,
+      "step": 1154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8693802652431396e-05,
+      "loss": 0.5325,
+      "step": 1155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8692134456585206e-05,
+      "loss": 0.5252,
+      "step": 1156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8690466260739012e-05,
+      "loss": 0.5504,
+      "step": 1157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8688798064892818e-05,
+      "loss": 0.516,
+      "step": 1158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8687129869046627e-05,
+      "loss": 0.5022,
+      "step": 1159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8685461673200437e-05,
+      "loss": 0.5493,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8683793477354243e-05,
+      "loss": 0.5172,
+      "step": 1161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.868212528150805e-05,
+      "loss": 0.5031,
+      "step": 1162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.868045708566186e-05,
+      "loss": 0.5001,
+      "step": 1163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8678788889815665e-05,
+      "loss": 0.5049,
+      "step": 1164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.867712069396947e-05,
+      "loss": 0.5127,
+      "step": 1165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.867545249812328e-05,
+      "loss": 0.5274,
+      "step": 1166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.867378430227709e-05,
+      "loss": 0.5359,
+      "step": 1167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8672116106430897e-05,
+      "loss": 0.5475,
+      "step": 1168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8670447910584703e-05,
+      "loss": 0.545,
+      "step": 1169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8668779714738513e-05,
+      "loss": 0.5067,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.866711151889232e-05,
+      "loss": 0.5301,
+      "step": 1171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8665443323046125e-05,
+      "loss": 0.5304,
+      "step": 1172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8663775127199935e-05,
+      "loss": 0.5234,
+      "step": 1173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8662106931353744e-05,
+      "loss": 0.4805,
+      "step": 1174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866043873550755e-05,
+      "loss": 0.5048,
+      "step": 1175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8658770539661356e-05,
+      "loss": 0.5458,
+      "step": 1176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8657102343815166e-05,
+      "loss": 0.5619,
+      "step": 1177
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8655434147968972e-05,
+      "loss": 0.512,
+      "step": 1178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865376595212278e-05,
+      "loss": 0.5199,
+      "step": 1179
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8652097756276588e-05,
+      "loss": 0.5077,
+      "step": 1180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8650429560430398e-05,
+      "loss": 0.5259,
+      "step": 1181
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8648761364584204e-05,
+      "loss": 0.5082,
+      "step": 1182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864709316873801e-05,
+      "loss": 0.4803,
+      "step": 1183
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864542497289182e-05,
+      "loss": 0.5159,
+      "step": 1184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8643756777045626e-05,
+      "loss": 0.5435,
+      "step": 1185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8642088581199432e-05,
+      "loss": 0.5599,
+      "step": 1186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864042038535324e-05,
+      "loss": 0.5327,
+      "step": 1187
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863875218950705e-05,
+      "loss": 0.5524,
+      "step": 1188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8637083993660857e-05,
+      "loss": 0.5243,
+      "step": 1189
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8635415797814664e-05,
+      "loss": 0.4756,
+      "step": 1190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8633747601968473e-05,
+      "loss": 0.5298,
+      "step": 1191
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863207940612228e-05,
+      "loss": 0.5192,
+      "step": 1192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8630411210276085e-05,
+      "loss": 0.5259,
+      "step": 1193
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8628743014429895e-05,
+      "loss": 0.4653,
+      "step": 1194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8627074818583705e-05,
+      "loss": 0.5216,
+      "step": 1195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862540662273751e-05,
+      "loss": 0.5126,
+      "step": 1196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8623738426891317e-05,
+      "loss": 0.4775,
+      "step": 1197
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622070231045127e-05,
+      "loss": 0.5161,
+      "step": 1198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8620402035198936e-05,
+      "loss": 0.4987,
+      "step": 1199
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861873383935274e-05,
+      "loss": 0.5127,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861706564350655e-05,
+      "loss": 0.5377,
+      "step": 1201
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8615397447660358e-05,
+      "loss": 0.4964,
+      "step": 1202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8613729251814164e-05,
+      "loss": 0.522,
+      "step": 1203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861206105596797e-05,
+      "loss": 0.5271,
+      "step": 1204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861039286012178e-05,
+      "loss": 0.5079,
+      "step": 1205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860872466427559e-05,
+      "loss": 0.508,
+      "step": 1206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8607056468429393e-05,
+      "loss": 0.5167,
+      "step": 1207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8605388272583202e-05,
+      "loss": 0.52,
+      "step": 1208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860372007673701e-05,
+      "loss": 0.5576,
+      "step": 1209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8602051880890818e-05,
+      "loss": 0.4808,
+      "step": 1210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8600383685044624e-05,
+      "loss": 0.5336,
+      "step": 1211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8598715489198434e-05,
+      "loss": 0.5537,
+      "step": 1212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8597047293352243e-05,
+      "loss": 0.5098,
+      "step": 1213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859537909750605e-05,
+      "loss": 0.528,
+      "step": 1214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8593710901659856e-05,
+      "loss": 0.5347,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8592042705813665e-05,
+      "loss": 0.5411,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.859037450996747e-05,
+      "loss": 0.5337,
+      "step": 1217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8588706314121278e-05,
+      "loss": 0.5279,
+      "step": 1218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8587038118275087e-05,
+      "loss": 0.5174,
+      "step": 1219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8585369922428897e-05,
+      "loss": 0.4953,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8583701726582703e-05,
+      "loss": 0.4701,
+      "step": 1221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.858203353073651e-05,
+      "loss": 0.4867,
+      "step": 1222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.858036533489032e-05,
+      "loss": 0.507,
+      "step": 1223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8578697139044125e-05,
+      "loss": 0.4925,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857702894319793e-05,
+      "loss": 0.5125,
+      "step": 1225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857536074735174e-05,
+      "loss": 0.5383,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.857369255150555e-05,
+      "loss": 0.492,
+      "step": 1227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8572024355659356e-05,
+      "loss": 0.5674,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8570356159813163e-05,
+      "loss": 0.5149,
+      "step": 1229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8568687963966972e-05,
+      "loss": 0.5241,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.856701976812078e-05,
+      "loss": 0.5124,
+      "step": 1231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8565351572274585e-05,
+      "loss": 0.5093,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8563683376428394e-05,
+      "loss": 0.5272,
+      "step": 1233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8562015180582204e-05,
+      "loss": 0.4726,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.856034698473601e-05,
+      "loss": 0.5318,
+      "step": 1235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8558678788889816e-05,
+      "loss": 0.5586,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8557010593043626e-05,
+      "loss": 0.49,
+      "step": 1237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8555342397197432e-05,
+      "loss": 0.4858,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8553674201351238e-05,
+      "loss": 0.5223,
+      "step": 1239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8552006005505048e-05,
+      "loss": 0.5073,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8550337809658857e-05,
+      "loss": 0.5318,
+      "step": 1241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8548669613812663e-05,
+      "loss": 0.5093,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.854700141796647e-05,
+      "loss": 0.4762,
+      "step": 1243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.854533322212028e-05,
+      "loss": 0.5164,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8543665026274085e-05,
+      "loss": 0.5433,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.854199683042789e-05,
+      "loss": 0.4726,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.85403286345817e-05,
+      "loss": 0.4948,
+      "step": 1247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853866043873551e-05,
+      "loss": 0.5504,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8536992242889317e-05,
+      "loss": 0.5729,
+      "step": 1249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8535324047043123e-05,
+      "loss": 0.5213,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8533655851196933e-05,
+      "loss": 0.5406,
+      "step": 1251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853198765535074e-05,
+      "loss": 0.5198,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8530319459504545e-05,
+      "loss": 0.5193,
+      "step": 1253
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8528651263658355e-05,
+      "loss": 0.5036,
+      "step": 1254
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8526983067812164e-05,
+      "loss": 0.5321,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852531487196597e-05,
+      "loss": 0.5099,
+      "step": 1256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8523646676119777e-05,
+      "loss": 0.5229,
+      "step": 1257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8521978480273586e-05,
+      "loss": 0.5149,
+      "step": 1258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8520310284427392e-05,
+      "loss": 0.5001,
+      "step": 1259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.85186420885812e-05,
+      "loss": 0.5334,
+      "step": 1260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8516973892735008e-05,
+      "loss": 0.4522,
+      "step": 1261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8515305696888818e-05,
+      "loss": 0.5184,
+      "step": 1262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8513637501042624e-05,
+      "loss": 0.5171,
+      "step": 1263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.851196930519643e-05,
+      "loss": 0.5288,
+      "step": 1264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.851030110935024e-05,
+      "loss": 0.4885,
+      "step": 1265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8508632913504046e-05,
+      "loss": 0.5236,
+      "step": 1266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8506964717657852e-05,
+      "loss": 0.5093,
+      "step": 1267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8505296521811662e-05,
+      "loss": 0.519,
+      "step": 1268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.850362832596547e-05,
+      "loss": 0.5385,
+      "step": 1269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8501960130119278e-05,
+      "loss": 0.5258,
+      "step": 1270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8500291934273084e-05,
+      "loss": 0.5371,
+      "step": 1271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8498623738426893e-05,
+      "loss": 0.5244,
+      "step": 1272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8496955542580703e-05,
+      "loss": 0.5095,
+      "step": 1273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8495287346734506e-05,
+      "loss": 0.5081,
+      "step": 1274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8493619150888315e-05,
+      "loss": 0.4944,
+      "step": 1275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8491950955042125e-05,
+      "loss": 0.5159,
+      "step": 1276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.849028275919593e-05,
+      "loss": 0.5409,
+      "step": 1277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8488614563349737e-05,
+      "loss": 0.5312,
+      "step": 1278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8486946367503547e-05,
+      "loss": 0.4968,
+      "step": 1279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8485278171657356e-05,
+      "loss": 0.5185,
+      "step": 1280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.848360997581116e-05,
+      "loss": 0.519,
+      "step": 1281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.848194177996497e-05,
+      "loss": 0.485,
+      "step": 1282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.848027358411878e-05,
+      "loss": 0.5444,
+      "step": 1283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8478605388272585e-05,
+      "loss": 0.511,
+      "step": 1284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.847693719242639e-05,
+      "loss": 0.4794,
+      "step": 1285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.84752689965802e-05,
+      "loss": 0.542,
+      "step": 1286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.847360080073401e-05,
+      "loss": 0.5203,
+      "step": 1287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8471932604887816e-05,
+      "loss": 0.5049,
+      "step": 1288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8470264409041622e-05,
+      "loss": 0.5176,
+      "step": 1289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8468596213195432e-05,
+      "loss": 0.5126,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8466928017349238e-05,
+      "loss": 0.5025,
+      "step": 1291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8465259821503044e-05,
+      "loss": 0.4989,
+      "step": 1292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8463591625656854e-05,
+      "loss": 0.5223,
+      "step": 1293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8461923429810663e-05,
+      "loss": 0.5386,
+      "step": 1294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.846025523396447e-05,
+      "loss": 0.5207,
+      "step": 1295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8458587038118276e-05,
+      "loss": 0.5497,
+      "step": 1296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8456918842272085e-05,
+      "loss": 0.5312,
+      "step": 1297
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.845525064642589e-05,
+      "loss": 0.5034,
+      "step": 1298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8453582450579698e-05,
+      "loss": 0.5133,
+      "step": 1299
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8451914254733507e-05,
+      "loss": 0.5132,
+      "step": 1300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8450246058887317e-05,
+      "loss": 0.5421,
+      "step": 1301
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8448577863041123e-05,
+      "loss": 0.529,
+      "step": 1302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844690966719493e-05,
+      "loss": 0.5057,
+      "step": 1303
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844524147134874e-05,
+      "loss": 0.5163,
+      "step": 1304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8443573275502545e-05,
+      "loss": 0.5003,
+      "step": 1305
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844190507965635e-05,
+      "loss": 0.4851,
+      "step": 1306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.844023688381016e-05,
+      "loss": 0.5237,
+      "step": 1307
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.843856868796397e-05,
+      "loss": 0.5219,
+      "step": 1308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8436900492117777e-05,
+      "loss": 0.5605,
+      "step": 1309
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8435232296271583e-05,
+      "loss": 0.4873,
+      "step": 1310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8433564100425392e-05,
+      "loss": 0.5515,
+      "step": 1311
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.84318959045792e-05,
+      "loss": 0.529,
+      "step": 1312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8430227708733005e-05,
+      "loss": 0.5185,
+      "step": 1313
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8428559512886814e-05,
+      "loss": 0.5268,
+      "step": 1314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8426891317040624e-05,
+      "loss": 0.5147,
+      "step": 1315
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.842522312119443e-05,
+      "loss": 0.4907,
+      "step": 1316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8423554925348236e-05,
+      "loss": 0.5141,
+      "step": 1317
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8421886729502046e-05,
+      "loss": 0.539,
+      "step": 1318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8420218533655852e-05,
+      "loss": 0.5047,
+      "step": 1319
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.841855033780966e-05,
+      "loss": 0.502,
+      "step": 1320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8416882141963468e-05,
+      "loss": 0.4939,
+      "step": 1321
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8415213946117278e-05,
+      "loss": 0.5482,
+      "step": 1322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8413545750271084e-05,
+      "loss": 0.5196,
+      "step": 1323
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.841187755442489e-05,
+      "loss": 0.5402,
+      "step": 1324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.84102093585787e-05,
+      "loss": 0.5188,
+      "step": 1325
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8408541162732506e-05,
+      "loss": 0.4892,
+      "step": 1326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8406872966886312e-05,
+      "loss": 0.5353,
+      "step": 1327
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.840520477104012e-05,
+      "loss": 0.566,
+      "step": 1328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.840353657519393e-05,
+      "loss": 0.5103,
+      "step": 1329
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8401868379347737e-05,
+      "loss": 0.4938,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8400200183501543e-05,
+      "loss": 0.5381,
+      "step": 1331
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8398531987655353e-05,
+      "loss": 0.5114,
+      "step": 1332
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.839686379180916e-05,
+      "loss": 0.5166,
+      "step": 1333
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8395195595962965e-05,
+      "loss": 0.4902,
+      "step": 1334
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8393527400116775e-05,
+      "loss": 0.5081,
+      "step": 1335
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8391859204270585e-05,
+      "loss": 0.5381,
+      "step": 1336
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.839019100842439e-05,
+      "loss": 0.5243,
+      "step": 1337
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8388522812578197e-05,
+      "loss": 0.5346,
+      "step": 1338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8386854616732007e-05,
+      "loss": 0.542,
+      "step": 1339
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8385186420885816e-05,
+      "loss": 0.5122,
+      "step": 1340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.838351822503962e-05,
+      "loss": 0.4958,
+      "step": 1341
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.838185002919343e-05,
+      "loss": 0.5283,
+      "step": 1342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8380181833347238e-05,
+      "loss": 0.5017,
+      "step": 1343
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8378513637501044e-05,
+      "loss": 0.5265,
+      "step": 1344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.837684544165485e-05,
+      "loss": 0.4861,
+      "step": 1345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.837517724580866e-05,
+      "loss": 0.4716,
+      "step": 1346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.837350904996247e-05,
+      "loss": 0.5544,
+      "step": 1347
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8371840854116272e-05,
+      "loss": 0.5463,
+      "step": 1348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8370172658270082e-05,
+      "loss": 0.5389,
+      "step": 1349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.836850446242389e-05,
+      "loss": 0.5221,
+      "step": 1350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8366836266577698e-05,
+      "loss": 0.5358,
+      "step": 1351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8365168070731504e-05,
+      "loss": 0.5044,
+      "step": 1352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8363499874885314e-05,
+      "loss": 0.5487,
+      "step": 1353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8361831679039123e-05,
+      "loss": 0.5328,
+      "step": 1354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.836016348319293e-05,
+      "loss": 0.515,
+      "step": 1355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8358495287346736e-05,
+      "loss": 0.5427,
+      "step": 1356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8356827091500545e-05,
+      "loss": 0.5244,
+      "step": 1357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.835515889565435e-05,
+      "loss": 0.5227,
+      "step": 1358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8353490699808157e-05,
+      "loss": 0.485,
+      "step": 1359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8351822503961967e-05,
+      "loss": 0.5397,
+      "step": 1360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8350154308115777e-05,
+      "loss": 0.4975,
+      "step": 1361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8348486112269583e-05,
+      "loss": 0.5098,
+      "step": 1362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.834681791642339e-05,
+      "loss": 0.5472,
+      "step": 1363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.83451497205772e-05,
+      "loss": 0.5246,
+      "step": 1364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8343481524731005e-05,
+      "loss": 0.5145,
+      "step": 1365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.834181332888481e-05,
+      "loss": 0.5115,
+      "step": 1366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.834014513303862e-05,
+      "loss": 0.5135,
+      "step": 1367
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.833847693719243e-05,
+      "loss": 0.5271,
+      "step": 1368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8336808741346236e-05,
+      "loss": 0.4875,
+      "step": 1369
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8335140545500043e-05,
+      "loss": 0.5079,
+      "step": 1370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8333472349653852e-05,
+      "loss": 0.5099,
+      "step": 1371
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.833180415380766e-05,
+      "loss": 0.4888,
+      "step": 1372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8330135957961464e-05,
+      "loss": 0.5014,
+      "step": 1373
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8328467762115274e-05,
+      "loss": 0.5074,
+      "step": 1374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8326799566269084e-05,
+      "loss": 0.5098,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.832513137042289e-05,
+      "loss": 0.5087,
+      "step": 1376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8323463174576696e-05,
+      "loss": 0.5028,
+      "step": 1377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8321794978730506e-05,
+      "loss": 0.46,
+      "step": 1378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8320126782884312e-05,
+      "loss": 0.5386,
+      "step": 1379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8318458587038118e-05,
+      "loss": 0.5003,
+      "step": 1380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8316790391191928e-05,
+      "loss": 0.5291,
+      "step": 1381
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8315122195345737e-05,
+      "loss": 0.5139,
+      "step": 1382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8313453999499543e-05,
+      "loss": 0.5354,
+      "step": 1383
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.831178580365335e-05,
+      "loss": 0.5363,
+      "step": 1384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.831011760780716e-05,
+      "loss": 0.5285,
+      "step": 1385
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8308449411960965e-05,
+      "loss": 0.5739,
+      "step": 1386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.830678121611477e-05,
+      "loss": 0.4711,
+      "step": 1387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.830511302026858e-05,
+      "loss": 0.4862,
+      "step": 1388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.830344482442239e-05,
+      "loss": 0.513,
+      "step": 1389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8301776628576197e-05,
+      "loss": 0.4987,
+      "step": 1390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8300108432730003e-05,
+      "loss": 0.4763,
+      "step": 1391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8298440236883813e-05,
+      "loss": 0.5222,
+      "step": 1392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.829677204103762e-05,
+      "loss": 0.4853,
+      "step": 1393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8295103845191425e-05,
+      "loss": 0.4966,
+      "step": 1394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8293435649345235e-05,
+      "loss": 0.5279,
+      "step": 1395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8291767453499044e-05,
+      "loss": 0.5311,
+      "step": 1396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.829009925765285e-05,
+      "loss": 0.4815,
+      "step": 1397
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8288431061806657e-05,
+      "loss": 0.5136,
+      "step": 1398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8286762865960466e-05,
+      "loss": 0.4994,
+      "step": 1399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8285094670114272e-05,
+      "loss": 0.4955,
+      "step": 1400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.828342647426808e-05,
+      "loss": 0.5586,
+      "step": 1401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8281758278421888e-05,
+      "loss": 0.4877,
+      "step": 1402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8280090082575698e-05,
+      "loss": 0.5041,
+      "step": 1403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8278421886729504e-05,
+      "loss": 0.4871,
+      "step": 1404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.827675369088331e-05,
+      "loss": 0.5086,
+      "step": 1405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.827508549503712e-05,
+      "loss": 0.4817,
+      "step": 1406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8273417299190926e-05,
+      "loss": 0.5129,
+      "step": 1407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8271749103344732e-05,
+      "loss": 0.4968,
+      "step": 1408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.827008090749854e-05,
+      "loss": 0.5223,
+      "step": 1409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.826841271165235e-05,
+      "loss": 0.5211,
+      "step": 1410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8266744515806157e-05,
+      "loss": 0.51,
+      "step": 1411
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8265076319959964e-05,
+      "loss": 0.5113,
+      "step": 1412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8263408124113773e-05,
+      "loss": 0.517,
+      "step": 1413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8261739928267583e-05,
+      "loss": 0.5521,
+      "step": 1414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8260071732421386e-05,
+      "loss": 0.5198,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8258403536575195e-05,
+      "loss": 0.5437,
+      "step": 1416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8256735340729005e-05,
+      "loss": 0.5075,
+      "step": 1417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.825506714488281e-05,
+      "loss": 0.5303,
+      "step": 1418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8253398949036617e-05,
+      "loss": 0.5388,
+      "step": 1419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8251730753190427e-05,
+      "loss": 0.489,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8250062557344236e-05,
+      "loss": 0.4841,
+      "step": 1421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.824839436149804e-05,
+      "loss": 0.4912,
+      "step": 1422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.824672616565185e-05,
+      "loss": 0.5125,
+      "step": 1423
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8245057969805658e-05,
+      "loss": 0.4862,
+      "step": 1424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8243389773959464e-05,
+      "loss": 0.5344,
+      "step": 1425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.824172157811327e-05,
+      "loss": 0.4946,
+      "step": 1426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.824005338226708e-05,
+      "loss": 0.5042,
+      "step": 1427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.823838518642089e-05,
+      "loss": 0.4744,
+      "step": 1428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8236716990574696e-05,
+      "loss": 0.5343,
+      "step": 1429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8235048794728502e-05,
+      "loss": 0.5365,
+      "step": 1430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8233380598882312e-05,
+      "loss": 0.5118,
+      "step": 1431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8231712403036118e-05,
+      "loss": 0.5111,
+      "step": 1432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8230044207189924e-05,
+      "loss": 0.5219,
+      "step": 1433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8228376011343734e-05,
+      "loss": 0.472,
+      "step": 1434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8226707815497543e-05,
+      "loss": 0.4843,
+      "step": 1435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.822503961965135e-05,
+      "loss": 0.5304,
+      "step": 1436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8223371423805156e-05,
+      "loss": 0.484,
+      "step": 1437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8221703227958965e-05,
+      "loss": 0.5126,
+      "step": 1438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.822003503211277e-05,
+      "loss": 0.5599,
+      "step": 1439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8218366836266578e-05,
+      "loss": 0.467,
+      "step": 1440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8216698640420387e-05,
+      "loss": 0.5086,
+      "step": 1441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8215030444574197e-05,
+      "loss": 0.4851,
+      "step": 1442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8213362248728003e-05,
+      "loss": 0.5223,
+      "step": 1443
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.821169405288181e-05,
+      "loss": 0.5212,
+      "step": 1444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.821002585703562e-05,
+      "loss": 0.5609,
+      "step": 1445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8208357661189425e-05,
+      "loss": 0.5406,
+      "step": 1446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.820668946534323e-05,
+      "loss": 0.5426,
+      "step": 1447
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.820502126949704e-05,
+      "loss": 0.5461,
+      "step": 1448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.820335307365085e-05,
+      "loss": 0.5201,
+      "step": 1449
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8201684877804657e-05,
+      "loss": 0.5202,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8200016681958463e-05,
+      "loss": 0.5487,
+      "step": 1451
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8198348486112272e-05,
+      "loss": 0.4977,
+      "step": 1452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.819668029026608e-05,
+      "loss": 0.5362,
+      "step": 1453
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8195012094419885e-05,
+      "loss": 0.5406,
+      "step": 1454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8193343898573694e-05,
+      "loss": 0.5023,
+      "step": 1455
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8191675702727504e-05,
+      "loss": 0.5283,
+      "step": 1456
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.819000750688131e-05,
+      "loss": 0.501,
+      "step": 1457
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8188339311035116e-05,
+      "loss": 0.5349,
+      "step": 1458
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8186671115188926e-05,
+      "loss": 0.4903,
+      "step": 1459
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8185002919342732e-05,
+      "loss": 0.5316,
+      "step": 1460
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8183334723496538e-05,
+      "loss": 0.4944,
+      "step": 1461
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8181666527650348e-05,
+      "loss": 0.5452,
+      "step": 1462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8179998331804157e-05,
+      "loss": 0.4918,
+      "step": 1463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8178330135957964e-05,
+      "loss": 0.5243,
+      "step": 1464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.817666194011177e-05,
+      "loss": 0.5018,
+      "step": 1465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.817499374426558e-05,
+      "loss": 0.5074,
+      "step": 1466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8173325548419386e-05,
+      "loss": 0.5368,
+      "step": 1467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8171657352573192e-05,
+      "loss": 0.5379,
+      "step": 1468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8169989156727e-05,
+      "loss": 0.5707,
+      "step": 1469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.816832096088081e-05,
+      "loss": 0.5218,
+      "step": 1470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8166652765034617e-05,
+      "loss": 0.5023,
+      "step": 1471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8164984569188423e-05,
+      "loss": 0.5009,
+      "step": 1472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8163316373342233e-05,
+      "loss": 0.5473,
+      "step": 1473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.816164817749604e-05,
+      "loss": 0.5166,
+      "step": 1474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8159979981649845e-05,
+      "loss": 0.504,
+      "step": 1475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8158311785803655e-05,
+      "loss": 0.5279,
+      "step": 1476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8156643589957464e-05,
+      "loss": 0.5279,
+      "step": 1477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.815497539411127e-05,
+      "loss": 0.5402,
+      "step": 1478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8153307198265077e-05,
+      "loss": 0.4989,
+      "step": 1479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8151639002418886e-05,
+      "loss": 0.467,
+      "step": 1480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8149970806572693e-05,
+      "loss": 0.5119,
+      "step": 1481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.81483026107265e-05,
+      "loss": 0.5054,
+      "step": 1482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.814663441488031e-05,
+      "loss": 0.4797,
+      "step": 1483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8144966219034118e-05,
+      "loss": 0.5077,
+      "step": 1484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8143298023187924e-05,
+      "loss": 0.53,
+      "step": 1485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.814162982734173e-05,
+      "loss": 0.515,
+      "step": 1486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.813996163149554e-05,
+      "loss": 0.508,
+      "step": 1487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8138293435649346e-05,
+      "loss": 0.5176,
+      "step": 1488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8136625239803152e-05,
+      "loss": 0.4898,
+      "step": 1489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8134957043956962e-05,
+      "loss": 0.5241,
+      "step": 1490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.813328884811077e-05,
+      "loss": 0.5316,
+      "step": 1491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8131620652264578e-05,
+      "loss": 0.4859,
+      "step": 1492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8129952456418384e-05,
+      "loss": 0.5162,
+      "step": 1493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8128284260572193e-05,
+      "loss": 0.4885,
+      "step": 1494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8126616064726e-05,
+      "loss": 0.5168,
+      "step": 1495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.812494786887981e-05,
+      "loss": 0.5199,
+      "step": 1496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8123279673033615e-05,
+      "loss": 0.4976,
+      "step": 1497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8121611477187425e-05,
+      "loss": 0.5432,
+      "step": 1498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.811994328134123e-05,
+      "loss": 0.531,
+      "step": 1499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8118275085495037e-05,
+      "loss": 0.5293,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8116606889648847e-05,
+      "loss": 0.5024,
+      "step": 1501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8114938693802657e-05,
+      "loss": 0.4748,
+      "step": 1502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8113270497956463e-05,
+      "loss": 0.4975,
+      "step": 1503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.811160230211027e-05,
+      "loss": 0.5101,
+      "step": 1504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810993410626408e-05,
+      "loss": 0.4959,
+      "step": 1505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8108265910417885e-05,
+      "loss": 0.5694,
+      "step": 1506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810659771457169e-05,
+      "loss": 0.524,
+      "step": 1507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.81049295187255e-05,
+      "loss": 0.545,
+      "step": 1508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.810326132287931e-05,
+      "loss": 0.5597,
+      "step": 1509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8101593127033116e-05,
+      "loss": 0.4748,
+      "step": 1510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8099924931186922e-05,
+      "loss": 0.5486,
+      "step": 1511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8098256735340732e-05,
+      "loss": 0.5822,
+      "step": 1512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8096588539494538e-05,
+      "loss": 0.5176,
+      "step": 1513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8094920343648344e-05,
+      "loss": 0.4983,
+      "step": 1514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8093252147802154e-05,
+      "loss": 0.5191,
+      "step": 1515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8091583951955964e-05,
+      "loss": 0.5011,
+      "step": 1516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.808991575610977e-05,
+      "loss": 0.4868,
+      "step": 1517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8088247560263576e-05,
+      "loss": 0.5249,
+      "step": 1518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8086579364417386e-05,
+      "loss": 0.5216,
+      "step": 1519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8084911168571192e-05,
+      "loss": 0.5191,
+      "step": 1520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8083242972724998e-05,
+      "loss": 0.5108,
+      "step": 1521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8081574776878807e-05,
+      "loss": 0.4845,
+      "step": 1522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8079906581032617e-05,
+      "loss": 0.4682,
+      "step": 1523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8078238385186423e-05,
+      "loss": 0.4598,
+      "step": 1524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.807657018934023e-05,
+      "loss": 0.5788,
+      "step": 1525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.807490199349404e-05,
+      "loss": 0.4896,
+      "step": 1526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8073233797647845e-05,
+      "loss": 0.5347,
+      "step": 1527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.807156560180165e-05,
+      "loss": 0.4755,
+      "step": 1528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.806989740595546e-05,
+      "loss": 0.5073,
+      "step": 1529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.806822921010927e-05,
+      "loss": 0.508,
+      "step": 1530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8066561014263077e-05,
+      "loss": 0.5222,
+      "step": 1531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8064892818416883e-05,
+      "loss": 0.517,
+      "step": 1532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8063224622570693e-05,
+      "loss": 0.4773,
+      "step": 1533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.80615564267245e-05,
+      "loss": 0.466,
+      "step": 1534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8059888230878305e-05,
+      "loss": 0.4988,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8058220035032115e-05,
+      "loss": 0.5272,
+      "step": 1536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8056551839185924e-05,
+      "loss": 0.5252,
+      "step": 1537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.805488364333973e-05,
+      "loss": 0.5268,
+      "step": 1538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8053215447493536e-05,
+      "loss": 0.5228,
+      "step": 1539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8051547251647346e-05,
+      "loss": 0.5317,
+      "step": 1540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8049879055801152e-05,
+      "loss": 0.4792,
+      "step": 1541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.804821085995496e-05,
+      "loss": 0.4924,
+      "step": 1542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8046542664108768e-05,
+      "loss": 0.5315,
+      "step": 1543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8044874468262578e-05,
+      "loss": 0.4897,
+      "step": 1544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8043206272416384e-05,
+      "loss": 0.5234,
+      "step": 1545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.804153807657019e-05,
+      "loss": 0.5147,
+      "step": 1546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8039869880724e-05,
+      "loss": 0.4908,
+      "step": 1547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8038201684877806e-05,
+      "loss": 0.5398,
+      "step": 1548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8036533489031612e-05,
+      "loss": 0.542,
+      "step": 1549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.803486529318542e-05,
+      "loss": 0.511,
+      "step": 1550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.803319709733923e-05,
+      "loss": 0.5191,
+      "step": 1551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8031528901493037e-05,
+      "loss": 0.4826,
+      "step": 1552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8029860705646844e-05,
+      "loss": 0.5351,
+      "step": 1553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8028192509800653e-05,
+      "loss": 0.4901,
+      "step": 1554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.802652431395446e-05,
+      "loss": 0.5166,
+      "step": 1555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8024856118108265e-05,
+      "loss": 0.5333,
+      "step": 1556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8023187922262075e-05,
+      "loss": 0.5153,
+      "step": 1557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8021519726415885e-05,
+      "loss": 0.4868,
+      "step": 1558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.801985153056969e-05,
+      "loss": 0.5143,
+      "step": 1559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8018183334723497e-05,
+      "loss": 0.5062,
+      "step": 1560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8016515138877307e-05,
+      "loss": 0.4964,
+      "step": 1561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8014846943031113e-05,
+      "loss": 0.4981,
+      "step": 1562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.801317874718492e-05,
+      "loss": 0.4979,
+      "step": 1563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.801151055133873e-05,
+      "loss": 0.4949,
+      "step": 1564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8009842355492538e-05,
+      "loss": 0.518,
+      "step": 1565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8008174159646344e-05,
+      "loss": 0.5228,
+      "step": 1566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.800650596380015e-05,
+      "loss": 0.4801,
+      "step": 1567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.800483776795396e-05,
+      "loss": 0.5197,
+      "step": 1568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8003169572107766e-05,
+      "loss": 0.5232,
+      "step": 1569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8001501376261576e-05,
+      "loss": 0.5026,
+      "step": 1570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7999833180415382e-05,
+      "loss": 0.4884,
+      "step": 1571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.799816498456919e-05,
+      "loss": 0.585,
+      "step": 1572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7996496788722998e-05,
+      "loss": 0.516,
+      "step": 1573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7994828592876804e-05,
+      "loss": 0.5264,
+      "step": 1574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7993160397030614e-05,
+      "loss": 0.5251,
+      "step": 1575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.799149220118442e-05,
+      "loss": 0.4992,
+      "step": 1576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.798982400533823e-05,
+      "loss": 0.5037,
+      "step": 1577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7988155809492036e-05,
+      "loss": 0.4831,
+      "step": 1578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7986487613645845e-05,
+      "loss": 0.5049,
+      "step": 1579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.798481941779965e-05,
+      "loss": 0.476,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7983151221953458e-05,
+      "loss": 0.4821,
+      "step": 1581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7981483026107267e-05,
+      "loss": 0.5204,
+      "step": 1582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7979814830261073e-05,
+      "loss": 0.5478,
+      "step": 1583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.7978146634414883e-05,
+      "loss": 0.563,
+      "step": 1584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.797647843856869e-05,
+      "loss": 0.5406,
+      "step": 1585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.79748102427225e-05,
+      "loss": 0.4898,
+      "step": 1586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7973142046876305e-05,
+      "loss": 0.5371,
+      "step": 1587
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.797147385103011e-05,
+      "loss": 0.5383,
+      "step": 1588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.796980565518392e-05,
+      "loss": 0.4807,
+      "step": 1589
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7968137459337727e-05,
+      "loss": 0.5284,
+      "step": 1590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7966469263491536e-05,
+      "loss": 0.4967,
+      "step": 1591
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7964801067645343e-05,
+      "loss": 0.5214,
+      "step": 1592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7963132871799152e-05,
+      "loss": 0.5311,
+      "step": 1593
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.796146467595296e-05,
+      "loss": 0.5338,
+      "step": 1594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7959796480106765e-05,
+      "loss": 0.5378,
+      "step": 1595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7958128284260574e-05,
+      "loss": 0.5298,
+      "step": 1596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.795646008841438e-05,
+      "loss": 0.5031,
+      "step": 1597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.795479189256819e-05,
+      "loss": 0.5465,
+      "step": 1598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7953123696721996e-05,
+      "loss": 0.4969,
+      "step": 1599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7951455500875806e-05,
+      "loss": 0.5155,
+      "step": 1600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7949787305029612e-05,
+      "loss": 0.4848,
+      "step": 1601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7948119109183418e-05,
+      "loss": 0.4602,
+      "step": 1602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7946450913337228e-05,
+      "loss": 0.5297,
+      "step": 1603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7944782717491034e-05,
+      "loss": 0.5309,
+      "step": 1604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7943114521644843e-05,
+      "loss": 0.5092,
+      "step": 1605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.794144632579865e-05,
+      "loss": 0.521,
+      "step": 1606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793977812995246e-05,
+      "loss": 0.5501,
+      "step": 1607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7938109934106265e-05,
+      "loss": 0.5049,
+      "step": 1608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793644173826007e-05,
+      "loss": 0.4876,
+      "step": 1609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.793477354241388e-05,
+      "loss": 0.5042,
+      "step": 1610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7933105346567687e-05,
+      "loss": 0.5125,
+      "step": 1611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7931437150721497e-05,
+      "loss": 0.529,
+      "step": 1612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7929768954875303e-05,
+      "loss": 0.5264,
+      "step": 1613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7928100759029113e-05,
+      "loss": 0.4949,
+      "step": 1614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.792643256318292e-05,
+      "loss": 0.4582,
+      "step": 1615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7924764367336725e-05,
+      "loss": 0.4988,
+      "step": 1616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7923096171490535e-05,
+      "loss": 0.5236,
+      "step": 1617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.792142797564434e-05,
+      "loss": 0.5061,
+      "step": 1618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.791975977979815e-05,
+      "loss": 0.52,
+      "step": 1619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7918091583951957e-05,
+      "loss": 0.5056,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7916423388105766e-05,
+      "loss": 0.5551,
+      "step": 1621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7914755192259572e-05,
+      "loss": 0.5185,
+      "step": 1622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.791308699641338e-05,
+      "loss": 0.4914,
+      "step": 1623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7911418800567188e-05,
+      "loss": 0.5182,
+      "step": 1624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7909750604720998e-05,
+      "loss": 0.5188,
+      "step": 1625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.7908082408874804e-05,
+      "loss": 0.5025,
+      "step": 1626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.790641421302861e-05,
+      "loss": 0.5173,
+      "step": 1627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.790474601718242e-05,
+      "loss": 0.5084,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7903077821336226e-05,
+      "loss": 0.4844,
+      "step": 1629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7901409625490032e-05,
+      "loss": 0.4785,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7899741429643842e-05,
+      "loss": 0.5031,
+      "step": 1631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.789807323379765e-05,
+      "loss": 0.4854,
+      "step": 1632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7896405037951458e-05,
+      "loss": 0.5217,
+      "step": 1633
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.501,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7893068646259073e-05,
+      "loss": 0.4916,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.789140045041288e-05,
+      "loss": 0.5018,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7889732254566686e-05,
+      "loss": 0.5268,
+      "step": 1637
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7888064058720495e-05,
+      "loss": 0.5062,
+      "step": 1638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7886395862874305e-05,
+      "loss": 0.5402,
+      "step": 1639
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.788472766702811e-05,
+      "loss": 0.5296,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7883059471181917e-05,
+      "loss": 0.5051,
+      "step": 1641
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7881391275335727e-05,
+      "loss": 0.5164,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7879723079489533e-05,
+      "loss": 0.4981,
+      "step": 1643
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7878054883643343e-05,
+      "loss": 0.4813,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.787638668779715e-05,
+      "loss": 0.5179,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.787471849195096e-05,
+      "loss": 0.5001,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7873050296104765e-05,
+      "loss": 0.5356,
+      "step": 1647
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.787138210025857e-05,
+      "loss": 0.5174,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.786971390441238e-05,
+      "loss": 0.492,
+      "step": 1649
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7868045708566187e-05,
+      "loss": 0.503,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7866377512719996e-05,
+      "loss": 0.5052,
+      "step": 1651
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7864709316873802e-05,
+      "loss": 0.5032,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7863041121027612e-05,
+      "loss": 0.4889,
+      "step": 1653
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7861372925181418e-05,
+      "loss": 0.5426,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7859704729335224e-05,
+      "loss": 0.4861,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7858036533489034e-05,
+      "loss": 0.507,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.785636833764284e-05,
+      "loss": 0.5147,
+      "step": 1657
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.785470014179665e-05,
+      "loss": 0.4905,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7853031945950456e-05,
+      "loss": 0.4764,
+      "step": 1659
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7851363750104265e-05,
+      "loss": 0.5336,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.784969555425807e-05,
+      "loss": 0.4926,
+      "step": 1661
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7848027358411878e-05,
+      "loss": 0.4693,
+      "step": 1662
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7846359162565687e-05,
+      "loss": 0.5235,
+      "step": 1663
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7844690966719494e-05,
+      "loss": 0.4701,
+      "step": 1664
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7843022770873303e-05,
+      "loss": 0.498,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.784135457502711e-05,
+      "loss": 0.4987,
+      "step": 1666
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.783968637918092e-05,
+      "loss": 0.5457,
+      "step": 1667
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.7838018183334725e-05,
+      "loss": 0.4942,
+      "step": 1668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.783634998748853e-05,
+      "loss": 0.5005,
+      "step": 1669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.783468179164234e-05,
+      "loss": 0.5125,
+      "step": 1670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7833013595796147e-05,
+      "loss": 0.4972,
+      "step": 1671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7831345399949957e-05,
+      "loss": 0.4798,
+      "step": 1672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7829677204103763e-05,
+      "loss": 0.4754,
+      "step": 1673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7828009008257572e-05,
+      "loss": 0.5084,
+      "step": 1674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.782634081241138e-05,
+      "loss": 0.4996,
+      "step": 1675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7824672616565185e-05,
+      "loss": 0.5228,
+      "step": 1676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7823004420718994e-05,
+      "loss": 0.5113,
+      "step": 1677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.78213362248728e-05,
+      "loss": 0.5244,
+      "step": 1678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.781966802902661e-05,
+      "loss": 0.4962,
+      "step": 1679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7817999833180416e-05,
+      "loss": 0.5099,
+      "step": 1680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7816331637334226e-05,
+      "loss": 0.5522,
+      "step": 1681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7814663441488032e-05,
+      "loss": 0.5228,
+      "step": 1682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.781299524564184e-05,
+      "loss": 0.4916,
+      "step": 1683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7811327049795648e-05,
+      "loss": 0.4933,
+      "step": 1684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7809658853949454e-05,
+      "loss": 0.5217,
+      "step": 1685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7807990658103264e-05,
+      "loss": 0.5163,
+      "step": 1686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.780632246225707e-05,
+      "loss": 0.4554,
+      "step": 1687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.780465426641088e-05,
+      "loss": 0.4942,
+      "step": 1688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7802986070564686e-05,
+      "loss": 0.4762,
+      "step": 1689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7801317874718492e-05,
+      "loss": 0.5198,
+      "step": 1690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.77996496788723e-05,
+      "loss": 0.5147,
+      "step": 1691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7797981483026108e-05,
+      "loss": 0.475,
+      "step": 1692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7796313287179917e-05,
+      "loss": 0.5152,
+      "step": 1693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7794645091333723e-05,
+      "loss": 0.4521,
+      "step": 1694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7792976895487533e-05,
+      "loss": 0.5394,
+      "step": 1695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.779130869964134e-05,
+      "loss": 0.5568,
+      "step": 1696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7789640503795145e-05,
+      "loss": 0.4731,
+      "step": 1697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7787972307948955e-05,
+      "loss": 0.5161,
+      "step": 1698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.778630411210276e-05,
+      "loss": 0.4787,
+      "step": 1699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.778463591625657e-05,
+      "loss": 0.5204,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7782967720410377e-05,
+      "loss": 0.4829,
+      "step": 1701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7781299524564186e-05,
+      "loss": 0.4662,
+      "step": 1702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7779631328717993e-05,
+      "loss": 0.4818,
+      "step": 1703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.77779631328718e-05,
+      "loss": 0.5373,
+      "step": 1704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.777629493702561e-05,
+      "loss": 0.5279,
+      "step": 1705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7774626741179415e-05,
+      "loss": 0.506,
+      "step": 1706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7772958545333224e-05,
+      "loss": 0.5068,
+      "step": 1707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.777129034948703e-05,
+      "loss": 0.4927,
+      "step": 1708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.776962215364084e-05,
+      "loss": 0.5239,
+      "step": 1709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7767953957794646e-05,
+      "loss": 0.5163,
+      "step": 1710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7766285761948456e-05,
+      "loss": 0.5475,
+      "step": 1711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7764617566102262e-05,
+      "loss": 0.5216,
+      "step": 1712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7762949370256068e-05,
+      "loss": 0.4624,
+      "step": 1713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7761281174409878e-05,
+      "loss": 0.5053,
+      "step": 1714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7759612978563684e-05,
+      "loss": 0.4962,
+      "step": 1715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7757944782717494e-05,
+      "loss": 0.552,
+      "step": 1716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.77562765868713e-05,
+      "loss": 0.5012,
+      "step": 1717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.775460839102511e-05,
+      "loss": 0.5163,
+      "step": 1718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7752940195178915e-05,
+      "loss": 0.4758,
+      "step": 1719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.775127199933272e-05,
+      "loss": 0.5049,
+      "step": 1720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.774960380348653e-05,
+      "loss": 0.5078,
+      "step": 1721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7747935607640337e-05,
+      "loss": 0.486,
+      "step": 1722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7746267411794147e-05,
+      "loss": 0.4983,
+      "step": 1723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7744599215947953e-05,
+      "loss": 0.5061,
+      "step": 1724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7742931020101763e-05,
+      "loss": 0.5309,
+      "step": 1725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.774126282425557e-05,
+      "loss": 0.5173,
+      "step": 1726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7739594628409375e-05,
+      "loss": 0.5566,
+      "step": 1727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7737926432563185e-05,
+      "loss": 0.4978,
+      "step": 1728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.773625823671699e-05,
+      "loss": 0.5007,
+      "step": 1729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.77345900408708e-05,
+      "loss": 0.5278,
+      "step": 1730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7732921845024607e-05,
+      "loss": 0.5463,
+      "step": 1731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7731253649178416e-05,
+      "loss": 0.5518,
+      "step": 1732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7729585453332223e-05,
+      "loss": 0.4761,
+      "step": 1733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.772791725748603e-05,
+      "loss": 0.4895,
+      "step": 1734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7726249061639838e-05,
+      "loss": 0.5136,
+      "step": 1735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7724580865793644e-05,
+      "loss": 0.5061,
+      "step": 1736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7722912669947454e-05,
+      "loss": 0.507,
+      "step": 1737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.772124447410126e-05,
+      "loss": 0.4866,
+      "step": 1738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.771957627825507e-05,
+      "loss": 0.5151,
+      "step": 1739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7717908082408876e-05,
+      "loss": 0.4862,
+      "step": 1740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7716239886562686e-05,
+      "loss": 0.5273,
+      "step": 1741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7714571690716492e-05,
+      "loss": 0.4409,
+      "step": 1742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7712903494870298e-05,
+      "loss": 0.5301,
+      "step": 1743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7711235299024108e-05,
+      "loss": 0.5198,
+      "step": 1744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7709567103177914e-05,
+      "loss": 0.4924,
+      "step": 1745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7707898907331723e-05,
+      "loss": 0.4984,
+      "step": 1746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.770623071148553e-05,
+      "loss": 0.5422,
+      "step": 1747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.770456251563934e-05,
+      "loss": 0.5477,
+      "step": 1748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7702894319793145e-05,
+      "loss": 0.5284,
+      "step": 1749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.770122612394695e-05,
+      "loss": 0.5103,
+      "step": 1750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.769955792810076e-05,
+      "loss": 0.5398,
+      "step": 1751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7697889732254567e-05,
+      "loss": 0.5243,
+      "step": 1752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7696221536408377e-05,
+      "loss": 0.5238,
+      "step": 1753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7694553340562183e-05,
+      "loss": 0.4974,
+      "step": 1754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7692885144715993e-05,
+      "loss": 0.4975,
+      "step": 1755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.76912169488698e-05,
+      "loss": 0.5098,
+      "step": 1756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7689548753023605e-05,
+      "loss": 0.4681,
+      "step": 1757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7687880557177415e-05,
+      "loss": 0.4797,
+      "step": 1758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.768621236133122e-05,
+      "loss": 0.4859,
+      "step": 1759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.768454416548503e-05,
+      "loss": 0.4958,
+      "step": 1760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7682875969638837e-05,
+      "loss": 0.4682,
+      "step": 1761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7681207773792646e-05,
+      "loss": 0.5016,
+      "step": 1762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7679539577946452e-05,
+      "loss": 0.5133,
+      "step": 1763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.767787138210026e-05,
+      "loss": 0.5302,
+      "step": 1764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7676203186254068e-05,
+      "loss": 0.4863,
+      "step": 1765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7674534990407874e-05,
+      "loss": 0.5438,
+      "step": 1766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7672866794561684e-05,
+      "loss": 0.5317,
+      "step": 1767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.767119859871549e-05,
+      "loss": 0.509,
+      "step": 1768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.76695304028693e-05,
+      "loss": 0.5283,
+      "step": 1769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7667862207023106e-05,
+      "loss": 0.5302,
+      "step": 1770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7666194011176912e-05,
+      "loss": 0.4896,
+      "step": 1771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.766452581533072e-05,
+      "loss": 0.4858,
+      "step": 1772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7662857619484528e-05,
+      "loss": 0.5088,
+      "step": 1773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7661189423638337e-05,
+      "loss": 0.5009,
+      "step": 1774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7659521227792144e-05,
+      "loss": 0.5119,
+      "step": 1775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7657853031945953e-05,
+      "loss": 0.5269,
+      "step": 1776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.765618483609976e-05,
+      "loss": 0.5183,
+      "step": 1777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7654516640253566e-05,
+      "loss": 0.5297,
+      "step": 1778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7652848444407375e-05,
+      "loss": 0.4478,
+      "step": 1779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.765118024856118e-05,
+      "loss": 0.4753,
+      "step": 1780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.764951205271499e-05,
+      "loss": 0.5088,
+      "step": 1781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7647843856868797e-05,
+      "loss": 0.5357,
+      "step": 1782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7646175661022607e-05,
+      "loss": 0.5113,
+      "step": 1783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7644507465176413e-05,
+      "loss": 0.4757,
+      "step": 1784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7642839269330222e-05,
+      "loss": 0.5116,
+      "step": 1785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.764117107348403e-05,
+      "loss": 0.5365,
+      "step": 1786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7639502877637835e-05,
+      "loss": 0.5067,
+      "step": 1787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7637834681791644e-05,
+      "loss": 0.5346,
+      "step": 1788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.763616648594545e-05,
+      "loss": 0.4796,
+      "step": 1789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.763449829009926e-05,
+      "loss": 0.5138,
+      "step": 1790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7632830094253066e-05,
+      "loss": 0.4955,
+      "step": 1791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7631161898406876e-05,
+      "loss": 0.4864,
+      "step": 1792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7629493702560682e-05,
+      "loss": 0.4883,
+      "step": 1793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.762782550671449e-05,
+      "loss": 0.4726,
+      "step": 1794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7626157310868298e-05,
+      "loss": 0.4807,
+      "step": 1795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7624489115022104e-05,
+      "loss": 0.5143,
+      "step": 1796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7622820919175914e-05,
+      "loss": 0.5244,
+      "step": 1797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.762115272332972e-05,
+      "loss": 0.5618,
+      "step": 1798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.761948452748353e-05,
+      "loss": 0.5105,
+      "step": 1799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7617816331637336e-05,
+      "loss": 0.5123,
+      "step": 1800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7616148135791142e-05,
+      "loss": 0.5162,
+      "step": 1801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.761447993994495e-05,
+      "loss": 0.5124,
+      "step": 1802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7612811744098758e-05,
+      "loss": 0.5233,
+      "step": 1803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7611143548252567e-05,
+      "loss": 0.4655,
+      "step": 1804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7609475352406373e-05,
+      "loss": 0.5144,
+      "step": 1805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7607807156560183e-05,
+      "loss": 0.5043,
+      "step": 1806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.760613896071399e-05,
+      "loss": 0.479,
+      "step": 1807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7604470764867795e-05,
+      "loss": 0.5192,
+      "step": 1808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7602802569021605e-05,
+      "loss": 0.4989,
+      "step": 1809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.760113437317541e-05,
+      "loss": 0.4758,
+      "step": 1810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.759946617732922e-05,
+      "loss": 0.4771,
+      "step": 1811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7597797981483027e-05,
+      "loss": 0.5233,
+      "step": 1812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7596129785636837e-05,
+      "loss": 0.4953,
+      "step": 1813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7594461589790643e-05,
+      "loss": 0.5576,
+      "step": 1814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.759279339394445e-05,
+      "loss": 0.4933,
+      "step": 1815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.759112519809826e-05,
+      "loss": 0.4856,
+      "step": 1816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7589457002252065e-05,
+      "loss": 0.4876,
+      "step": 1817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7587788806405874e-05,
+      "loss": 0.505,
+      "step": 1818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.758612061055968e-05,
+      "loss": 0.4921,
+      "step": 1819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.758445241471349e-05,
+      "loss": 0.5012,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7582784218867296e-05,
+      "loss": 0.5236,
+      "step": 1821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7581116023021102e-05,
+      "loss": 0.4799,
+      "step": 1822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7579447827174912e-05,
+      "loss": 0.5281,
+      "step": 1823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7577779631328718e-05,
+      "loss": 0.5227,
+      "step": 1824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7576111435482528e-05,
+      "loss": 0.5177,
+      "step": 1825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7574443239636334e-05,
+      "loss": 0.5015,
+      "step": 1826
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7572775043790144e-05,
+      "loss": 0.5071,
+      "step": 1827
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.757110684794395e-05,
+      "loss": 0.5301,
+      "step": 1828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7569438652097756e-05,
+      "loss": 0.502,
+      "step": 1829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7567770456251566e-05,
+      "loss": 0.4981,
+      "step": 1830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7566102260405372e-05,
+      "loss": 0.5163,
+      "step": 1831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.756443406455918e-05,
+      "loss": 0.5161,
+      "step": 1832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7562765868712988e-05,
+      "loss": 0.5641,
+      "step": 1833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7561097672866797e-05,
+      "loss": 0.4956,
+      "step": 1834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7559429477020603e-05,
+      "loss": 0.5087,
+      "step": 1835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.755776128117441e-05,
+      "loss": 0.5122,
+      "step": 1836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.755609308532822e-05,
+      "loss": 0.5079,
+      "step": 1837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7554424889482025e-05,
+      "loss": 0.52,
+      "step": 1838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7552756693635835e-05,
+      "loss": 0.4762,
+      "step": 1839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.755108849778964e-05,
+      "loss": 0.4972,
+      "step": 1840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.754942030194345e-05,
+      "loss": 0.5006,
+      "step": 1841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7547752106097257e-05,
+      "loss": 0.5167,
+      "step": 1842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7546083910251063e-05,
+      "loss": 0.5075,
+      "step": 1843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7544415714404873e-05,
+      "loss": 0.5201,
+      "step": 1844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.754274751855868e-05,
+      "loss": 0.5276,
+      "step": 1845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.754107932271249e-05,
+      "loss": 0.4578,
+      "step": 1846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7539411126866295e-05,
+      "loss": 0.5329,
+      "step": 1847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7537742931020104e-05,
+      "loss": 0.4942,
+      "step": 1848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.753607473517391e-05,
+      "loss": 0.5047,
+      "step": 1849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7534406539327717e-05,
+      "loss": 0.53,
+      "step": 1850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7532738343481526e-05,
+      "loss": 0.4903,
+      "step": 1851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7531070147635336e-05,
+      "loss": 0.532,
+      "step": 1852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7529401951789142e-05,
+      "loss": 0.5318,
+      "step": 1853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7527733755942948e-05,
+      "loss": 0.4978,
+      "step": 1854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7526065560096758e-05,
+      "loss": 0.5151,
+      "step": 1855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7524397364250564e-05,
+      "loss": 0.5074,
+      "step": 1856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7522729168404373e-05,
+      "loss": 0.4898,
+      "step": 1857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.752106097255818e-05,
+      "loss": 0.5073,
+      "step": 1858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.751939277671199e-05,
+      "loss": 0.5196,
+      "step": 1859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7517724580865795e-05,
+      "loss": 0.4927,
+      "step": 1860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.75160563850196e-05,
+      "loss": 0.4807,
+      "step": 1861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.751438818917341e-05,
+      "loss": 0.5158,
+      "step": 1862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7512719993327217e-05,
+      "loss": 0.5288,
+      "step": 1863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7511051797481027e-05,
+      "loss": 0.5484,
+      "step": 1864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7509383601634833e-05,
+      "loss": 0.5079,
+      "step": 1865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7507715405788643e-05,
+      "loss": 0.5146,
+      "step": 1866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.750604720994245e-05,
+      "loss": 0.5204,
+      "step": 1867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7504379014096255e-05,
+      "loss": 0.4968,
+      "step": 1868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7502710818250065e-05,
+      "loss": 0.4906,
+      "step": 1869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.750104262240387e-05,
+      "loss": 0.5001,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.749937442655768e-05,
+      "loss": 0.5006,
+      "step": 1871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7497706230711487e-05,
+      "loss": 0.5273,
+      "step": 1872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7496038034865296e-05,
+      "loss": 0.4905,
+      "step": 1873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.7494369839019102e-05,
+      "loss": 0.5282,
+      "step": 1874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.749270164317291e-05,
+      "loss": 0.4471,
+      "step": 1875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7491033447326718e-05,
+      "loss": 0.5151,
+      "step": 1876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7489365251480524e-05,
+      "loss": 0.4696,
+      "step": 1877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7487697055634334e-05,
+      "loss": 0.5116,
+      "step": 1878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748602885978814e-05,
+      "loss": 0.4898,
+      "step": 1879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.748436066394195e-05,
+      "loss": 0.5237,
+      "step": 1880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7482692468095756e-05,
+      "loss": 0.5046,
+      "step": 1881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7481024272249562e-05,
+      "loss": 0.4786,
+      "step": 1882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7479356076403372e-05,
+      "loss": 0.4777,
+      "step": 1883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7477687880557178e-05,
+      "loss": 0.5168,
+      "step": 1884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7476019684710987e-05,
+      "loss": 0.458,
+      "step": 1885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7474351488864794e-05,
+      "loss": 0.538,
+      "step": 1886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7472683293018603e-05,
+      "loss": 0.4898,
+      "step": 1887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.747101509717241e-05,
+      "loss": 0.5118,
+      "step": 1888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7469346901326216e-05,
+      "loss": 0.5115,
+      "step": 1889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7467678705480025e-05,
+      "loss": 0.4939,
+      "step": 1890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.746601050963383e-05,
+      "loss": 0.4794,
+      "step": 1891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.746434231378764e-05,
+      "loss": 0.4839,
+      "step": 1892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7462674117941447e-05,
+      "loss": 0.4944,
+      "step": 1893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7461005922095257e-05,
+      "loss": 0.5311,
+      "step": 1894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7459337726249063e-05,
+      "loss": 0.5149,
+      "step": 1895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745766953040287e-05,
+      "loss": 0.5054,
+      "step": 1896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.745600133455668e-05,
+      "loss": 0.5281,
+      "step": 1897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7454333138710485e-05,
+      "loss": 0.4996,
+      "step": 1898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7452664942864295e-05,
+      "loss": 0.5315,
+      "step": 1899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.74509967470181e-05,
+      "loss": 0.5341,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.744932855117191e-05,
+      "loss": 0.5055,
+      "step": 1901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7447660355325716e-05,
+      "loss": 0.4909,
+      "step": 1902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7445992159479523e-05,
+      "loss": 0.4877,
+      "step": 1903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7444323963633332e-05,
+      "loss": 0.4711,
+      "step": 1904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.744265576778714e-05,
+      "loss": 0.5107,
+      "step": 1905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7440987571940948e-05,
+      "loss": 0.517,
+      "step": 1906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7439319376094754e-05,
+      "loss": 0.5284,
+      "step": 1907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7437651180248564e-05,
+      "loss": 0.5076,
+      "step": 1908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.743598298440237e-05,
+      "loss": 0.4667,
+      "step": 1909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7434314788556176e-05,
+      "loss": 0.4513,
+      "step": 1910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7432646592709986e-05,
+      "loss": 0.539,
+      "step": 1911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7430978396863792e-05,
+      "loss": 0.5292,
+      "step": 1912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.74293102010176e-05,
+      "loss": 0.5102,
+      "step": 1913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7427642005171408e-05,
+      "loss": 0.5491,
+      "step": 1914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7425973809325217e-05,
+      "loss": 0.5577,
+      "step": 1915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.7424305613479024e-05,
+      "loss": 0.4932,
+      "step": 1916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.742263741763283e-05,
+      "loss": 0.4948,
+      "step": 1917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.742096922178664e-05,
+      "loss": 0.5061,
+      "step": 1918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7419301025940445e-05,
+      "loss": 0.5118,
+      "step": 1919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7417632830094255e-05,
+      "loss": 0.5295,
+      "step": 1920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.741596463424806e-05,
+      "loss": 0.4985,
+      "step": 1921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.741429643840187e-05,
+      "loss": 0.5029,
+      "step": 1922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7412628242555677e-05,
+      "loss": 0.538,
+      "step": 1923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7410960046709483e-05,
+      "loss": 0.5223,
+      "step": 1924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7409291850863293e-05,
+      "loss": 0.5172,
+      "step": 1925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7407623655017102e-05,
+      "loss": 0.5714,
+      "step": 1926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.740595545917091e-05,
+      "loss": 0.5178,
+      "step": 1927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7404287263324715e-05,
+      "loss": 0.5046,
+      "step": 1928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7402619067478524e-05,
+      "loss": 0.4751,
+      "step": 1929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.740095087163233e-05,
+      "loss": 0.4841,
+      "step": 1930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7399282675786137e-05,
+      "loss": 0.502,
+      "step": 1931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7397614479939946e-05,
+      "loss": 0.5468,
+      "step": 1932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7395946284093756e-05,
+      "loss": 0.4735,
+      "step": 1933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7394278088247562e-05,
+      "loss": 0.458,
+      "step": 1934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7392609892401368e-05,
+      "loss": 0.5122,
+      "step": 1935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7390941696555178e-05,
+      "loss": 0.4723,
+      "step": 1936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7389273500708984e-05,
+      "loss": 0.5112,
+      "step": 1937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.738760530486279e-05,
+      "loss": 0.5196,
+      "step": 1938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.73859371090166e-05,
+      "loss": 0.4871,
+      "step": 1939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.738426891317041e-05,
+      "loss": 0.5007,
+      "step": 1940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7382600717324216e-05,
+      "loss": 0.4657,
+      "step": 1941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7380932521478022e-05,
+      "loss": 0.5044,
+      "step": 1942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.737926432563183e-05,
+      "loss": 0.5217,
+      "step": 1943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7377596129785638e-05,
+      "loss": 0.5044,
+      "step": 1944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7375927933939444e-05,
+      "loss": 0.5289,
+      "step": 1945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7374259738093253e-05,
+      "loss": 0.4956,
+      "step": 1946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7372591542247063e-05,
+      "loss": 0.5143,
+      "step": 1947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.737092334640087e-05,
+      "loss": 0.4874,
+      "step": 1948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7369255150554675e-05,
+      "loss": 0.5134,
+      "step": 1949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7367586954708485e-05,
+      "loss": 0.4705,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.736591875886229e-05,
+      "loss": 0.4776,
+      "step": 1951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7364250563016097e-05,
+      "loss": 0.4969,
+      "step": 1952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7362582367169907e-05,
+      "loss": 0.5337,
+      "step": 1953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7360914171323716e-05,
+      "loss": 0.4911,
+      "step": 1954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.7359245975477523e-05,
+      "loss": 0.4553,
+      "step": 1955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.735757777963133e-05,
+      "loss": 0.4743,
+      "step": 1956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.735590958378514e-05,
+      "loss": 0.511,
+      "step": 1957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7354241387938945e-05,
+      "loss": 0.4864,
+      "step": 1958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.735257319209275e-05,
+      "loss": 0.5273,
+      "step": 1959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.735090499624656e-05,
+      "loss": 0.497,
+      "step": 1960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.734923680040037e-05,
+      "loss": 0.5233,
+      "step": 1961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7347568604554176e-05,
+      "loss": 0.5076,
+      "step": 1962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7345900408707982e-05,
+      "loss": 0.4568,
+      "step": 1963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7344232212861792e-05,
+      "loss": 0.5194,
+      "step": 1964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7342564017015598e-05,
+      "loss": 0.482,
+      "step": 1965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7340895821169404e-05,
+      "loss": 0.5083,
+      "step": 1966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7339227625323214e-05,
+      "loss": 0.5144,
+      "step": 1967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7337559429477023e-05,
+      "loss": 0.5117,
+      "step": 1968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.733589123363083e-05,
+      "loss": 0.5045,
+      "step": 1969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7334223037784636e-05,
+      "loss": 0.5678,
+      "step": 1970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7332554841938445e-05,
+      "loss": 0.4722,
+      "step": 1971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.733088664609225e-05,
+      "loss": 0.5001,
+      "step": 1972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7329218450246058e-05,
+      "loss": 0.5055,
+      "step": 1973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7327550254399867e-05,
+      "loss": 0.4924,
+      "step": 1974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7325882058553677e-05,
+      "loss": 0.507,
+      "step": 1975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7324213862707483e-05,
+      "loss": 0.5136,
+      "step": 1976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.732254566686129e-05,
+      "loss": 0.4992,
+      "step": 1977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.73208774710151e-05,
+      "loss": 0.5241,
+      "step": 1978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7319209275168905e-05,
+      "loss": 0.4932,
+      "step": 1979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7317541079322715e-05,
+      "loss": 0.4813,
+      "step": 1980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.731587288347652e-05,
+      "loss": 0.4741,
+      "step": 1981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.731420468763033e-05,
+      "loss": 0.4901,
+      "step": 1982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7312536491784137e-05,
+      "loss": 0.5255,
+      "step": 1983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7310868295937943e-05,
+      "loss": 0.5034,
+      "step": 1984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7309200100091752e-05,
+      "loss": 0.4855,
+      "step": 1985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.730753190424556e-05,
+      "loss": 0.4792,
+      "step": 1986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7305863708399368e-05,
+      "loss": 0.4681,
+      "step": 1987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7304195512553174e-05,
+      "loss": 0.5178,
+      "step": 1988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7302527316706984e-05,
+      "loss": 0.5186,
+      "step": 1989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.730085912086079e-05,
+      "loss": 0.494,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7299190925014596e-05,
+      "loss": 0.5305,
+      "step": 1991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7297522729168406e-05,
+      "loss": 0.5124,
+      "step": 1992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7295854533322212e-05,
+      "loss": 0.5457,
+      "step": 1993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7294186337476022e-05,
+      "loss": 0.5204,
+      "step": 1994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7292518141629828e-05,
+      "loss": 0.4978,
+      "step": 1995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7290849945783638e-05,
+      "loss": 0.4934,
+      "step": 1996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7289181749937444e-05,
+      "loss": 0.5155,
+      "step": 1997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.728751355409125e-05,
+      "loss": 0.509,
+      "step": 1998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.728584535824506e-05,
+      "loss": 0.4891,
+      "step": 1999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.728417716239887e-05,
+      "loss": 0.4894,
+      "step": 2000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7282508966552675e-05,
+      "loss": 0.5024,
+      "step": 2001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.728084077070648e-05,
+      "loss": 0.5166,
+      "step": 2002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.727917257486029e-05,
+      "loss": 0.5308,
+      "step": 2003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7277504379014097e-05,
+      "loss": 0.4989,
+      "step": 2004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7275836183167903e-05,
+      "loss": 0.4994,
+      "step": 2005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7274167987321713e-05,
+      "loss": 0.509,
+      "step": 2006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7272499791475523e-05,
+      "loss": 0.5088,
+      "step": 2007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.727083159562933e-05,
+      "loss": 0.491,
+      "step": 2008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7269163399783135e-05,
+      "loss": 0.5064,
+      "step": 2009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7267495203936945e-05,
+      "loss": 0.4504,
+      "step": 2010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.726582700809075e-05,
+      "loss": 0.4838,
+      "step": 2011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7264158812244557e-05,
+      "loss": 0.4883,
+      "step": 2012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7262490616398367e-05,
+      "loss": 0.4778,
+      "step": 2013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7260822420552176e-05,
+      "loss": 0.4934,
+      "step": 2014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7259154224705982e-05,
+      "loss": 0.5177,
+      "step": 2015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.725748602885979e-05,
+      "loss": 0.4916,
+      "step": 2016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7255817833013598e-05,
+      "loss": 0.5167,
+      "step": 2017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7254149637167404e-05,
+      "loss": 0.4722,
+      "step": 2018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.725248144132121e-05,
+      "loss": 0.5279,
+      "step": 2019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.725081324547502e-05,
+      "loss": 0.481,
+      "step": 2020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.724914504962883e-05,
+      "loss": 0.5276,
+      "step": 2021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7247476853782636e-05,
+      "loss": 0.504,
+      "step": 2022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7245808657936442e-05,
+      "loss": 0.5384,
+      "step": 2023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.724414046209025e-05,
+      "loss": 0.4936,
+      "step": 2024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7242472266244058e-05,
+      "loss": 0.5358,
+      "step": 2025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7240804070397864e-05,
+      "loss": 0.5028,
+      "step": 2026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7239135874551674e-05,
+      "loss": 0.4654,
+      "step": 2027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7237467678705483e-05,
+      "loss": 0.5102,
+      "step": 2028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.723579948285929e-05,
+      "loss": 0.4717,
+      "step": 2029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7234131287013096e-05,
+      "loss": 0.4732,
+      "step": 2030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7232463091166905e-05,
+      "loss": 0.504,
+      "step": 2031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.723079489532071e-05,
+      "loss": 0.5209,
+      "step": 2032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7229126699474517e-05,
+      "loss": 0.5442,
+      "step": 2033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7227458503628327e-05,
+      "loss": 0.5012,
+      "step": 2034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7225790307782137e-05,
+      "loss": 0.4881,
+      "step": 2035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7224122111935943e-05,
+      "loss": 0.5029,
+      "step": 2036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.722245391608975e-05,
+      "loss": 0.4992,
+      "step": 2037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.722078572024356e-05,
+      "loss": 0.5106,
+      "step": 2038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7219117524397365e-05,
+      "loss": 0.5474,
+      "step": 2039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.721744932855117e-05,
+      "loss": 0.5002,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.721578113270498e-05,
+      "loss": 0.535,
+      "step": 2041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.721411293685879e-05,
+      "loss": 0.5372,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7212444741012596e-05,
+      "loss": 0.4992,
+      "step": 2043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7210776545166403e-05,
+      "loss": 0.521,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7209108349320212e-05,
+      "loss": 0.5095,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.720744015347402e-05,
+      "loss": 0.5346,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7205771957627825e-05,
+      "loss": 0.4973,
+      "step": 2047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7204103761781634e-05,
+      "loss": 0.4792,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7202435565935444e-05,
+      "loss": 0.5541,
+      "step": 2049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.720076737008925e-05,
+      "loss": 0.5084,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7199099174243056e-05,
+      "loss": 0.4974,
+      "step": 2051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7197430978396866e-05,
+      "loss": 0.509,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7195762782550672e-05,
+      "loss": 0.511,
+      "step": 2053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7194094586704478e-05,
+      "loss": 0.5174,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7192426390858288e-05,
+      "loss": 0.4744,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7190758195012097e-05,
+      "loss": 0.5033,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7189089999165903e-05,
+      "loss": 0.4923,
+      "step": 2057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.718742180331971e-05,
+      "loss": 0.4798,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.718575360747352e-05,
+      "loss": 0.5251,
+      "step": 2059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7184085411627325e-05,
+      "loss": 0.5015,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.718241721578113e-05,
+      "loss": 0.532,
+      "step": 2061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.718074901993494e-05,
+      "loss": 0.5042,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.717908082408875e-05,
+      "loss": 0.5204,
+      "step": 2063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7177412628242557e-05,
+      "loss": 0.5148,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7175744432396363e-05,
+      "loss": 0.4992,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7174076236550173e-05,
+      "loss": 0.5022,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7172408040703982e-05,
+      "loss": 0.5115,
+      "step": 2067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7170739844857785e-05,
+      "loss": 0.4692,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7169071649011595e-05,
+      "loss": 0.4851,
+      "step": 2069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7167403453165404e-05,
+      "loss": 0.4945,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.716573525731921e-05,
+      "loss": 0.4951,
+      "step": 2071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7164067061473017e-05,
+      "loss": 0.5054,
+      "step": 2072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7162398865626826e-05,
+      "loss": 0.4964,
+      "step": 2073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7160730669780636e-05,
+      "loss": 0.4745,
+      "step": 2074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.715906247393444e-05,
+      "loss": 0.5071,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7157394278088248e-05,
+      "loss": 0.5042,
+      "step": 2076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7155726082242058e-05,
+      "loss": 0.5257,
+      "step": 2077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7154057886395864e-05,
+      "loss": 0.4589,
+      "step": 2078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.715238969054967e-05,
+      "loss": 0.4887,
+      "step": 2079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.715072149470348e-05,
+      "loss": 0.4908,
+      "step": 2080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.714905329885729e-05,
+      "loss": 0.5014,
+      "step": 2081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7147385103011095e-05,
+      "loss": 0.4833,
+      "step": 2082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.71457169071649e-05,
+      "loss": 0.502,
+      "step": 2083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.714404871131871e-05,
+      "loss": 0.5406,
+      "step": 2084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7142380515472517e-05,
+      "loss": 0.5161,
+      "step": 2085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7140712319626324e-05,
+      "loss": 0.501,
+      "step": 2086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7139044123780133e-05,
+      "loss": 0.5242,
+      "step": 2087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7137375927933943e-05,
+      "loss": 0.5028,
+      "step": 2088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.713570773208775e-05,
+      "loss": 0.4756,
+      "step": 2089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7134039536241555e-05,
+      "loss": 0.4814,
+      "step": 2090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7132371340395365e-05,
+      "loss": 0.5125,
+      "step": 2091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.713070314454917e-05,
+      "loss": 0.5082,
+      "step": 2092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7129034948702977e-05,
+      "loss": 0.5259,
+      "step": 2093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7127366752856787e-05,
+      "loss": 0.5115,
+      "step": 2094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7125698557010596e-05,
+      "loss": 0.4655,
+      "step": 2095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7124030361164403e-05,
+      "loss": 0.4975,
+      "step": 2096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.712236216531821e-05,
+      "loss": 0.466,
+      "step": 2097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7120693969472018e-05,
+      "loss": 0.4827,
+      "step": 2098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7119025773625824e-05,
+      "loss": 0.5249,
+      "step": 2099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711735757777963e-05,
+      "loss": 0.479,
+      "step": 2100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711568938193344e-05,
+      "loss": 0.5364,
+      "step": 2101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.711402118608725e-05,
+      "loss": 0.4941,
+      "step": 2102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7112352990241056e-05,
+      "loss": 0.5016,
+      "step": 2103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7110684794394862e-05,
+      "loss": 0.5066,
+      "step": 2104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7109016598548672e-05,
+      "loss": 0.4871,
+      "step": 2105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7107348402702478e-05,
+      "loss": 0.4714,
+      "step": 2106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7105680206856284e-05,
+      "loss": 0.475,
+      "step": 2107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7104012011010094e-05,
+      "loss": 0.5546,
+      "step": 2108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7102343815163903e-05,
+      "loss": 0.5109,
+      "step": 2109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.710067561931771e-05,
+      "loss": 0.4917,
+      "step": 2110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7099007423471516e-05,
+      "loss": 0.5095,
+      "step": 2111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7097339227625325e-05,
+      "loss": 0.5111,
+      "step": 2112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.709567103177913e-05,
+      "loss": 0.5105,
+      "step": 2113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7094002835932938e-05,
+      "loss": 0.4929,
+      "step": 2114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7092334640086747e-05,
+      "loss": 0.5317,
+      "step": 2115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7090666444240557e-05,
+      "loss": 0.5128,
+      "step": 2116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7088998248394363e-05,
+      "loss": 0.4715,
+      "step": 2117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708733005254817e-05,
+      "loss": 0.4926,
+      "step": 2118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708566185670198e-05,
+      "loss": 0.4946,
+      "step": 2119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7083993660855785e-05,
+      "loss": 0.5324,
+      "step": 2120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.708232546500959e-05,
+      "loss": 0.5111,
+      "step": 2121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.70806572691634e-05,
+      "loss": 0.4903,
+      "step": 2122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.707898907331721e-05,
+      "loss": 0.5064,
+      "step": 2123
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7077320877471017e-05,
+      "loss": 0.4617,
+      "step": 2124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7075652681624823e-05,
+      "loss": 0.4977,
+      "step": 2125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7073984485778632e-05,
+      "loss": 0.5035,
+      "step": 2126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.707231628993244e-05,
+      "loss": 0.489,
+      "step": 2127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7070648094086245e-05,
+      "loss": 0.4993,
+      "step": 2128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7068979898240054e-05,
+      "loss": 0.5165,
+      "step": 2129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7067311702393864e-05,
+      "loss": 0.5116,
+      "step": 2130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.706564350654767e-05,
+      "loss": 0.5469,
+      "step": 2131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7063975310701476e-05,
+      "loss": 0.5236,
+      "step": 2132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7062307114855286e-05,
+      "loss": 0.5136,
+      "step": 2133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7060638919009092e-05,
+      "loss": 0.4899,
+      "step": 2134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7058970723162898e-05,
+      "loss": 0.4786,
+      "step": 2135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7057302527316708e-05,
+      "loss": 0.5646,
+      "step": 2136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7055634331470517e-05,
+      "loss": 0.523,
+      "step": 2137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7053966135624324e-05,
+      "loss": 0.4746,
+      "step": 2138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.705229793977813e-05,
+      "loss": 0.5006,
+      "step": 2139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.705062974393194e-05,
+      "loss": 0.5139,
+      "step": 2140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.704896154808575e-05,
+      "loss": 0.5065,
+      "step": 2141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7047293352239552e-05,
+      "loss": 0.5445,
+      "step": 2142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.704562515639336e-05,
+      "loss": 0.5173,
+      "step": 2143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.704395696054717e-05,
+      "loss": 0.4927,
+      "step": 2144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7042288764700977e-05,
+      "loss": 0.5434,
+      "step": 2145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7040620568854783e-05,
+      "loss": 0.5055,
+      "step": 2146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7038952373008593e-05,
+      "loss": 0.4896,
+      "step": 2147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7037284177162402e-05,
+      "loss": 0.4881,
+      "step": 2148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7035615981316205e-05,
+      "loss": 0.4781,
+      "step": 2149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7033947785470015e-05,
+      "loss": 0.4845,
+      "step": 2150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7032279589623824e-05,
+      "loss": 0.4983,
+      "step": 2151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.703061139377763e-05,
+      "loss": 0.5258,
+      "step": 2152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7028943197931437e-05,
+      "loss": 0.5325,
+      "step": 2153
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7027275002085246e-05,
+      "loss": 0.5026,
+      "step": 2154
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7025606806239056e-05,
+      "loss": 0.5161,
+      "step": 2155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7023938610392862e-05,
+      "loss": 0.5422,
+      "step": 2156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.702227041454667e-05,
+      "loss": 0.5425,
+      "step": 2157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7020602218700478e-05,
+      "loss": 0.5262,
+      "step": 2158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7018934022854284e-05,
+      "loss": 0.5123,
+      "step": 2159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.701726582700809e-05,
+      "loss": 0.5566,
+      "step": 2160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.70155976311619e-05,
+      "loss": 0.5396,
+      "step": 2161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.701392943531571e-05,
+      "loss": 0.4941,
+      "step": 2162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7012261239469516e-05,
+      "loss": 0.4682,
+      "step": 2163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7010593043623322e-05,
+      "loss": 0.5211,
+      "step": 2164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.700892484777713e-05,
+      "loss": 0.5223,
+      "step": 2165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7007256651930938e-05,
+      "loss": 0.495,
+      "step": 2166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7005588456084744e-05,
+      "loss": 0.5195,
+      "step": 2167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7003920260238553e-05,
+      "loss": 0.4827,
+      "step": 2168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7002252064392363e-05,
+      "loss": 0.4923,
+      "step": 2169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.700058386854617e-05,
+      "loss": 0.4882,
+      "step": 2170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6998915672699975e-05,
+      "loss": 0.5254,
+      "step": 2171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6997247476853785e-05,
+      "loss": 0.5312,
+      "step": 2172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.699557928100759e-05,
+      "loss": 0.5015,
+      "step": 2173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6993911085161397e-05,
+      "loss": 0.516,
+      "step": 2174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6992242889315207e-05,
+      "loss": 0.5361,
+      "step": 2175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6990574693469017e-05,
+      "loss": 0.49,
+      "step": 2176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6988906497622823e-05,
+      "loss": 0.49,
+      "step": 2177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698723830177663e-05,
+      "loss": 0.495,
+      "step": 2178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698557010593044e-05,
+      "loss": 0.4714,
+      "step": 2179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6983901910084245e-05,
+      "loss": 0.4751,
+      "step": 2180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698223371423805e-05,
+      "loss": 0.499,
+      "step": 2181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.698056551839186e-05,
+      "loss": 0.4901,
+      "step": 2182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.697889732254567e-05,
+      "loss": 0.4914,
+      "step": 2183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6977229126699476e-05,
+      "loss": 0.4605,
+      "step": 2184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6975560930853282e-05,
+      "loss": 0.4645,
+      "step": 2185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6973892735007092e-05,
+      "loss": 0.5074,
+      "step": 2186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6972224539160898e-05,
+      "loss": 0.5349,
+      "step": 2187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6970556343314704e-05,
+      "loss": 0.5079,
+      "step": 2188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6968888147468514e-05,
+      "loss": 0.4864,
+      "step": 2189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6967219951622324e-05,
+      "loss": 0.4662,
+      "step": 2190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.696555175577613e-05,
+      "loss": 0.5552,
+      "step": 2191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6963883559929936e-05,
+      "loss": 0.5313,
+      "step": 2192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6962215364083746e-05,
+      "loss": 0.5279,
+      "step": 2193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6960547168237552e-05,
+      "loss": 0.5173,
+      "step": 2194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6958878972391358e-05,
+      "loss": 0.4873,
+      "step": 2195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6957210776545168e-05,
+      "loss": 0.4903,
+      "step": 2196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6955542580698977e-05,
+      "loss": 0.4924,
+      "step": 2197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6953874384852783e-05,
+      "loss": 0.5042,
+      "step": 2198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.695220618900659e-05,
+      "loss": 0.4802,
+      "step": 2199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.69505379931604e-05,
+      "loss": 0.5256,
+      "step": 2200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.6948869797314205e-05,
+      "loss": 0.5146,
+      "step": 2201
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.694720160146801e-05,
+      "loss": 0.5125,
+      "step": 2202
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.694553340562182e-05,
+      "loss": 0.5251,
+      "step": 2203
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.694386520977563e-05,
+      "loss": 0.4903,
+      "step": 2204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6942197013929437e-05,
+      "loss": 0.4751,
+      "step": 2205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6940528818083243e-05,
+      "loss": 0.5167,
+      "step": 2206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6938860622237053e-05,
+      "loss": 0.5073,
+      "step": 2207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6937192426390862e-05,
+      "loss": 0.548,
+      "step": 2208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6935524230544665e-05,
+      "loss": 0.4723,
+      "step": 2209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6933856034698475e-05,
+      "loss": 0.4759,
+      "step": 2210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6932187838852284e-05,
+      "loss": 0.5267,
+      "step": 2211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.693051964300609e-05,
+      "loss": 0.4962,
+      "step": 2212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6928851447159897e-05,
+      "loss": 0.5185,
+      "step": 2213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6927183251313706e-05,
+      "loss": 0.495,
+      "step": 2214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6925515055467516e-05,
+      "loss": 0.525,
+      "step": 2215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.692384685962132e-05,
+      "loss": 0.5082,
+      "step": 2216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6922178663775128e-05,
+      "loss": 0.4964,
+      "step": 2217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6920510467928938e-05,
+      "loss": 0.5061,
+      "step": 2218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6918842272082744e-05,
+      "loss": 0.5359,
+      "step": 2219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691717407623655e-05,
+      "loss": 0.5055,
+      "step": 2220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691550588039036e-05,
+      "loss": 0.4826,
+      "step": 2221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691383768454417e-05,
+      "loss": 0.4667,
+      "step": 2222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6912169488697975e-05,
+      "loss": 0.5267,
+      "step": 2223
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.691050129285178e-05,
+      "loss": 0.5073,
+      "step": 2224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.690883309700559e-05,
+      "loss": 0.477,
+      "step": 2225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6907164901159397e-05,
+      "loss": 0.5056,
+      "step": 2226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6905496705313204e-05,
+      "loss": 0.5064,
+      "step": 2227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6903828509467013e-05,
+      "loss": 0.4898,
+      "step": 2228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6902160313620823e-05,
+      "loss": 0.495,
+      "step": 2229
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.690049211777463e-05,
+      "loss": 0.4669,
+      "step": 2230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6898823921928435e-05,
+      "loss": 0.5129,
+      "step": 2231
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6897155726082245e-05,
+      "loss": 0.491,
+      "step": 2232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.689548753023605e-05,
+      "loss": 0.4891,
+      "step": 2233
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6893819334389857e-05,
+      "loss": 0.4972,
+      "step": 2234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6892151138543667e-05,
+      "loss": 0.4753,
+      "step": 2235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6890482942697476e-05,
+      "loss": 0.5059,
+      "step": 2236
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6888814746851282e-05,
+      "loss": 0.5078,
+      "step": 2237
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.688714655100509e-05,
+      "loss": 0.5317,
+      "step": 2238
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6885478355158898e-05,
+      "loss": 0.5203,
+      "step": 2239
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6883810159312704e-05,
+      "loss": 0.5367,
+      "step": 2240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.688214196346651e-05,
+      "loss": 0.5083,
+      "step": 2241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.688047376762032e-05,
+      "loss": 0.4954,
+      "step": 2242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.687880557177413e-05,
+      "loss": 0.5166,
+      "step": 2243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6877137375927936e-05,
+      "loss": 0.5105,
+      "step": 2244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.6875469180081742e-05,
+      "loss": 0.5176,
+      "step": 2245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6873800984235552e-05,
+      "loss": 0.5283,
+      "step": 2246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6872132788389358e-05,
+      "loss": 0.5209,
+      "step": 2247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6870464592543164e-05,
+      "loss": 0.5143,
+      "step": 2248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6868796396696974e-05,
+      "loss": 0.4501,
+      "step": 2249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6867128200850783e-05,
+      "loss": 0.4749,
+      "step": 2250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686546000500459e-05,
+      "loss": 0.4989,
+      "step": 2251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6863791809158396e-05,
+      "loss": 0.4554,
+      "step": 2252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6862123613312205e-05,
+      "loss": 0.5418,
+      "step": 2253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.686045541746601e-05,
+      "loss": 0.45,
+      "step": 2254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6858787221619818e-05,
+      "loss": 0.4931,
+      "step": 2255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6857119025773627e-05,
+      "loss": 0.5348,
+      "step": 2256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6855450829927437e-05,
+      "loss": 0.5209,
+      "step": 2257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6853782634081243e-05,
+      "loss": 0.5082,
+      "step": 2258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.685211443823505e-05,
+      "loss": 0.5218,
+      "step": 2259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.685044624238886e-05,
+      "loss": 0.4701,
+      "step": 2260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6848778046542665e-05,
+      "loss": 0.5042,
+      "step": 2261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.684710985069647e-05,
+      "loss": 0.5257,
+      "step": 2262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.684544165485028e-05,
+      "loss": 0.4832,
+      "step": 2263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.684377345900409e-05,
+      "loss": 0.5458,
+      "step": 2264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.5498,
+      "step": 2265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6840437067311703e-05,
+      "loss": 0.5228,
+      "step": 2266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6838768871465512e-05,
+      "loss": 0.4883,
+      "step": 2267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.683710067561932e-05,
+      "loss": 0.4998,
+      "step": 2268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6835432479773125e-05,
+      "loss": 0.4969,
+      "step": 2269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6833764283926934e-05,
+      "loss": 0.5226,
+      "step": 2270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6832096088080744e-05,
+      "loss": 0.5096,
+      "step": 2271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.683042789223455e-05,
+      "loss": 0.5051,
+      "step": 2272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6828759696388356e-05,
+      "loss": 0.5201,
+      "step": 2273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6827091500542166e-05,
+      "loss": 0.51,
+      "step": 2274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6825423304695972e-05,
+      "loss": 0.4406,
+      "step": 2275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6823755108849778e-05,
+      "loss": 0.4965,
+      "step": 2276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6822086913003588e-05,
+      "loss": 0.4996,
+      "step": 2277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6820418717157397e-05,
+      "loss": 0.5032,
+      "step": 2278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6818750521311204e-05,
+      "loss": 0.4874,
+      "step": 2279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681708232546501e-05,
+      "loss": 0.4638,
+      "step": 2280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681541412961882e-05,
+      "loss": 0.4945,
+      "step": 2281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681374593377263e-05,
+      "loss": 0.4949,
+      "step": 2282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681207773792643e-05,
+      "loss": 0.4833,
+      "step": 2283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.681040954208024e-05,
+      "loss": 0.4685,
+      "step": 2284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.680874134623405e-05,
+      "loss": 0.5021,
+      "step": 2285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.6807073150387857e-05,
+      "loss": 0.4783,
+      "step": 2286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6805404954541663e-05,
+      "loss": 0.5205,
+      "step": 2287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6803736758695473e-05,
+      "loss": 0.4859,
+      "step": 2288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6802068562849282e-05,
+      "loss": 0.53,
+      "step": 2289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6800400367003085e-05,
+      "loss": 0.503,
+      "step": 2290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6798732171156895e-05,
+      "loss": 0.4641,
+      "step": 2291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6797063975310704e-05,
+      "loss": 0.4812,
+      "step": 2292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.679539577946451e-05,
+      "loss": 0.5049,
+      "step": 2293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6793727583618317e-05,
+      "loss": 0.5061,
+      "step": 2294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6792059387772126e-05,
+      "loss": 0.53,
+      "step": 2295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6790391191925936e-05,
+      "loss": 0.4729,
+      "step": 2296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6788722996079742e-05,
+      "loss": 0.5592,
+      "step": 2297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6787054800233548e-05,
+      "loss": 0.5317,
+      "step": 2298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6785386604387358e-05,
+      "loss": 0.5091,
+      "step": 2299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6783718408541164e-05,
+      "loss": 0.5351,
+      "step": 2300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.678205021269497e-05,
+      "loss": 0.4926,
+      "step": 2301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.678038201684878e-05,
+      "loss": 0.4819,
+      "step": 2302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677871382100259e-05,
+      "loss": 0.5151,
+      "step": 2303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6777045625156396e-05,
+      "loss": 0.5149,
+      "step": 2304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6775377429310202e-05,
+      "loss": 0.5203,
+      "step": 2305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.677370923346401e-05,
+      "loss": 0.5009,
+      "step": 2306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6772041037617818e-05,
+      "loss": 0.4808,
+      "step": 2307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6770372841771624e-05,
+      "loss": 0.5045,
+      "step": 2308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6768704645925433e-05,
+      "loss": 0.5031,
+      "step": 2309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6767036450079243e-05,
+      "loss": 0.4886,
+      "step": 2310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.676536825423305e-05,
+      "loss": 0.475,
+      "step": 2311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6763700058386855e-05,
+      "loss": 0.4985,
+      "step": 2312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6762031862540665e-05,
+      "loss": 0.4997,
+      "step": 2313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.676036366669447e-05,
+      "loss": 0.4572,
+      "step": 2314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6758695470848277e-05,
+      "loss": 0.5172,
+      "step": 2315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6757027275002087e-05,
+      "loss": 0.5246,
+      "step": 2316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6755359079155896e-05,
+      "loss": 0.5272,
+      "step": 2317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6753690883309703e-05,
+      "loss": 0.4892,
+      "step": 2318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.675202268746351e-05,
+      "loss": 0.5186,
+      "step": 2319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.675035449161732e-05,
+      "loss": 0.5159,
+      "step": 2320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6748686295771125e-05,
+      "loss": 0.4974,
+      "step": 2321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.674701809992493e-05,
+      "loss": 0.5083,
+      "step": 2322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.674534990407874e-05,
+      "loss": 0.4498,
+      "step": 2323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.674368170823255e-05,
+      "loss": 0.5006,
+      "step": 2324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6742013512386356e-05,
+      "loss": 0.5133,
+      "step": 2325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6740345316540162e-05,
+      "loss": 0.4755,
+      "step": 2326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6738677120693972e-05,
+      "loss": 0.5319,
+      "step": 2327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.6737008924847778e-05,
+      "loss": 0.5,
+      "step": 2328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6735340729001584e-05,
+      "loss": 0.5247,
+      "step": 2329
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6733672533155394e-05,
+      "loss": 0.5272,
+      "step": 2330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6732004337309203e-05,
+      "loss": 0.5303,
+      "step": 2331
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.673033614146301e-05,
+      "loss": 0.5084,
+      "step": 2332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6728667945616816e-05,
+      "loss": 0.4719,
+      "step": 2333
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6726999749770625e-05,
+      "loss": 0.4767,
+      "step": 2334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.672533155392443e-05,
+      "loss": 0.4894,
+      "step": 2335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6723663358078238e-05,
+      "loss": 0.5135,
+      "step": 2336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6721995162232047e-05,
+      "loss": 0.4914,
+      "step": 2337
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6720326966385857e-05,
+      "loss": 0.5046,
+      "step": 2338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6718658770539663e-05,
+      "loss": 0.5643,
+      "step": 2339
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671699057469347e-05,
+      "loss": 0.4918,
+      "step": 2340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671532237884728e-05,
+      "loss": 0.504,
+      "step": 2341
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6713654183001085e-05,
+      "loss": 0.5259,
+      "step": 2342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.671198598715489e-05,
+      "loss": 0.5131,
+      "step": 2343
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.67103177913087e-05,
+      "loss": 0.5374,
+      "step": 2344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.670864959546251e-05,
+      "loss": 0.5239,
+      "step": 2345
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6706981399616317e-05,
+      "loss": 0.5272,
+      "step": 2346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6705313203770123e-05,
+      "loss": 0.4922,
+      "step": 2347
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6703645007923932e-05,
+      "loss": 0.543,
+      "step": 2348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6701976812077742e-05,
+      "loss": 0.5045,
+      "step": 2349
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6700308616231545e-05,
+      "loss": 0.5312,
+      "step": 2350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6698640420385354e-05,
+      "loss": 0.4896,
+      "step": 2351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6696972224539164e-05,
+      "loss": 0.5284,
+      "step": 2352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.669530402869297e-05,
+      "loss": 0.5013,
+      "step": 2353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6693635832846776e-05,
+      "loss": 0.4801,
+      "step": 2354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6691967637000586e-05,
+      "loss": 0.5096,
+      "step": 2355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6690299441154396e-05,
+      "loss": 0.4983,
+      "step": 2356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.66886312453082e-05,
+      "loss": 0.5424,
+      "step": 2357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6686963049462008e-05,
+      "loss": 0.4708,
+      "step": 2358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6685294853615818e-05,
+      "loss": 0.4867,
+      "step": 2359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6683626657769624e-05,
+      "loss": 0.5218,
+      "step": 2360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.668195846192343e-05,
+      "loss": 0.4824,
+      "step": 2361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.668029026607724e-05,
+      "loss": 0.474,
+      "step": 2362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667862207023105e-05,
+      "loss": 0.5114,
+      "step": 2363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6676953874384855e-05,
+      "loss": 0.5095,
+      "step": 2364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667528567853866e-05,
+      "loss": 0.4968,
+      "step": 2365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.667361748269247e-05,
+      "loss": 0.5105,
+      "step": 2366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6671949286846277e-05,
+      "loss": 0.5344,
+      "step": 2367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6670281091000083e-05,
+      "loss": 0.5013,
+      "step": 2368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.6668612895153893e-05,
+      "loss": 0.4869,
+      "step": 2369
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6666944699307703e-05,
+      "loss": 0.4795,
+      "step": 2370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666527650346151e-05,
+      "loss": 0.519,
+      "step": 2371
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6663608307615315e-05,
+      "loss": 0.5144,
+      "step": 2372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6661940111769125e-05,
+      "loss": 0.4716,
+      "step": 2373
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.666027191592293e-05,
+      "loss": 0.5106,
+      "step": 2374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6658603720076737e-05,
+      "loss": 0.4655,
+      "step": 2375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6656935524230547e-05,
+      "loss": 0.5025,
+      "step": 2376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6655267328384356e-05,
+      "loss": 0.5335,
+      "step": 2377
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6653599132538162e-05,
+      "loss": 0.4444,
+      "step": 2378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.665193093669197e-05,
+      "loss": 0.4984,
+      "step": 2379
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6650262740845778e-05,
+      "loss": 0.4972,
+      "step": 2380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6648594544999584e-05,
+      "loss": 0.4815,
+      "step": 2381
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664692634915339e-05,
+      "loss": 0.5325,
+      "step": 2382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66452581533072e-05,
+      "loss": 0.5037,
+      "step": 2383
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.664358995746101e-05,
+      "loss": 0.4628,
+      "step": 2384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6641921761614816e-05,
+      "loss": 0.4907,
+      "step": 2385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6640253565768622e-05,
+      "loss": 0.5261,
+      "step": 2386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663858536992243e-05,
+      "loss": 0.4786,
+      "step": 2387
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6636917174076238e-05,
+      "loss": 0.5003,
+      "step": 2388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6635248978230044e-05,
+      "loss": 0.505,
+      "step": 2389
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6633580782383854e-05,
+      "loss": 0.503,
+      "step": 2390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6631912586537663e-05,
+      "loss": 0.4868,
+      "step": 2391
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.663024439069147e-05,
+      "loss": 0.5014,
+      "step": 2392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6628576194845276e-05,
+      "loss": 0.5495,
+      "step": 2393
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6626907998999085e-05,
+      "loss": 0.5144,
+      "step": 2394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.662523980315289e-05,
+      "loss": 0.5255,
+      "step": 2395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6623571607306697e-05,
+      "loss": 0.4653,
+      "step": 2396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6621903411460507e-05,
+      "loss": 0.4954,
+      "step": 2397
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6620235215614317e-05,
+      "loss": 0.5003,
+      "step": 2398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6618567019768123e-05,
+      "loss": 0.4843,
+      "step": 2399
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661689882392193e-05,
+      "loss": 0.4933,
+      "step": 2400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661523062807574e-05,
+      "loss": 0.4979,
+      "step": 2401
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6613562432229545e-05,
+      "loss": 0.492,
+      "step": 2402
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661189423638335e-05,
+      "loss": 0.5036,
+      "step": 2403
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.661022604053716e-05,
+      "loss": 0.5251,
+      "step": 2404
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.660855784469097e-05,
+      "loss": 0.4451,
+      "step": 2405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6606889648844776e-05,
+      "loss": 0.4949,
+      "step": 2406
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6605221452998583e-05,
+      "loss": 0.5121,
+      "step": 2407
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6603553257152392e-05,
+      "loss": 0.4882,
+      "step": 2408
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.66018850613062e-05,
+      "loss": 0.4962,
+      "step": 2409
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.6600216865460005e-05,
+      "loss": 0.4903,
+      "step": 2410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6598548669613814e-05,
+      "loss": 0.5084,
+      "step": 2411
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6596880473767624e-05,
+      "loss": 0.4999,
+      "step": 2412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.659521227792143e-05,
+      "loss": 0.4993,
+      "step": 2413
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6593544082075236e-05,
+      "loss": 0.5034,
+      "step": 2414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6591875886229046e-05,
+      "loss": 0.5048,
+      "step": 2415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6590207690382852e-05,
+      "loss": 0.4915,
+      "step": 2416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6588539494536658e-05,
+      "loss": 0.5227,
+      "step": 2417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6586871298690468e-05,
+      "loss": 0.4584,
+      "step": 2418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6585203102844277e-05,
+      "loss": 0.4927,
+      "step": 2419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6583534906998083e-05,
+      "loss": 0.4687,
+      "step": 2420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.658186671115189e-05,
+      "loss": 0.5601,
+      "step": 2421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.65801985153057e-05,
+      "loss": 0.5391,
+      "step": 2422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6578530319459505e-05,
+      "loss": 0.5619,
+      "step": 2423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657686212361331e-05,
+      "loss": 0.5091,
+      "step": 2424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657519392776712e-05,
+      "loss": 0.4872,
+      "step": 2425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.657352573192093e-05,
+      "loss": 0.5143,
+      "step": 2426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6571857536074737e-05,
+      "loss": 0.5155,
+      "step": 2427
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6570189340228543e-05,
+      "loss": 0.5306,
+      "step": 2428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6568521144382353e-05,
+      "loss": 0.4873,
+      "step": 2429
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656685294853616e-05,
+      "loss": 0.5042,
+      "step": 2430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6565184752689965e-05,
+      "loss": 0.5185,
+      "step": 2431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6563516556843775e-05,
+      "loss": 0.5059,
+      "step": 2432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6561848360997584e-05,
+      "loss": 0.4905,
+      "step": 2433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.656018016515139e-05,
+      "loss": 0.5001,
+      "step": 2434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6558511969305197e-05,
+      "loss": 0.5304,
+      "step": 2435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6556843773459006e-05,
+      "loss": 0.4517,
+      "step": 2436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6555175577612812e-05,
+      "loss": 0.4739,
+      "step": 2437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6553507381766622e-05,
+      "loss": 0.4896,
+      "step": 2438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6551839185920428e-05,
+      "loss": 0.508,
+      "step": 2439
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6550170990074238e-05,
+      "loss": 0.551,
+      "step": 2440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6548502794228044e-05,
+      "loss": 0.5306,
+      "step": 2441
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654683459838185e-05,
+      "loss": 0.4509,
+      "step": 2442
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654516640253566e-05,
+      "loss": 0.4954,
+      "step": 2443
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6543498206689466e-05,
+      "loss": 0.4776,
+      "step": 2444
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6541830010843275e-05,
+      "loss": 0.4706,
+      "step": 2445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.654016181499708e-05,
+      "loss": 0.5125,
+      "step": 2446
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.653849361915089e-05,
+      "loss": 0.4839,
+      "step": 2447
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6536825423304697e-05,
+      "loss": 0.4768,
+      "step": 2448
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6535157227458504e-05,
+      "loss": 0.4432,
+      "step": 2449
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6533489031612313e-05,
+      "loss": 0.4579,
+      "step": 2450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.6531820835766123e-05,
+      "loss": 0.4493,
+      "step": 2451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.653015263991993e-05,
+      "loss": 0.4766,
+      "step": 2452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6528484444073735e-05,
+      "loss": 0.4553,
+      "step": 2453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6526816248227545e-05,
+      "loss": 0.497,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.652514805238135e-05,
+      "loss": 0.4495,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6523479856535157e-05,
+      "loss": 0.4413,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6521811660688967e-05,
+      "loss": 0.4897,
+      "step": 2457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6520143464842776e-05,
+      "loss": 0.5286,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6518475268996583e-05,
+      "loss": 0.5013,
+      "step": 2459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.651680707315039e-05,
+      "loss": 0.5346,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6515138877304198e-05,
+      "loss": 0.5335,
+      "step": 2461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6513470681458004e-05,
+      "loss": 0.5037,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.651180248561181e-05,
+      "loss": 0.5007,
+      "step": 2463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.651013428976562e-05,
+      "loss": 0.4695,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.650846609391943e-05,
+      "loss": 0.4967,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6506797898073236e-05,
+      "loss": 0.4775,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6505129702227042e-05,
+      "loss": 0.5208,
+      "step": 2467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6503461506380852e-05,
+      "loss": 0.4857,
+      "step": 2468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6501793310534658e-05,
+      "loss": 0.5027,
+      "step": 2469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6500125114688464e-05,
+      "loss": 0.5254,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6498456918842274e-05,
+      "loss": 0.4992,
+      "step": 2471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6496788722996083e-05,
+      "loss": 0.5011,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.649512052714989e-05,
+      "loss": 0.4716,
+      "step": 2473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6493452331303696e-05,
+      "loss": 0.4918,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6491784135457505e-05,
+      "loss": 0.4854,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.649011593961131e-05,
+      "loss": 0.5045,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6488447743765118e-05,
+      "loss": 0.5284,
+      "step": 2477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6486779547918927e-05,
+      "loss": 0.4916,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6485111352072737e-05,
+      "loss": 0.5112,
+      "step": 2479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6483443156226543e-05,
+      "loss": 0.5154,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.648177496038035e-05,
+      "loss": 0.4921,
+      "step": 2481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.648010676453416e-05,
+      "loss": 0.4827,
+      "step": 2482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6478438568687965e-05,
+      "loss": 0.4255,
+      "step": 2483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647677037284177e-05,
+      "loss": 0.5149,
+      "step": 2484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647510217699558e-05,
+      "loss": 0.5172,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.647343398114939e-05,
+      "loss": 0.4859,
+      "step": 2486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6471765785303197e-05,
+      "loss": 0.5137,
+      "step": 2487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6470097589457003e-05,
+      "loss": 0.4893,
+      "step": 2488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6468429393610812e-05,
+      "loss": 0.5188,
+      "step": 2489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.646676119776462e-05,
+      "loss": 0.4873,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6465093001918425e-05,
+      "loss": 0.5145,
+      "step": 2491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6463424806072234e-05,
+      "loss": 0.5482,
+      "step": 2492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6461756610226044e-05,
+      "loss": 0.505,
+      "step": 2493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.646008841437985e-05,
+      "loss": 0.502,
+      "step": 2494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6458420218533656e-05,
+      "loss": 0.4941,
+      "step": 2495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6456752022687466e-05,
+      "loss": 0.4506,
+      "step": 2496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6455083826841272e-05,
+      "loss": 0.4895,
+      "step": 2497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6453415630995078e-05,
+      "loss": 0.52,
+      "step": 2498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6451747435148888e-05,
+      "loss": 0.5588,
+      "step": 2499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6450079239302697e-05,
+      "loss": 0.4823,
+      "step": 2500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6448411043456504e-05,
+      "loss": 0.4901,
+      "step": 2501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644674284761031e-05,
+      "loss": 0.5027,
+      "step": 2502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644507465176412e-05,
+      "loss": 0.468,
+      "step": 2503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6443406455917926e-05,
+      "loss": 0.5176,
+      "step": 2504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6441738260071732e-05,
+      "loss": 0.5045,
+      "step": 2505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.644007006422554e-05,
+      "loss": 0.5011,
+      "step": 2506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.643840186837935e-05,
+      "loss": 0.4977,
+      "step": 2507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6436733672533157e-05,
+      "loss": 0.5276,
+      "step": 2508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6435065476686963e-05,
+      "loss": 0.4681,
+      "step": 2509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6433397280840773e-05,
+      "loss": 0.4871,
+      "step": 2510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.643172908499458e-05,
+      "loss": 0.4739,
+      "step": 2511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.643006088914839e-05,
+      "loss": 0.4628,
+      "step": 2512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6428392693302195e-05,
+      "loss": 0.528,
+      "step": 2513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6426724497456004e-05,
+      "loss": 0.4697,
+      "step": 2514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.642505630160981e-05,
+      "loss": 0.5375,
+      "step": 2515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6423388105763617e-05,
+      "loss": 0.5039,
+      "step": 2516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6421719909917426e-05,
+      "loss": 0.4976,
+      "step": 2517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6420051714071233e-05,
+      "loss": 0.5102,
+      "step": 2518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6418383518225042e-05,
+      "loss": 0.5117,
+      "step": 2519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641671532237885e-05,
+      "loss": 0.5092,
+      "step": 2520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6415047126532658e-05,
+      "loss": 0.5265,
+      "step": 2521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6413378930686464e-05,
+      "loss": 0.4887,
+      "step": 2522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641171073484027e-05,
+      "loss": 0.4941,
+      "step": 2523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641004253899408e-05,
+      "loss": 0.5339,
+      "step": 2524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6408374343147886e-05,
+      "loss": 0.5203,
+      "step": 2525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6406706147301696e-05,
+      "loss": 0.5167,
+      "step": 2526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6405037951455502e-05,
+      "loss": 0.4817,
+      "step": 2527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.640336975560931e-05,
+      "loss": 0.4723,
+      "step": 2528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6401701559763118e-05,
+      "loss": 0.5021,
+      "step": 2529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6400033363916924e-05,
+      "loss": 0.4708,
+      "step": 2530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6398365168070733e-05,
+      "loss": 0.4954,
+      "step": 2531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.639669697222454e-05,
+      "loss": 0.5504,
+      "step": 2532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.639502877637835e-05,
+      "loss": 0.4953,
+      "step": 2533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6393360580532155e-05,
+      "loss": 0.4901,
+      "step": 2534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6391692384685965e-05,
+      "loss": 0.51,
+      "step": 2535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.639002418883977e-05,
+      "loss": 0.4803,
+      "step": 2536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6388355992993577e-05,
+      "loss": 0.4482,
+      "step": 2537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6386687797147387e-05,
+      "loss": 0.4925,
+      "step": 2538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6385019601301193e-05,
+      "loss": 0.5293,
+      "step": 2539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6383351405455003e-05,
+      "loss": 0.5048,
+      "step": 2540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.638168320960881e-05,
+      "loss": 0.4971,
+      "step": 2541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.638001501376262e-05,
+      "loss": 0.5133,
+      "step": 2542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6378346817916425e-05,
+      "loss": 0.4883,
+      "step": 2543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637667862207023e-05,
+      "loss": 0.4678,
+      "step": 2544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.637501042622404e-05,
+      "loss": 0.5196,
+      "step": 2545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6373342230377847e-05,
+      "loss": 0.5071,
+      "step": 2546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6371674034531656e-05,
+      "loss": 0.4474,
+      "step": 2547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6370005838685462e-05,
+      "loss": 0.5392,
+      "step": 2548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6368337642839272e-05,
+      "loss": 0.4574,
+      "step": 2549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6366669446993078e-05,
+      "loss": 0.4858,
+      "step": 2550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6365001251146884e-05,
+      "loss": 0.4503,
+      "step": 2551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6363333055300694e-05,
+      "loss": 0.5147,
+      "step": 2552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.63616648594545e-05,
+      "loss": 0.4507,
+      "step": 2553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.635999666360831e-05,
+      "loss": 0.5144,
+      "step": 2554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6358328467762116e-05,
+      "loss": 0.5077,
+      "step": 2555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6356660271915926e-05,
+      "loss": 0.5126,
+      "step": 2556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6354992076069732e-05,
+      "loss": 0.5056,
+      "step": 2557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6353323880223538e-05,
+      "loss": 0.4581,
+      "step": 2558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6351655684377348e-05,
+      "loss": 0.4781,
+      "step": 2559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6349987488531154e-05,
+      "loss": 0.4748,
+      "step": 2560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6348319292684963e-05,
+      "loss": 0.4555,
+      "step": 2561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.634665109683877e-05,
+      "loss": 0.5514,
+      "step": 2562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.634498290099258e-05,
+      "loss": 0.4954,
+      "step": 2563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6343314705146385e-05,
+      "loss": 0.4871,
+      "step": 2564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.634164650930019e-05,
+      "loss": 0.4973,
+      "step": 2565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6339978313454e-05,
+      "loss": 0.5035,
+      "step": 2566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6338310117607807e-05,
+      "loss": 0.4946,
+      "step": 2567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6336641921761617e-05,
+      "loss": 0.5384,
+      "step": 2568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6334973725915423e-05,
+      "loss": 0.5386,
+      "step": 2569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6333305530069233e-05,
+      "loss": 0.4747,
+      "step": 2570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.633163733422304e-05,
+      "loss": 0.5278,
+      "step": 2571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6329969138376845e-05,
+      "loss": 0.4706,
+      "step": 2572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6328300942530655e-05,
+      "loss": 0.4984,
+      "step": 2573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6326632746684464e-05,
+      "loss": 0.5063,
+      "step": 2574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.632496455083827e-05,
+      "loss": 0.462,
+      "step": 2575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6323296354992077e-05,
+      "loss": 0.4941,
+      "step": 2576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6321628159145886e-05,
+      "loss": 0.5185,
+      "step": 2577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6319959963299692e-05,
+      "loss": 0.496,
+      "step": 2578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6318291767453502e-05,
+      "loss": 0.4925,
+      "step": 2579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6316623571607308e-05,
+      "loss": 0.5032,
+      "step": 2580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6314955375761118e-05,
+      "loss": 0.4908,
+      "step": 2581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6313287179914924e-05,
+      "loss": 0.4814,
+      "step": 2582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.631161898406873e-05,
+      "loss": 0.4757,
+      "step": 2583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.630995078822254e-05,
+      "loss": 0.5163,
+      "step": 2584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6308282592376346e-05,
+      "loss": 0.5032,
+      "step": 2585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6306614396530155e-05,
+      "loss": 0.4844,
+      "step": 2586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.630494620068396e-05,
+      "loss": 0.4981,
+      "step": 2587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.630327800483777e-05,
+      "loss": 0.5345,
+      "step": 2588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6301609808991577e-05,
+      "loss": 0.4959,
+      "step": 2589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6299941613145384e-05,
+      "loss": 0.5152,
+      "step": 2590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6298273417299193e-05,
+      "loss": 0.5276,
+      "step": 2591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6296605221453e-05,
+      "loss": 0.5125,
+      "step": 2592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.629493702560681e-05,
+      "loss": 0.5138,
+      "step": 2593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6293268829760615e-05,
+      "loss": 0.5051,
+      "step": 2594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6291600633914425e-05,
+      "loss": 0.5212,
+      "step": 2595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.628993243806823e-05,
+      "loss": 0.4647,
+      "step": 2596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6288264242222037e-05,
+      "loss": 0.5416,
+      "step": 2597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6286596046375847e-05,
+      "loss": 0.4869,
+      "step": 2598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6284927850529653e-05,
+      "loss": 0.5348,
+      "step": 2599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6283259654683462e-05,
+      "loss": 0.4513,
+      "step": 2600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.628159145883727e-05,
+      "loss": 0.4854,
+      "step": 2601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6279923262991078e-05,
+      "loss": 0.5014,
+      "step": 2602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6278255067144884e-05,
+      "loss": 0.5197,
+      "step": 2603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.627658687129869e-05,
+      "loss": 0.5227,
+      "step": 2604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.62749186754525e-05,
+      "loss": 0.4879,
+      "step": 2605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6273250479606306e-05,
+      "loss": 0.5184,
+      "step": 2606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6271582283760116e-05,
+      "loss": 0.5187,
+      "step": 2607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6269914087913922e-05,
+      "loss": 0.5119,
+      "step": 2608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.626824589206773e-05,
+      "loss": 0.5479,
+      "step": 2609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6266577696221538e-05,
+      "loss": 0.4798,
+      "step": 2610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6264909500375344e-05,
+      "loss": 0.4735,
+      "step": 2611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6263241304529154e-05,
+      "loss": 0.5174,
+      "step": 2612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.626157310868296e-05,
+      "loss": 0.4488,
+      "step": 2613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.625990491283677e-05,
+      "loss": 0.5186,
+      "step": 2614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6258236716990576e-05,
+      "loss": 0.4731,
+      "step": 2615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6256568521144385e-05,
+      "loss": 0.5176,
+      "step": 2616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.625490032529819e-05,
+      "loss": 0.4683,
+      "step": 2617
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6253232129451998e-05,
+      "loss": 0.4784,
+      "step": 2618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6251563933605807e-05,
+      "loss": 0.4768,
+      "step": 2619
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6249895737759613e-05,
+      "loss": 0.5174,
+      "step": 2620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6248227541913423e-05,
+      "loss": 0.5384,
+      "step": 2621
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624655934606723e-05,
+      "loss": 0.512,
+      "step": 2622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624489115022104e-05,
+      "loss": 0.5065,
+      "step": 2623
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6243222954374845e-05,
+      "loss": 0.453,
+      "step": 2624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.624155475852865e-05,
+      "loss": 0.4934,
+      "step": 2625
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.623988656268246e-05,
+      "loss": 0.4781,
+      "step": 2626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6238218366836267e-05,
+      "loss": 0.4838,
+      "step": 2627
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6236550170990076e-05,
+      "loss": 0.5076,
+      "step": 2628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6234881975143883e-05,
+      "loss": 0.484,
+      "step": 2629
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6233213779297692e-05,
+      "loss": 0.5117,
+      "step": 2630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.62315455834515e-05,
+      "loss": 0.5168,
+      "step": 2631
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6229877387605305e-05,
+      "loss": 0.5054,
+      "step": 2632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6228209191759114e-05,
+      "loss": 0.5099,
+      "step": 2633
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.622654099591292e-05,
+      "loss": 0.4961,
+      "step": 2634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.622487280006673e-05,
+      "loss": 0.5411,
+      "step": 2635
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6223204604220536e-05,
+      "loss": 0.5073,
+      "step": 2636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6221536408374346e-05,
+      "loss": 0.4894,
+      "step": 2637
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6219868212528152e-05,
+      "loss": 0.4517,
+      "step": 2638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6218200016681958e-05,
+      "loss": 0.5308,
+      "step": 2639
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6216531820835768e-05,
+      "loss": 0.505,
+      "step": 2640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6214863624989574e-05,
+      "loss": 0.5161,
+      "step": 2641
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6213195429143383e-05,
+      "loss": 0.5082,
+      "step": 2642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.621152723329719e-05,
+      "loss": 0.5049,
+      "step": 2643
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6209859037451e-05,
+      "loss": 0.4581,
+      "step": 2644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6208190841604805e-05,
+      "loss": 0.4847,
+      "step": 2645
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.620652264575861e-05,
+      "loss": 0.494,
+      "step": 2646
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.620485444991242e-05,
+      "loss": 0.485,
+      "step": 2647
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6203186254066227e-05,
+      "loss": 0.4847,
+      "step": 2648
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6201518058220037e-05,
+      "loss": 0.4653,
+      "step": 2649
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6199849862373843e-05,
+      "loss": 0.4968,
+      "step": 2650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6198181666527653e-05,
+      "loss": 0.5193,
+      "step": 2651
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.619651347068146e-05,
+      "loss": 0.4775,
+      "step": 2652
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.619484527483527e-05,
+      "loss": 0.4818,
+      "step": 2653
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6193177078989075e-05,
+      "loss": 0.4964,
+      "step": 2654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.619150888314288e-05,
+      "loss": 0.5097,
+      "step": 2655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.618984068729669e-05,
+      "loss": 0.5374,
+      "step": 2656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.6188172491450497e-05,
+      "loss": 0.4814,
+      "step": 2657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6186504295604306e-05,
+      "loss": 0.4965,
+      "step": 2658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6184836099758112e-05,
+      "loss": 0.4794,
+      "step": 2659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6183167903911922e-05,
+      "loss": 0.4959,
+      "step": 2660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6181499708065728e-05,
+      "loss": 0.5286,
+      "step": 2661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6179831512219534e-05,
+      "loss": 0.4986,
+      "step": 2662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6178163316373344e-05,
+      "loss": 0.4955,
+      "step": 2663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.617649512052715e-05,
+      "loss": 0.5072,
+      "step": 2664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.617482692468096e-05,
+      "loss": 0.4866,
+      "step": 2665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6173158728834766e-05,
+      "loss": 0.4597,
+      "step": 2666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6171490532988576e-05,
+      "loss": 0.5045,
+      "step": 2667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6169822337142382e-05,
+      "loss": 0.4488,
+      "step": 2668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6168154141296188e-05,
+      "loss": 0.4725,
+      "step": 2669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6166485945449998e-05,
+      "loss": 0.5453,
+      "step": 2670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6164817749603804e-05,
+      "loss": 0.53,
+      "step": 2671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6163149553757613e-05,
+      "loss": 0.4844,
+      "step": 2672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.616148135791142e-05,
+      "loss": 0.5232,
+      "step": 2673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.615981316206523e-05,
+      "loss": 0.4697,
+      "step": 2674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6158144966219035e-05,
+      "loss": 0.4984,
+      "step": 2675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.615647677037284e-05,
+      "loss": 0.5355,
+      "step": 2676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.615480857452665e-05,
+      "loss": 0.5241,
+      "step": 2677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6153140378680457e-05,
+      "loss": 0.4732,
+      "step": 2678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6151472182834267e-05,
+      "loss": 0.4702,
+      "step": 2679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6149803986988073e-05,
+      "loss": 0.4917,
+      "step": 2680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6148135791141883e-05,
+      "loss": 0.5156,
+      "step": 2681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.614646759529569e-05,
+      "loss": 0.5037,
+      "step": 2682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6144799399449495e-05,
+      "loss": 0.4977,
+      "step": 2683
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6143131203603305e-05,
+      "loss": 0.4834,
+      "step": 2684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.614146300775711e-05,
+      "loss": 0.4808,
+      "step": 2685
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.613979481191092e-05,
+      "loss": 0.4845,
+      "step": 2686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6138126616064727e-05,
+      "loss": 0.4636,
+      "step": 2687
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6136458420218536e-05,
+      "loss": 0.4804,
+      "step": 2688
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6134790224372342e-05,
+      "loss": 0.4719,
+      "step": 2689
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6133122028526152e-05,
+      "loss": 0.5183,
+      "step": 2690
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6131453832679958e-05,
+      "loss": 0.487,
+      "step": 2691
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6129785636833764e-05,
+      "loss": 0.4939,
+      "step": 2692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6128117440987574e-05,
+      "loss": 0.4833,
+      "step": 2693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.612644924514138e-05,
+      "loss": 0.5081,
+      "step": 2694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.612478104929519e-05,
+      "loss": 0.49,
+      "step": 2695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6123112853448996e-05,
+      "loss": 0.516,
+      "step": 2696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.6121444657602805e-05,
+      "loss": 0.5268,
+      "step": 2697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.611977646175661e-05,
+      "loss": 0.475,
+      "step": 2698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6118108265910418e-05,
+      "loss": 0.4812,
+      "step": 2699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6116440070064227e-05,
+      "loss": 0.4553,
+      "step": 2700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6114771874218034e-05,
+      "loss": 0.541,
+      "step": 2701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6113103678371843e-05,
+      "loss": 0.5453,
+      "step": 2702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.611143548252565e-05,
+      "loss": 0.5015,
+      "step": 2703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.610976728667946e-05,
+      "loss": 0.5071,
+      "step": 2704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6108099090833265e-05,
+      "loss": 0.4982,
+      "step": 2705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.610643089498707e-05,
+      "loss": 0.4611,
+      "step": 2706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.610476269914088e-05,
+      "loss": 0.4833,
+      "step": 2707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6103094503294687e-05,
+      "loss": 0.5122,
+      "step": 2708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6101426307448497e-05,
+      "loss": 0.4801,
+      "step": 2709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6099758111602303e-05,
+      "loss": 0.4724,
+      "step": 2710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6098089915756112e-05,
+      "loss": 0.497,
+      "step": 2711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.609642171990992e-05,
+      "loss": 0.496,
+      "step": 2712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6094753524063725e-05,
+      "loss": 0.4583,
+      "step": 2713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6093085328217534e-05,
+      "loss": 0.4514,
+      "step": 2714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.609141713237134e-05,
+      "loss": 0.4827,
+      "step": 2715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.608974893652515e-05,
+      "loss": 0.502,
+      "step": 2716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6088080740678956e-05,
+      "loss": 0.4844,
+      "step": 2717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6086412544832766e-05,
+      "loss": 0.4988,
+      "step": 2718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6084744348986572e-05,
+      "loss": 0.4814,
+      "step": 2719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6083076153140382e-05,
+      "loss": 0.4992,
+      "step": 2720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6081407957294188e-05,
+      "loss": 0.5194,
+      "step": 2721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6079739761447994e-05,
+      "loss": 0.4626,
+      "step": 2722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6078071565601804e-05,
+      "loss": 0.5084,
+      "step": 2723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.607640336975561e-05,
+      "loss": 0.4936,
+      "step": 2724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.607473517390942e-05,
+      "loss": 0.4926,
+      "step": 2725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6073066978063226e-05,
+      "loss": 0.482,
+      "step": 2726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6071398782217035e-05,
+      "loss": 0.4681,
+      "step": 2727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.606973058637084e-05,
+      "loss": 0.4872,
+      "step": 2728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6068062390524648e-05,
+      "loss": 0.4569,
+      "step": 2729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6066394194678457e-05,
+      "loss": 0.4758,
+      "step": 2730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6064725998832263e-05,
+      "loss": 0.4796,
+      "step": 2731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6063057802986073e-05,
+      "loss": 0.521,
+      "step": 2732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.606138960713988e-05,
+      "loss": 0.4693,
+      "step": 2733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.605972141129369e-05,
+      "loss": 0.5032,
+      "step": 2734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6058053215447495e-05,
+      "loss": 0.4982,
+      "step": 2735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.60563850196013e-05,
+      "loss": 0.4883,
+      "step": 2736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.605471682375511e-05,
+      "loss": 0.4987,
+      "step": 2737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6053048627908917e-05,
+      "loss": 0.5056,
+      "step": 2738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6051380432062727e-05,
+      "loss": 0.4738,
+      "step": 2739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.6049712236216533e-05,
+      "loss": 0.5147,
+      "step": 2740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6048044040370342e-05,
+      "loss": 0.5161,
+      "step": 2741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.604637584452415e-05,
+      "loss": 0.4785,
+      "step": 2742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6044707648677955e-05,
+      "loss": 0.4683,
+      "step": 2743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6043039452831764e-05,
+      "loss": 0.4719,
+      "step": 2744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.604137125698557e-05,
+      "loss": 0.4817,
+      "step": 2745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.603970306113938e-05,
+      "loss": 0.5365,
+      "step": 2746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6038034865293186e-05,
+      "loss": 0.479,
+      "step": 2747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6036366669446996e-05,
+      "loss": 0.479,
+      "step": 2748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6034698473600802e-05,
+      "loss": 0.4837,
+      "step": 2749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6033030277754608e-05,
+      "loss": 0.4754,
+      "step": 2750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6031362081908418e-05,
+      "loss": 0.5107,
+      "step": 2751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6029693886062224e-05,
+      "loss": 0.5237,
+      "step": 2752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6028025690216034e-05,
+      "loss": 0.4991,
+      "step": 2753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.602635749436984e-05,
+      "loss": 0.4862,
+      "step": 2754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.602468929852365e-05,
+      "loss": 0.5182,
+      "step": 2755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6023021102677456e-05,
+      "loss": 0.4713,
+      "step": 2756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6021352906831262e-05,
+      "loss": 0.5233,
+      "step": 2757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.601968471098507e-05,
+      "loss": 0.4963,
+      "step": 2758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6018016515138877e-05,
+      "loss": 0.4956,
+      "step": 2759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6016348319292687e-05,
+      "loss": 0.4766,
+      "step": 2760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6014680123446493e-05,
+      "loss": 0.4822,
+      "step": 2761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6013011927600303e-05,
+      "loss": 0.5229,
+      "step": 2762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.601134373175411e-05,
+      "loss": 0.4906,
+      "step": 2763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6009675535907915e-05,
+      "loss": 0.5225,
+      "step": 2764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6008007340061725e-05,
+      "loss": 0.5226,
+      "step": 2765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.600633914421553e-05,
+      "loss": 0.4776,
+      "step": 2766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.600467094836934e-05,
+      "loss": 0.4962,
+      "step": 2767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6003002752523147e-05,
+      "loss": 0.5235,
+      "step": 2768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6001334556676956e-05,
+      "loss": 0.5014,
+      "step": 2769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5999666360830763e-05,
+      "loss": 0.5064,
+      "step": 2770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.599799816498457e-05,
+      "loss": 0.4866,
+      "step": 2771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.599632996913838e-05,
+      "loss": 0.506,
+      "step": 2772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5994661773292185e-05,
+      "loss": 0.4963,
+      "step": 2773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5992993577445994e-05,
+      "loss": 0.4711,
+      "step": 2774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.59913253815998e-05,
+      "loss": 0.4808,
+      "step": 2775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.598965718575361e-05,
+      "loss": 0.5202,
+      "step": 2776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5987988989907416e-05,
+      "loss": 0.4783,
+      "step": 2777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5986320794061222e-05,
+      "loss": 0.4957,
+      "step": 2778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5984652598215032e-05,
+      "loss": 0.4768,
+      "step": 2779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5982984402368838e-05,
+      "loss": 0.4876,
+      "step": 2780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.5981316206522648e-05,
+      "loss": 0.444,
+      "step": 2781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5979648010676454e-05,
+      "loss": 0.4643,
+      "step": 2782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5977979814830263e-05,
+      "loss": 0.5098,
+      "step": 2783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597631161898407e-05,
+      "loss": 0.5545,
+      "step": 2784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5974643423137876e-05,
+      "loss": 0.4903,
+      "step": 2785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5972975227291685e-05,
+      "loss": 0.4631,
+      "step": 2786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.597130703144549e-05,
+      "loss": 0.4994,
+      "step": 2787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59696388355993e-05,
+      "loss": 0.4944,
+      "step": 2788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5967970639753107e-05,
+      "loss": 0.4695,
+      "step": 2789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5966302443906917e-05,
+      "loss": 0.4652,
+      "step": 2790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5964634248060723e-05,
+      "loss": 0.5529,
+      "step": 2791
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.596296605221453e-05,
+      "loss": 0.4812,
+      "step": 2792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.596129785636834e-05,
+      "loss": 0.5019,
+      "step": 2793
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595962966052215e-05,
+      "loss": 0.4944,
+      "step": 2794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5957961464675955e-05,
+      "loss": 0.4868,
+      "step": 2795
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595629326882976e-05,
+      "loss": 0.4871,
+      "step": 2796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.595462507298357e-05,
+      "loss": 0.5266,
+      "step": 2797
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5952956877137377e-05,
+      "loss": 0.4706,
+      "step": 2798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5951288681291183e-05,
+      "loss": 0.5055,
+      "step": 2799
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5949620485444992e-05,
+      "loss": 0.4827,
+      "step": 2800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5947952289598802e-05,
+      "loss": 0.4982,
+      "step": 2801
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5946284093752608e-05,
+      "loss": 0.5153,
+      "step": 2802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5944615897906414e-05,
+      "loss": 0.4971,
+      "step": 2803
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5942947702060224e-05,
+      "loss": 0.4539,
+      "step": 2804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.594127950621403e-05,
+      "loss": 0.5196,
+      "step": 2805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593961131036784e-05,
+      "loss": 0.492,
+      "step": 2806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5937943114521646e-05,
+      "loss": 0.5152,
+      "step": 2807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5936274918675455e-05,
+      "loss": 0.5354,
+      "step": 2808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.593460672282926e-05,
+      "loss": 0.5074,
+      "step": 2809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5932938526983068e-05,
+      "loss": 0.4538,
+      "step": 2810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5931270331136877e-05,
+      "loss": 0.4996,
+      "step": 2811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5929602135290684e-05,
+      "loss": 0.5382,
+      "step": 2812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5927933939444493e-05,
+      "loss": 0.5141,
+      "step": 2813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.59262657435983e-05,
+      "loss": 0.4913,
+      "step": 2814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.592459754775211e-05,
+      "loss": 0.5036,
+      "step": 2815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5922929351905915e-05,
+      "loss": 0.489,
+      "step": 2816
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.592126115605972e-05,
+      "loss": 0.4537,
+      "step": 2817
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.591959296021353e-05,
+      "loss": 0.4704,
+      "step": 2818
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5917924764367337e-05,
+      "loss": 0.4953,
+      "step": 2819
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5916256568521147e-05,
+      "loss": 0.4967,
+      "step": 2820
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5914588372674953e-05,
+      "loss": 0.5115,
+      "step": 2821
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.5912920176828763e-05,
+      "loss": 0.4676,
+      "step": 2822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.591125198098257e-05,
+      "loss": 0.4614,
+      "step": 2823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5909583785136375e-05,
+      "loss": 0.4942,
+      "step": 2824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5907915589290184e-05,
+      "loss": 0.4695,
+      "step": 2825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.590624739344399e-05,
+      "loss": 0.4834,
+      "step": 2826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.59045791975978e-05,
+      "loss": 0.4622,
+      "step": 2827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5902911001751606e-05,
+      "loss": 0.4407,
+      "step": 2828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5901242805905416e-05,
+      "loss": 0.5175,
+      "step": 2829
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5899574610059222e-05,
+      "loss": 0.496,
+      "step": 2830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.589790641421303e-05,
+      "loss": 0.5336,
+      "step": 2831
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5896238218366838e-05,
+      "loss": 0.4802,
+      "step": 2832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5894570022520644e-05,
+      "loss": 0.5501,
+      "step": 2833
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5892901826674454e-05,
+      "loss": 0.4681,
+      "step": 2834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.589123363082826e-05,
+      "loss": 0.5003,
+      "step": 2835
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.588956543498207e-05,
+      "loss": 0.5136,
+      "step": 2836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5887897239135876e-05,
+      "loss": 0.5101,
+      "step": 2837
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5886229043289682e-05,
+      "loss": 0.4923,
+      "step": 2838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.588456084744349e-05,
+      "loss": 0.4878,
+      "step": 2839
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5882892651597298e-05,
+      "loss": 0.4816,
+      "step": 2840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5881224455751107e-05,
+      "loss": 0.5333,
+      "step": 2841
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5879556259904913e-05,
+      "loss": 0.5162,
+      "step": 2842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5877888064058723e-05,
+      "loss": 0.4949,
+      "step": 2843
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.587621986821253e-05,
+      "loss": 0.5148,
+      "step": 2844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5874551672366335e-05,
+      "loss": 0.4971,
+      "step": 2845
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5872883476520145e-05,
+      "loss": 0.4979,
+      "step": 2846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.587121528067395e-05,
+      "loss": 0.5057,
+      "step": 2847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586954708482776e-05,
+      "loss": 0.506,
+      "step": 2848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5867878888981567e-05,
+      "loss": 0.4848,
+      "step": 2849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5866210693135377e-05,
+      "loss": 0.4929,
+      "step": 2850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5864542497289183e-05,
+      "loss": 0.4847,
+      "step": 2851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.586287430144299e-05,
+      "loss": 0.4742,
+      "step": 2852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.58612061055968e-05,
+      "loss": 0.4796,
+      "step": 2853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5859537909750605e-05,
+      "loss": 0.5134,
+      "step": 2854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5857869713904414e-05,
+      "loss": 0.4859,
+      "step": 2855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.585620151805822e-05,
+      "loss": 0.4926,
+      "step": 2856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.585453332221203e-05,
+      "loss": 0.5371,
+      "step": 2857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5852865126365836e-05,
+      "loss": 0.5374,
+      "step": 2858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5851196930519642e-05,
+      "loss": 0.4856,
+      "step": 2859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5849528734673452e-05,
+      "loss": 0.4629,
+      "step": 2860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.584786053882726e-05,
+      "loss": 0.4678,
+      "step": 2861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5846192342981068e-05,
+      "loss": 0.5132,
+      "step": 2862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.5844524147134874e-05,
+      "loss": 0.4994,
+      "step": 2863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5842855951288684e-05,
+      "loss": 0.5131,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.584118775544249e-05,
+      "loss": 0.4991,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5839519559596296e-05,
+      "loss": 0.506,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5837851363750106e-05,
+      "loss": 0.5037,
+      "step": 2867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5836183167903915e-05,
+      "loss": 0.5134,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.583451497205772e-05,
+      "loss": 0.5107,
+      "step": 2869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5832846776211528e-05,
+      "loss": 0.4693,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5831178580365337e-05,
+      "loss": 0.468,
+      "step": 2871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5829510384519143e-05,
+      "loss": 0.5053,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582784218867295e-05,
+      "loss": 0.4848,
+      "step": 2873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582617399282676e-05,
+      "loss": 0.5348,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582450579698057e-05,
+      "loss": 0.4999,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5822837601134375e-05,
+      "loss": 0.499,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.582116940528818e-05,
+      "loss": 0.4857,
+      "step": 2877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.581950120944199e-05,
+      "loss": 0.475,
+      "step": 2878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5817833013595797e-05,
+      "loss": 0.4762,
+      "step": 2879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5816164817749603e-05,
+      "loss": 0.5051,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5814496621903413e-05,
+      "loss": 0.4926,
+      "step": 2881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5812828426057222e-05,
+      "loss": 0.4982,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.581116023021103e-05,
+      "loss": 0.5039,
+      "step": 2883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5809492034364835e-05,
+      "loss": 0.4758,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5807823838518644e-05,
+      "loss": 0.51,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.580615564267245e-05,
+      "loss": 0.4754,
+      "step": 2886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5804487446826257e-05,
+      "loss": 0.4913,
+      "step": 2887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5802819250980066e-05,
+      "loss": 0.4956,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5801151055133876e-05,
+      "loss": 0.4952,
+      "step": 2889
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5799482859287682e-05,
+      "loss": 0.5033,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5797814663441488e-05,
+      "loss": 0.5378,
+      "step": 2891
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5796146467595298e-05,
+      "loss": 0.4723,
+      "step": 2892
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5794478271749104e-05,
+      "loss": 0.4838,
+      "step": 2893
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.579281007590291e-05,
+      "loss": 0.5192,
+      "step": 2894
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.579114188005672e-05,
+      "loss": 0.5023,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.4926,
+      "step": 2896
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5787805488364335e-05,
+      "loss": 0.5084,
+      "step": 2897
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.578613729251814e-05,
+      "loss": 0.528,
+      "step": 2898
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.578446909667195e-05,
+      "loss": 0.4556,
+      "step": 2899
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5782800900825757e-05,
+      "loss": 0.4698,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5781132704979564e-05,
+      "loss": 0.4801,
+      "step": 2901
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5779464509133373e-05,
+      "loss": 0.5238,
+      "step": 2902
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.5777796313287183e-05,
+      "loss": 0.459,
+      "step": 2903
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.577612811744099e-05,
+      "loss": 0.4952,
+      "step": 2904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5774459921594795e-05,
+      "loss": 0.4893,
+      "step": 2905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5772791725748605e-05,
+      "loss": 0.4478,
+      "step": 2906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.577112352990241e-05,
+      "loss": 0.4756,
+      "step": 2907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5769455334056217e-05,
+      "loss": 0.487,
+      "step": 2908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5767787138210027e-05,
+      "loss": 0.4551,
+      "step": 2909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5766118942363836e-05,
+      "loss": 0.4858,
+      "step": 2910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5764450746517642e-05,
+      "loss": 0.4847,
+      "step": 2911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.576278255067145e-05,
+      "loss": 0.4772,
+      "step": 2912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5761114354825258e-05,
+      "loss": 0.5004,
+      "step": 2913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5759446158979064e-05,
+      "loss": 0.4675,
+      "step": 2914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.575777796313287e-05,
+      "loss": 0.4862,
+      "step": 2915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.575610976728668e-05,
+      "loss": 0.4891,
+      "step": 2916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.575444157144049e-05,
+      "loss": 0.4912,
+      "step": 2917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5752773375594296e-05,
+      "loss": 0.4857,
+      "step": 2918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5751105179748102e-05,
+      "loss": 0.4998,
+      "step": 2919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5749436983901912e-05,
+      "loss": 0.4796,
+      "step": 2920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5747768788055718e-05,
+      "loss": 0.4936,
+      "step": 2921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5746100592209524e-05,
+      "loss": 0.4933,
+      "step": 2922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5744432396363334e-05,
+      "loss": 0.5202,
+      "step": 2923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5742764200517143e-05,
+      "loss": 0.4889,
+      "step": 2924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.574109600467095e-05,
+      "loss": 0.5313,
+      "step": 2925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5739427808824756e-05,
+      "loss": 0.4717,
+      "step": 2926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5737759612978565e-05,
+      "loss": 0.4833,
+      "step": 2927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.573609141713237e-05,
+      "loss": 0.5183,
+      "step": 2928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.573442322128618e-05,
+      "loss": 0.4931,
+      "step": 2929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5732755025439987e-05,
+      "loss": 0.4613,
+      "step": 2930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5731086829593797e-05,
+      "loss": 0.4835,
+      "step": 2931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5729418633747603e-05,
+      "loss": 0.5001,
+      "step": 2932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.572775043790141e-05,
+      "loss": 0.4907,
+      "step": 2933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.572608224205522e-05,
+      "loss": 0.4797,
+      "step": 2934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.572441404620903e-05,
+      "loss": 0.5046,
+      "step": 2935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5722745850362835e-05,
+      "loss": 0.4528,
+      "step": 2936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.572107765451664e-05,
+      "loss": 0.4794,
+      "step": 2937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.571940945867045e-05,
+      "loss": 0.5113,
+      "step": 2938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5717741262824256e-05,
+      "loss": 0.5084,
+      "step": 2939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5716073066978063e-05,
+      "loss": 0.4726,
+      "step": 2940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5714404871131872e-05,
+      "loss": 0.5055,
+      "step": 2941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5712736675285682e-05,
+      "loss": 0.4701,
+      "step": 2942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5711068479439488e-05,
+      "loss": 0.4706,
+      "step": 2943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5709400283593294e-05,
+      "loss": 0.5165,
+      "step": 2944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.5707732087747104e-05,
+      "loss": 0.4585,
+      "step": 2945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.570606389190091e-05,
+      "loss": 0.49,
+      "step": 2946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5704395696054716e-05,
+      "loss": 0.5233,
+      "step": 2947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5702727500208526e-05,
+      "loss": 0.5423,
+      "step": 2948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5701059304362335e-05,
+      "loss": 0.4978,
+      "step": 2949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.569939110851614e-05,
+      "loss": 0.4959,
+      "step": 2950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5697722912669948e-05,
+      "loss": 0.4702,
+      "step": 2951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5696054716823757e-05,
+      "loss": 0.4986,
+      "step": 2952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5694386520977564e-05,
+      "loss": 0.5177,
+      "step": 2953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.569271832513137e-05,
+      "loss": 0.5006,
+      "step": 2954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.569105012928518e-05,
+      "loss": 0.5086,
+      "step": 2955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.568938193343899e-05,
+      "loss": 0.455,
+      "step": 2956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5687713737592795e-05,
+      "loss": 0.4944,
+      "step": 2957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.56860455417466e-05,
+      "loss": 0.5115,
+      "step": 2958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.568437734590041e-05,
+      "loss": 0.4519,
+      "step": 2959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5682709150054217e-05,
+      "loss": 0.5169,
+      "step": 2960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5681040954208023e-05,
+      "loss": 0.4984,
+      "step": 2961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5679372758361833e-05,
+      "loss": 0.4975,
+      "step": 2962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5677704562515642e-05,
+      "loss": 0.5002,
+      "step": 2963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.567603636666945e-05,
+      "loss": 0.5289,
+      "step": 2964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5674368170823255e-05,
+      "loss": 0.4879,
+      "step": 2965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5672699974977064e-05,
+      "loss": 0.5035,
+      "step": 2966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.567103177913087e-05,
+      "loss": 0.4927,
+      "step": 2967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5669363583284677e-05,
+      "loss": 0.4988,
+      "step": 2968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5667695387438486e-05,
+      "loss": 0.5263,
+      "step": 2969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5666027191592296e-05,
+      "loss": 0.4869,
+      "step": 2970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5664358995746102e-05,
+      "loss": 0.4804,
+      "step": 2971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5662690799899908e-05,
+      "loss": 0.5028,
+      "step": 2972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5661022604053718e-05,
+      "loss": 0.4592,
+      "step": 2973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5659354408207524e-05,
+      "loss": 0.4967,
+      "step": 2974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.565768621236133e-05,
+      "loss": 0.4965,
+      "step": 2975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.565601801651514e-05,
+      "loss": 0.4532,
+      "step": 2976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.565434982066895e-05,
+      "loss": 0.4767,
+      "step": 2977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5652681624822756e-05,
+      "loss": 0.485,
+      "step": 2978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5651013428976562e-05,
+      "loss": 0.4605,
+      "step": 2979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.564934523313037e-05,
+      "loss": 0.5115,
+      "step": 2980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5647677037284178e-05,
+      "loss": 0.4803,
+      "step": 2981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5646008841437984e-05,
+      "loss": 0.5149,
+      "step": 2982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5644340645591793e-05,
+      "loss": 0.4972,
+      "step": 2983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5642672449745603e-05,
+      "loss": 0.4906,
+      "step": 2984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.564100425389941e-05,
+      "loss": 0.484,
+      "step": 2985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5639336058053215e-05,
+      "loss": 0.4751,
+      "step": 2986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.5637667862207025e-05,
+      "loss": 0.4774,
+      "step": 2987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.563599966636083e-05,
+      "loss": 0.5692,
+      "step": 2988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5634331470514637e-05,
+      "loss": 0.4902,
+      "step": 2989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5632663274668447e-05,
+      "loss": 0.5359,
+      "step": 2990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5630995078822256e-05,
+      "loss": 0.4783,
+      "step": 2991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5629326882976063e-05,
+      "loss": 0.4926,
+      "step": 2992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.562765868712987e-05,
+      "loss": 0.4556,
+      "step": 2993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.562599049128368e-05,
+      "loss": 0.4877,
+      "step": 2994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5624322295437485e-05,
+      "loss": 0.4896,
+      "step": 2995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.562265409959129e-05,
+      "loss": 0.4402,
+      "step": 2996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.56209859037451e-05,
+      "loss": 0.5057,
+      "step": 2997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.561931770789891e-05,
+      "loss": 0.4884,
+      "step": 2998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5617649512052716e-05,
+      "loss": 0.4715,
+      "step": 2999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5615981316206522e-05,
+      "loss": 0.4957,
+      "step": 3000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5614313120360332e-05,
+      "loss": 0.5332,
+      "step": 3001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5612644924514138e-05,
+      "loss": 0.5057,
+      "step": 3002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5610976728667944e-05,
+      "loss": 0.4918,
+      "step": 3003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5609308532821754e-05,
+      "loss": 0.5093,
+      "step": 3004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5607640336975563e-05,
+      "loss": 0.4984,
+      "step": 3005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.560597214112937e-05,
+      "loss": 0.4719,
+      "step": 3006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5604303945283176e-05,
+      "loss": 0.5369,
+      "step": 3007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5602635749436985e-05,
+      "loss": 0.4803,
+      "step": 3008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5600967553590795e-05,
+      "loss": 0.5366,
+      "step": 3009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5599299357744598e-05,
+      "loss": 0.4771,
+      "step": 3010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5597631161898407e-05,
+      "loss": 0.4759,
+      "step": 3011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5595962966052217e-05,
+      "loss": 0.4684,
+      "step": 3012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5594294770206023e-05,
+      "loss": 0.4807,
+      "step": 3013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.559262657435983e-05,
+      "loss": 0.4691,
+      "step": 3014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.559095837851364e-05,
+      "loss": 0.5411,
+      "step": 3015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.558929018266745e-05,
+      "loss": 0.459,
+      "step": 3016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.558762198682125e-05,
+      "loss": 0.4843,
+      "step": 3017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.558595379097506e-05,
+      "loss": 0.4722,
+      "step": 3018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.558428559512887e-05,
+      "loss": 0.5319,
+      "step": 3019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5582617399282677e-05,
+      "loss": 0.5168,
+      "step": 3020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5580949203436483e-05,
+      "loss": 0.4758,
+      "step": 3021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5579281007590292e-05,
+      "loss": 0.4887,
+      "step": 3022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5577612811744102e-05,
+      "loss": 0.4821,
+      "step": 3023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5575944615897908e-05,
+      "loss": 0.5113,
+      "step": 3024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5574276420051714e-05,
+      "loss": 0.4983,
+      "step": 3025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5572608224205524e-05,
+      "loss": 0.4619,
+      "step": 3026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.557094002835933e-05,
+      "loss": 0.4815,
+      "step": 3027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.5569271832513136e-05,
+      "loss": 0.479,
+      "step": 3028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5567603636666946e-05,
+      "loss": 0.4864,
+      "step": 3029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5565935440820756e-05,
+      "loss": 0.4844,
+      "step": 3030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5564267244974562e-05,
+      "loss": 0.4999,
+      "step": 3031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5562599049128368e-05,
+      "loss": 0.4465,
+      "step": 3032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5560930853282178e-05,
+      "loss": 0.4965,
+      "step": 3033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5559262657435984e-05,
+      "loss": 0.5068,
+      "step": 3034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555759446158979e-05,
+      "loss": 0.4476,
+      "step": 3035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.55559262657436e-05,
+      "loss": 0.4753,
+      "step": 3036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555425806989741e-05,
+      "loss": 0.4695,
+      "step": 3037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5552589874051215e-05,
+      "loss": 0.5094,
+      "step": 3038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555092167820502e-05,
+      "loss": 0.4873,
+      "step": 3039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.554925348235883e-05,
+      "loss": 0.4934,
+      "step": 3040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5547585286512637e-05,
+      "loss": 0.4789,
+      "step": 3041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5545917090666443e-05,
+      "loss": 0.4884,
+      "step": 3042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5544248894820253e-05,
+      "loss": 0.4611,
+      "step": 3043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5542580698974063e-05,
+      "loss": 0.5143,
+      "step": 3044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.554091250312787e-05,
+      "loss": 0.4651,
+      "step": 3045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5539244307281675e-05,
+      "loss": 0.5312,
+      "step": 3046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5537576111435485e-05,
+      "loss": 0.456,
+      "step": 3047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.553590791558929e-05,
+      "loss": 0.5015,
+      "step": 3048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5534239719743097e-05,
+      "loss": 0.4982,
+      "step": 3049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5532571523896907e-05,
+      "loss": 0.5059,
+      "step": 3050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5530903328050716e-05,
+      "loss": 0.513,
+      "step": 3051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5529235132204522e-05,
+      "loss": 0.4696,
+      "step": 3052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.552756693635833e-05,
+      "loss": 0.5303,
+      "step": 3053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5525898740512138e-05,
+      "loss": 0.4876,
+      "step": 3054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5524230544665944e-05,
+      "loss": 0.4596,
+      "step": 3055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.552256234881975e-05,
+      "loss": 0.4674,
+      "step": 3056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.552089415297356e-05,
+      "loss": 0.5136,
+      "step": 3057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.551922595712737e-05,
+      "loss": 0.4955,
+      "step": 3058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5517557761281176e-05,
+      "loss": 0.5321,
+      "step": 3059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5515889565434982e-05,
+      "loss": 0.4877,
+      "step": 3060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.551422136958879e-05,
+      "loss": 0.4637,
+      "step": 3061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5512553173742598e-05,
+      "loss": 0.4775,
+      "step": 3062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5510884977896404e-05,
+      "loss": 0.4874,
+      "step": 3063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5509216782050214e-05,
+      "loss": 0.5007,
+      "step": 3064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5507548586204023e-05,
+      "loss": 0.4738,
+      "step": 3065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.550588039035783e-05,
+      "loss": 0.4848,
+      "step": 3066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5504212194511636e-05,
+      "loss": 0.4758,
+      "step": 3067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5502543998665445e-05,
+      "loss": 0.5223,
+      "step": 3068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.550087580281925e-05,
+      "loss": 0.4426,
+      "step": 3069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5499207606973058e-05,
+      "loss": 0.4274,
+      "step": 3070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5497539411126867e-05,
+      "loss": 0.4684,
+      "step": 3071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5495871215280677e-05,
+      "loss": 0.4767,
+      "step": 3072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5494203019434483e-05,
+      "loss": 0.4771,
+      "step": 3073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.549253482358829e-05,
+      "loss": 0.4729,
+      "step": 3074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.54908666277421e-05,
+      "loss": 0.4874,
+      "step": 3075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5489198431895908e-05,
+      "loss": 0.5014,
+      "step": 3076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548753023604971e-05,
+      "loss": 0.4709,
+      "step": 3077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548586204020352e-05,
+      "loss": 0.476,
+      "step": 3078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548419384435733e-05,
+      "loss": 0.4613,
+      "step": 3079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5482525648511136e-05,
+      "loss": 0.5014,
+      "step": 3080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5480857452664943e-05,
+      "loss": 0.5004,
+      "step": 3081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5479189256818752e-05,
+      "loss": 0.5034,
+      "step": 3082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5477521060972562e-05,
+      "loss": 0.4975,
+      "step": 3083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5475852865126365e-05,
+      "loss": 0.4577,
+      "step": 3084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5474184669280174e-05,
+      "loss": 0.5131,
+      "step": 3085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5472516473433984e-05,
+      "loss": 0.4768,
+      "step": 3086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547084827758779e-05,
+      "loss": 0.503,
+      "step": 3087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5469180081741596e-05,
+      "loss": 0.4766,
+      "step": 3088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5467511885895406e-05,
+      "loss": 0.5112,
+      "step": 3089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5465843690049215e-05,
+      "loss": 0.4951,
+      "step": 3090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546417549420302e-05,
+      "loss": 0.5059,
+      "step": 3091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5462507298356828e-05,
+      "loss": 0.4776,
+      "step": 3092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5460839102510637e-05,
+      "loss": 0.482,
+      "step": 3093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5459170906664443e-05,
+      "loss": 0.4906,
+      "step": 3094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545750271081825e-05,
+      "loss": 0.4693,
+      "step": 3095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545583451497206e-05,
+      "loss": 0.4871,
+      "step": 3096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545416631912587e-05,
+      "loss": 0.4998,
+      "step": 3097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5452498123279675e-05,
+      "loss": 0.4786,
+      "step": 3098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545082992743348e-05,
+      "loss": 0.4906,
+      "step": 3099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544916173158729e-05,
+      "loss": 0.5325,
+      "step": 3100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5447493535741097e-05,
+      "loss": 0.4626,
+      "step": 3101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5445825339894903e-05,
+      "loss": 0.4616,
+      "step": 3102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5444157144048713e-05,
+      "loss": 0.4858,
+      "step": 3103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5442488948202522e-05,
+      "loss": 0.5008,
+      "step": 3104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544082075235633e-05,
+      "loss": 0.4565,
+      "step": 3105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5439152556510135e-05,
+      "loss": 0.4842,
+      "step": 3106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5437484360663944e-05,
+      "loss": 0.5028,
+      "step": 3107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.543581616481775e-05,
+      "loss": 0.4962,
+      "step": 3108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5434147968971557e-05,
+      "loss": 0.4838,
+      "step": 3109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5432479773125366e-05,
+      "loss": 0.509,
+      "step": 3110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5430811577279176e-05,
+      "loss": 0.501,
+      "step": 3111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5429143381432982e-05,
+      "loss": 0.478,
+      "step": 3112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5427475185586788e-05,
+      "loss": 0.4727,
+      "step": 3113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5425806989740598e-05,
+      "loss": 0.5176,
+      "step": 3114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5424138793894404e-05,
+      "loss": 0.4681,
+      "step": 3115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.542247059804821e-05,
+      "loss": 0.5223,
+      "step": 3116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.542080240220202e-05,
+      "loss": 0.5125,
+      "step": 3117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.541913420635583e-05,
+      "loss": 0.5019,
+      "step": 3118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5417466010509636e-05,
+      "loss": 0.4964,
+      "step": 3119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5415797814663442e-05,
+      "loss": 0.4851,
+      "step": 3120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.541412961881725e-05,
+      "loss": 0.4443,
+      "step": 3121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5412461422971057e-05,
+      "loss": 0.4948,
+      "step": 3122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5410793227124864e-05,
+      "loss": 0.4952,
+      "step": 3123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5409125031278673e-05,
+      "loss": 0.5158,
+      "step": 3124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5407456835432483e-05,
+      "loss": 0.5106,
+      "step": 3125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540578863958629e-05,
+      "loss": 0.4966,
+      "step": 3126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5404120443740095e-05,
+      "loss": 0.4724,
+      "step": 3127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5402452247893905e-05,
+      "loss": 0.4844,
+      "step": 3128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540078405204771e-05,
+      "loss": 0.508,
+      "step": 3129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5399115856201517e-05,
+      "loss": 0.4657,
+      "step": 3130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5397447660355327e-05,
+      "loss": 0.4749,
+      "step": 3131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5395779464509136e-05,
+      "loss": 0.5066,
+      "step": 3132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5394111268662943e-05,
+      "loss": 0.4948,
+      "step": 3133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.539244307281675e-05,
+      "loss": 0.4892,
+      "step": 3134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.539077487697056e-05,
+      "loss": 0.4659,
+      "step": 3135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5389106681124365e-05,
+      "loss": 0.4809,
+      "step": 3136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.538743848527817e-05,
+      "loss": 0.4982,
+      "step": 3137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.538577028943198e-05,
+      "loss": 0.4803,
+      "step": 3138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.538410209358579e-05,
+      "loss": 0.4438,
+      "step": 3139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5382433897739596e-05,
+      "loss": 0.4336,
+      "step": 3140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5380765701893402e-05,
+      "loss": 0.5275,
+      "step": 3141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5379097506047212e-05,
+      "loss": 0.516,
+      "step": 3142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5377429310201018e-05,
+      "loss": 0.5105,
+      "step": 3143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5375761114354824e-05,
+      "loss": 0.511,
+      "step": 3144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5374092918508634e-05,
+      "loss": 0.4849,
+      "step": 3145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5372424722662443e-05,
+      "loss": 0.5067,
+      "step": 3146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.537075652681625e-05,
+      "loss": 0.5698,
+      "step": 3147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5369088330970056e-05,
+      "loss": 0.4597,
+      "step": 3148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5367420135123865e-05,
+      "loss": 0.4733,
+      "step": 3149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5365751939277675e-05,
+      "loss": 0.5152,
+      "step": 3150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5364083743431478e-05,
+      "loss": 0.4933,
+      "step": 3151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5362415547585287e-05,
+      "loss": 0.4876,
+      "step": 3152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5360747351739097e-05,
+      "loss": 0.4809,
+      "step": 3153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5359079155892903e-05,
+      "loss": 0.513,
+      "step": 3154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535741096004671e-05,
+      "loss": 0.4678,
+      "step": 3155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535574276420052e-05,
+      "loss": 0.4835,
+      "step": 3156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535407456835433e-05,
+      "loss": 0.4966,
+      "step": 3157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535240637250813e-05,
+      "loss": 0.4754,
+      "step": 3158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535073817666194e-05,
+      "loss": 0.5576,
+      "step": 3159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.534906998081575e-05,
+      "loss": 0.4688,
+      "step": 3160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5347401784969557e-05,
+      "loss": 0.5047,
+      "step": 3161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5345733589123363e-05,
+      "loss": 0.495,
+      "step": 3162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5344065393277172e-05,
+      "loss": 0.5006,
+      "step": 3163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5342397197430982e-05,
+      "loss": 0.4936,
+      "step": 3164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5340729001584788e-05,
+      "loss": 0.4911,
+      "step": 3165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5339060805738594e-05,
+      "loss": 0.494,
+      "step": 3166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5337392609892404e-05,
+      "loss": 0.4572,
+      "step": 3167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.533572441404621e-05,
+      "loss": 0.4787,
+      "step": 3168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5334056218200016e-05,
+      "loss": 0.4637,
+      "step": 3169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5332388022353826e-05,
+      "loss": 0.5059,
+      "step": 3170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5330719826507635e-05,
+      "loss": 0.5294,
+      "step": 3171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.532905163066144e-05,
+      "loss": 0.517,
+      "step": 3172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5327383434815248e-05,
+      "loss": 0.4567,
+      "step": 3173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5325715238969057e-05,
+      "loss": 0.4816,
+      "step": 3174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5324047043122864e-05,
+      "loss": 0.4454,
+      "step": 3175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.532237884727667e-05,
+      "loss": 0.5273,
+      "step": 3176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.532071065143048e-05,
+      "loss": 0.4803,
+      "step": 3177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531904245558429e-05,
+      "loss": 0.4755,
+      "step": 3178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5317374259738095e-05,
+      "loss": 0.5087,
+      "step": 3179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.53157060638919e-05,
+      "loss": 0.4488,
+      "step": 3180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.531403786804571e-05,
+      "loss": 0.4698,
+      "step": 3181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5312369672199517e-05,
+      "loss": 0.4953,
+      "step": 3182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5310701476353323e-05,
+      "loss": 0.4959,
+      "step": 3183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5309033280507133e-05,
+      "loss": 0.4673,
+      "step": 3184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5307365084660943e-05,
+      "loss": 0.4841,
+      "step": 3185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.530569688881475e-05,
+      "loss": 0.4763,
+      "step": 3186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5304028692968555e-05,
+      "loss": 0.4548,
+      "step": 3187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5302360497122364e-05,
+      "loss": 0.532,
+      "step": 3188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.530069230127617e-05,
+      "loss": 0.4722,
+      "step": 3189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5299024105429977e-05,
+      "loss": 0.5018,
+      "step": 3190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5297355909583786e-05,
+      "loss": 0.4706,
+      "step": 3191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5295687713737596e-05,
+      "loss": 0.4663,
+      "step": 3192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5294019517891402e-05,
+      "loss": 0.5226,
+      "step": 3193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.529235132204521e-05,
+      "loss": 0.489,
+      "step": 3194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5290683126199018e-05,
+      "loss": 0.5182,
+      "step": 3195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5289014930352824e-05,
+      "loss": 0.5279,
+      "step": 3196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.528734673450663e-05,
+      "loss": 0.5226,
+      "step": 3197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.528567853866044e-05,
+      "loss": 0.4801,
+      "step": 3198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.528401034281425e-05,
+      "loss": 0.497,
+      "step": 3199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5282342146968056e-05,
+      "loss": 0.4872,
+      "step": 3200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5280673951121862e-05,
+      "loss": 0.5035,
+      "step": 3201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527900575527567e-05,
+      "loss": 0.4989,
+      "step": 3202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5277337559429478e-05,
+      "loss": 0.4805,
+      "step": 3203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5275669363583284e-05,
+      "loss": 0.4986,
+      "step": 3204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5274001167737093e-05,
+      "loss": 0.5304,
+      "step": 3205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5272332971890903e-05,
+      "loss": 0.4618,
+      "step": 3206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.527066477604471e-05,
+      "loss": 0.4393,
+      "step": 3207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5268996580198515e-05,
+      "loss": 0.447,
+      "step": 3208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5267328384352325e-05,
+      "loss": 0.4918,
+      "step": 3209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.526566018850613e-05,
+      "loss": 0.4973,
+      "step": 3210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5263991992659937e-05,
+      "loss": 0.4711,
+      "step": 3211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5262323796813747e-05,
+      "loss": 0.5008,
+      "step": 3212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5260655600967557e-05,
+      "loss": 0.4827,
+      "step": 3213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5258987405121363e-05,
+      "loss": 0.4762,
+      "step": 3214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.525731920927517e-05,
+      "loss": 0.4674,
+      "step": 3215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5255651013428979e-05,
+      "loss": 0.4692,
+      "step": 3216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5253982817582786e-05,
+      "loss": 0.481,
+      "step": 3217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5252314621736593e-05,
+      "loss": 0.4602,
+      "step": 3218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.52506464258904e-05,
+      "loss": 0.4824,
+      "step": 3219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5248978230044208e-05,
+      "loss": 0.4658,
+      "step": 3220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5247310034198015e-05,
+      "loss": 0.4608,
+      "step": 3221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5245641838351824e-05,
+      "loss": 0.4878,
+      "step": 3222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5243973642505632e-05,
+      "loss": 0.467,
+      "step": 3223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.524230544665944e-05,
+      "loss": 0.5012,
+      "step": 3224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5240637250813246e-05,
+      "loss": 0.4999,
+      "step": 3225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5238969054967054e-05,
+      "loss": 0.4816,
+      "step": 3226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5237300859120862e-05,
+      "loss": 0.4983,
+      "step": 3227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5235632663274668e-05,
+      "loss": 0.462,
+      "step": 3228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5233964467428478e-05,
+      "loss": 0.448,
+      "step": 3229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5232296271582286e-05,
+      "loss": 0.4736,
+      "step": 3230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5230628075736093e-05,
+      "loss": 0.5363,
+      "step": 3231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.52289598798899e-05,
+      "loss": 0.4653,
+      "step": 3232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5227291684043708e-05,
+      "loss": 0.5044,
+      "step": 3233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.5225623488197515e-05,
+      "loss": 0.4403,
+      "step": 3234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5223955292351322e-05,
+      "loss": 0.5134,
+      "step": 3235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5222287096505131e-05,
+      "loss": 0.4685,
+      "step": 3236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5220618900658939e-05,
+      "loss": 0.5301,
+      "step": 3237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5218950704812747e-05,
+      "loss": 0.5107,
+      "step": 3238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5217282508966553e-05,
+      "loss": 0.5105,
+      "step": 3239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5215614313120361e-05,
+      "loss": 0.5046,
+      "step": 3240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.521394611727417e-05,
+      "loss": 0.5267,
+      "step": 3241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5212277921427975e-05,
+      "loss": 0.4718,
+      "step": 3242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5210609725581785e-05,
+      "loss": 0.5146,
+      "step": 3243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5208941529735593e-05,
+      "loss": 0.5126,
+      "step": 3244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.52072733338894e-05,
+      "loss": 0.5045,
+      "step": 3245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5205605138043207e-05,
+      "loss": 0.4991,
+      "step": 3246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5203936942197015e-05,
+      "loss": 0.4702,
+      "step": 3247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5202268746350824e-05,
+      "loss": 0.4735,
+      "step": 3248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5200600550504629e-05,
+      "loss": 0.5044,
+      "step": 3249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5198932354658438e-05,
+      "loss": 0.4398,
+      "step": 3250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5197264158812246e-05,
+      "loss": 0.513,
+      "step": 3251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5195595962966054e-05,
+      "loss": 0.4536,
+      "step": 3252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.519392776711986e-05,
+      "loss": 0.5048,
+      "step": 3253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5192259571273668e-05,
+      "loss": 0.4723,
+      "step": 3254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5190591375427478e-05,
+      "loss": 0.4874,
+      "step": 3255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5188923179581284e-05,
+      "loss": 0.4387,
+      "step": 3256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5187254983735092e-05,
+      "loss": 0.4895,
+      "step": 3257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.51855867878889e-05,
+      "loss": 0.5122,
+      "step": 3258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5183918592042708e-05,
+      "loss": 0.4961,
+      "step": 3259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5182250396196514e-05,
+      "loss": 0.5044,
+      "step": 3260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5180582200350322e-05,
+      "loss": 0.5128,
+      "step": 3261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5178914004504131e-05,
+      "loss": 0.504,
+      "step": 3262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5177245808657937e-05,
+      "loss": 0.4607,
+      "step": 3263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5175577612811745e-05,
+      "loss": 0.4835,
+      "step": 3264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5173909416965553e-05,
+      "loss": 0.4674,
+      "step": 3265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5172241221119361e-05,
+      "loss": 0.495,
+      "step": 3266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5170573025273167e-05,
+      "loss": 0.533,
+      "step": 3267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5168904829426975e-05,
+      "loss": 0.4949,
+      "step": 3268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5167236633580785e-05,
+      "loss": 0.4823,
+      "step": 3269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5165568437734591e-05,
+      "loss": 0.4624,
+      "step": 3270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5163900241888399e-05,
+      "loss": 0.4818,
+      "step": 3271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5162232046042207e-05,
+      "loss": 0.509,
+      "step": 3272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5160563850196015e-05,
+      "loss": 0.4434,
+      "step": 3273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.515889565434982e-05,
+      "loss": 0.4715,
+      "step": 3274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.5157227458503629e-05,
+      "loss": 0.4905,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5155559262657438e-05,
+      "loss": 0.5411,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5153891066811244e-05,
+      "loss": 0.4709,
+      "step": 3277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5152222870965052e-05,
+      "loss": 0.511,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.515055467511886e-05,
+      "loss": 0.456,
+      "step": 3279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5148886479272668e-05,
+      "loss": 0.464,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5147218283426474e-05,
+      "loss": 0.5029,
+      "step": 3281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5145550087580284e-05,
+      "loss": 0.5317,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5143881891734092e-05,
+      "loss": 0.4688,
+      "step": 3283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.51422136958879e-05,
+      "loss": 0.528,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5140545500041706e-05,
+      "loss": 0.4389,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5138877304195514e-05,
+      "loss": 0.4774,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5137209108349322e-05,
+      "loss": 0.4707,
+      "step": 3287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5135540912503128e-05,
+      "loss": 0.4769,
+      "step": 3288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5133872716656937e-05,
+      "loss": 0.5068,
+      "step": 3289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5132204520810745e-05,
+      "loss": 0.4943,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5130536324964553e-05,
+      "loss": 0.469,
+      "step": 3291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.512886812911836e-05,
+      "loss": 0.4811,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5127199933272167e-05,
+      "loss": 0.5064,
+      "step": 3293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5125531737425975e-05,
+      "loss": 0.5137,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5123863541579781e-05,
+      "loss": 0.5036,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5122195345733591e-05,
+      "loss": 0.4871,
+      "step": 3296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5120527149887399e-05,
+      "loss": 0.4846,
+      "step": 3297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5118858954041207e-05,
+      "loss": 0.4498,
+      "step": 3298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5117190758195013e-05,
+      "loss": 0.4933,
+      "step": 3299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.511552256234882e-05,
+      "loss": 0.4694,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5113854366502629e-05,
+      "loss": 0.5565,
+      "step": 3301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5112186170656435e-05,
+      "loss": 0.4615,
+      "step": 3302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5110517974810244e-05,
+      "loss": 0.4959,
+      "step": 3303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5108849778964052e-05,
+      "loss": 0.4797,
+      "step": 3304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.510718158311786e-05,
+      "loss": 0.4641,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5105513387271666e-05,
+      "loss": 0.4544,
+      "step": 3306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5103845191425474e-05,
+      "loss": 0.5002,
+      "step": 3307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5102176995579282e-05,
+      "loss": 0.4827,
+      "step": 3308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5100508799733088e-05,
+      "loss": 0.4854,
+      "step": 3309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5098840603886898e-05,
+      "loss": 0.4931,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5097172408040706e-05,
+      "loss": 0.506,
+      "step": 3311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5095504212194514e-05,
+      "loss": 0.51,
+      "step": 3312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.509383601634832e-05,
+      "loss": 0.4912,
+      "step": 3313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5092167820502128e-05,
+      "loss": 0.469,
+      "step": 3314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5090499624655937e-05,
+      "loss": 0.4425,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5088831428809742e-05,
+      "loss": 0.4451,
+      "step": 3316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.5087163232963551e-05,
+      "loss": 0.4906,
+      "step": 3317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.508549503711736e-05,
+      "loss": 0.4694,
+      "step": 3318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5083826841271167e-05,
+      "loss": 0.5196,
+      "step": 3319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5082158645424973e-05,
+      "loss": 0.4639,
+      "step": 3320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5080490449578781e-05,
+      "loss": 0.4867,
+      "step": 3321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.507882225373259e-05,
+      "loss": 0.4919,
+      "step": 3322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5077154057886395e-05,
+      "loss": 0.4701,
+      "step": 3323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5075485862040205e-05,
+      "loss": 0.4751,
+      "step": 3324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5073817666194013e-05,
+      "loss": 0.5258,
+      "step": 3325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.507214947034782e-05,
+      "loss": 0.5029,
+      "step": 3326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5070481274501627e-05,
+      "loss": 0.521,
+      "step": 3327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5068813078655435e-05,
+      "loss": 0.5037,
+      "step": 3328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5067144882809244e-05,
+      "loss": 0.5347,
+      "step": 3329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.506547668696305e-05,
+      "loss": 0.4916,
+      "step": 3330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5063808491116858e-05,
+      "loss": 0.5555,
+      "step": 3331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5062140295270666e-05,
+      "loss": 0.4835,
+      "step": 3332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5060472099424474e-05,
+      "loss": 0.4967,
+      "step": 3333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.505880390357828e-05,
+      "loss": 0.5327,
+      "step": 3334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5057135707732088e-05,
+      "loss": 0.4972,
+      "step": 3335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5055467511885898e-05,
+      "loss": 0.4937,
+      "step": 3336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5053799316039704e-05,
+      "loss": 0.5188,
+      "step": 3337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5052131120193512e-05,
+      "loss": 0.5038,
+      "step": 3338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.505046292434732e-05,
+      "loss": 0.5144,
+      "step": 3339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5048794728501128e-05,
+      "loss": 0.4716,
+      "step": 3340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5047126532654934e-05,
+      "loss": 0.4649,
+      "step": 3341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5045458336808742e-05,
+      "loss": 0.4607,
+      "step": 3342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5043790140962551e-05,
+      "loss": 0.5148,
+      "step": 3343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5042121945116358e-05,
+      "loss": 0.46,
+      "step": 3344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5040453749270165e-05,
+      "loss": 0.4984,
+      "step": 3345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5038785553423973e-05,
+      "loss": 0.4991,
+      "step": 3346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5037117357577781e-05,
+      "loss": 0.4757,
+      "step": 3347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5035449161731587e-05,
+      "loss": 0.4487,
+      "step": 3348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5033780965885395e-05,
+      "loss": 0.4978,
+      "step": 3349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5032112770039205e-05,
+      "loss": 0.4684,
+      "step": 3350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5030444574193011e-05,
+      "loss": 0.4852,
+      "step": 3351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5028776378346819e-05,
+      "loss": 0.491,
+      "step": 3352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5027108182500627e-05,
+      "loss": 0.4993,
+      "step": 3353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5025439986654435e-05,
+      "loss": 0.5108,
+      "step": 3354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5023771790808241e-05,
+      "loss": 0.4954,
+      "step": 3355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.502210359496205e-05,
+      "loss": 0.4743,
+      "step": 3356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5020435399115858e-05,
+      "loss": 0.4903,
+      "step": 3357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.5018767203269665e-05,
+      "loss": 0.5064,
+      "step": 3358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5017099007423473e-05,
+      "loss": 0.4589,
+      "step": 3359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.501543081157728e-05,
+      "loss": 0.5171,
+      "step": 3360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5013762615731088e-05,
+      "loss": 0.4761,
+      "step": 3361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5012094419884894e-05,
+      "loss": 0.5208,
+      "step": 3362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5010426224038704e-05,
+      "loss": 0.4703,
+      "step": 3363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5008758028192512e-05,
+      "loss": 0.4558,
+      "step": 3364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5007089832346318e-05,
+      "loss": 0.5533,
+      "step": 3365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5005421636500126e-05,
+      "loss": 0.4818,
+      "step": 3366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5003753440653934e-05,
+      "loss": 0.5058,
+      "step": 3367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5002085244807742e-05,
+      "loss": 0.4468,
+      "step": 3368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5000417048961548e-05,
+      "loss": 0.4972,
+      "step": 3369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4998748853115358e-05,
+      "loss": 0.4769,
+      "step": 3370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4997080657269165e-05,
+      "loss": 0.5186,
+      "step": 3371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4995412461422972e-05,
+      "loss": 0.458,
+      "step": 3372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.499374426557678e-05,
+      "loss": 0.4861,
+      "step": 3373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4992076069730587e-05,
+      "loss": 0.4772,
+      "step": 3374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4990407873884395e-05,
+      "loss": 0.4812,
+      "step": 3375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4988739678038202e-05,
+      "loss": 0.4893,
+      "step": 3376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4987071482192011e-05,
+      "loss": 0.5026,
+      "step": 3377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4985403286345819e-05,
+      "loss": 0.4574,
+      "step": 3378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4983735090499625e-05,
+      "loss": 0.5162,
+      "step": 3379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4982066894653433e-05,
+      "loss": 0.4765,
+      "step": 3380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4980398698807241e-05,
+      "loss": 0.4753,
+      "step": 3381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.497873050296105e-05,
+      "loss": 0.5148,
+      "step": 3382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4977062307114855e-05,
+      "loss": 0.5098,
+      "step": 3383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4975394111268665e-05,
+      "loss": 0.4553,
+      "step": 3384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4973725915422472e-05,
+      "loss": 0.5011,
+      "step": 3385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4972057719576279e-05,
+      "loss": 0.5028,
+      "step": 3386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4970389523730087e-05,
+      "loss": 0.4644,
+      "step": 3387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4968721327883894e-05,
+      "loss": 0.466,
+      "step": 3388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4967053132037704e-05,
+      "loss": 0.4945,
+      "step": 3389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4965384936191509e-05,
+      "loss": 0.5091,
+      "step": 3390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4963716740345318e-05,
+      "loss": 0.4741,
+      "step": 3391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4962048544499126e-05,
+      "loss": 0.4994,
+      "step": 3392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4960380348652932e-05,
+      "loss": 0.479,
+      "step": 3393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.495871215280674e-05,
+      "loss": 0.4535,
+      "step": 3394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4957043956960548e-05,
+      "loss": 0.4574,
+      "step": 3395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4955375761114358e-05,
+      "loss": 0.5137,
+      "step": 3396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4953707565268164e-05,
+      "loss": 0.5058,
+      "step": 3397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.4952039369421972e-05,
+      "loss": 0.4476,
+      "step": 3398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.495037117357578e-05,
+      "loss": 0.4693,
+      "step": 3399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4948702977729587e-05,
+      "loss": 0.5203,
+      "step": 3400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4947034781883394e-05,
+      "loss": 0.469,
+      "step": 3401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4945366586037201e-05,
+      "loss": 0.4778,
+      "step": 3402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4943698390191011e-05,
+      "loss": 0.4633,
+      "step": 3403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4942030194344817e-05,
+      "loss": 0.5341,
+      "step": 3404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940361998498625e-05,
+      "loss": 0.4869,
+      "step": 3405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4938693802652433e-05,
+      "loss": 0.4608,
+      "step": 3406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937025606806241e-05,
+      "loss": 0.5258,
+      "step": 3407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4935357410960047e-05,
+      "loss": 0.5014,
+      "step": 3408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4933689215113855e-05,
+      "loss": 0.467,
+      "step": 3409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4932021019267665e-05,
+      "loss": 0.4996,
+      "step": 3410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493035282342147e-05,
+      "loss": 0.4736,
+      "step": 3411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4928684627575279e-05,
+      "loss": 0.4905,
+      "step": 3412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4927016431729087e-05,
+      "loss": 0.5047,
+      "step": 3413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4925348235882894e-05,
+      "loss": 0.4896,
+      "step": 3414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49236800400367e-05,
+      "loss": 0.4993,
+      "step": 3415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4922011844190509e-05,
+      "loss": 0.4959,
+      "step": 3416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4920343648344318e-05,
+      "loss": 0.4842,
+      "step": 3417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4918675452498124e-05,
+      "loss": 0.4992,
+      "step": 3418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4917007256651932e-05,
+      "loss": 0.5019,
+      "step": 3419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491533906080574e-05,
+      "loss": 0.478,
+      "step": 3420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4913670864959548e-05,
+      "loss": 0.451,
+      "step": 3421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4912002669113354e-05,
+      "loss": 0.4717,
+      "step": 3422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4910334473267164e-05,
+      "loss": 0.4937,
+      "step": 3423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4908666277420972e-05,
+      "loss": 0.5155,
+      "step": 3424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906998081574778e-05,
+      "loss": 0.472,
+      "step": 3425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4905329885728586e-05,
+      "loss": 0.4774,
+      "step": 3426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4903661689882394e-05,
+      "loss": 0.4806,
+      "step": 3427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4901993494036201e-05,
+      "loss": 0.4496,
+      "step": 3428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4900325298190008e-05,
+      "loss": 0.5016,
+      "step": 3429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4898657102343817e-05,
+      "loss": 0.4701,
+      "step": 3430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4896988906497625e-05,
+      "loss": 0.4698,
+      "step": 3431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4895320710651431e-05,
+      "loss": 0.4717,
+      "step": 3432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489365251480524e-05,
+      "loss": 0.4927,
+      "step": 3433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4891984318959047e-05,
+      "loss": 0.4846,
+      "step": 3434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4890316123112855e-05,
+      "loss": 0.4741,
+      "step": 3435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4888647927266661e-05,
+      "loss": 0.5125,
+      "step": 3436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.488697973142047e-05,
+      "loss": 0.4834,
+      "step": 3437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4885311535574279e-05,
+      "loss": 0.4708,
+      "step": 3438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4883643339728085e-05,
+      "loss": 0.4868,
+      "step": 3439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4881975143881893e-05,
+      "loss": 0.4952,
+      "step": 3440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.48803069480357e-05,
+      "loss": 0.4979,
+      "step": 3441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4878638752189508e-05,
+      "loss": 0.4957,
+      "step": 3442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4876970556343315e-05,
+      "loss": 0.4903,
+      "step": 3443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4875302360497124e-05,
+      "loss": 0.4587,
+      "step": 3444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4873634164650932e-05,
+      "loss": 0.4596,
+      "step": 3445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4871965968804738e-05,
+      "loss": 0.4829,
+      "step": 3446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4870297772958546e-05,
+      "loss": 0.4802,
+      "step": 3447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4868629577112354e-05,
+      "loss": 0.5063,
+      "step": 3448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4866961381266162e-05,
+      "loss": 0.4593,
+      "step": 3449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4865293185419968e-05,
+      "loss": 0.4996,
+      "step": 3450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4863624989573778e-05,
+      "loss": 0.4719,
+      "step": 3451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4861956793727586e-05,
+      "loss": 0.5053,
+      "step": 3452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4860288597881392e-05,
+      "loss": 0.4584,
+      "step": 3453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.48586204020352e-05,
+      "loss": 0.4552,
+      "step": 3454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4856952206189008e-05,
+      "loss": 0.5009,
+      "step": 3455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4855284010342817e-05,
+      "loss": 0.4991,
+      "step": 3456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4853615814496622e-05,
+      "loss": 0.5148,
+      "step": 3457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4851947618650431e-05,
+      "loss": 0.4896,
+      "step": 3458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.485027942280424e-05,
+      "loss": 0.4559,
+      "step": 3459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4848611226958045e-05,
+      "loss": 0.4936,
+      "step": 3460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4846943031111853e-05,
+      "loss": 0.5259,
+      "step": 3461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4845274835265661e-05,
+      "loss": 0.4935,
+      "step": 3462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.484360663941947e-05,
+      "loss": 0.4267,
+      "step": 3463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4841938443573275e-05,
+      "loss": 0.4733,
+      "step": 3464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4840270247727085e-05,
+      "loss": 0.4979,
+      "step": 3465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4838602051880893e-05,
+      "loss": 0.4861,
+      "step": 3466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4836933856034699e-05,
+      "loss": 0.4819,
+      "step": 3467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4835265660188507e-05,
+      "loss": 0.4933,
+      "step": 3468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4833597464342315e-05,
+      "loss": 0.4823,
+      "step": 3469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4831929268496124e-05,
+      "loss": 0.4715,
+      "step": 3470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.483026107264993e-05,
+      "loss": 0.4652,
+      "step": 3471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4828592876803738e-05,
+      "loss": 0.5016,
+      "step": 3472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4826924680957546e-05,
+      "loss": 0.4906,
+      "step": 3473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4825256485111352e-05,
+      "loss": 0.5,
+      "step": 3474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.482358828926516e-05,
+      "loss": 0.4976,
+      "step": 3475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4821920093418968e-05,
+      "loss": 0.5026,
+      "step": 3476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4820251897572778e-05,
+      "loss": 0.4747,
+      "step": 3477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4818583701726584e-05,
+      "loss": 0.52,
+      "step": 3478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4816915505880392e-05,
+      "loss": 0.4799,
+      "step": 3479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.48152473100342e-05,
+      "loss": 0.4895,
+      "step": 3480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4813579114188006e-05,
+      "loss": 0.4904,
+      "step": 3481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4811910918341814e-05,
+      "loss": 0.4772,
+      "step": 3482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4810242722495622e-05,
+      "loss": 0.499,
+      "step": 3483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4808574526649431e-05,
+      "loss": 0.464,
+      "step": 3484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4806906330803237e-05,
+      "loss": 0.4799,
+      "step": 3485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4805238134957045e-05,
+      "loss": 0.5,
+      "step": 3486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4803569939110853e-05,
+      "loss": 0.5442,
+      "step": 3487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.480190174326466e-05,
+      "loss": 0.494,
+      "step": 3488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4800233547418467e-05,
+      "loss": 0.4509,
+      "step": 3489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4798565351572275e-05,
+      "loss": 0.4716,
+      "step": 3490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4796897155726085e-05,
+      "loss": 0.5195,
+      "step": 3491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4795228959879891e-05,
+      "loss": 0.4821,
+      "step": 3492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4793560764033699e-05,
+      "loss": 0.5373,
+      "step": 3493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4791892568187507e-05,
+      "loss": 0.5086,
+      "step": 3494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4790224372341313e-05,
+      "loss": 0.5025,
+      "step": 3495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.478855617649512e-05,
+      "loss": 0.5095,
+      "step": 3496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.478688798064893e-05,
+      "loss": 0.5119,
+      "step": 3497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4785219784802738e-05,
+      "loss": 0.495,
+      "step": 3498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4783551588956545e-05,
+      "loss": 0.509,
+      "step": 3499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4781883393110352e-05,
+      "loss": 0.4967,
+      "step": 3500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.478021519726416e-05,
+      "loss": 0.5167,
+      "step": 3501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4778547001417966e-05,
+      "loss": 0.5071,
+      "step": 3502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4776878805571774e-05,
+      "loss": 0.4844,
+      "step": 3503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4775210609725584e-05,
+      "loss": 0.4913,
+      "step": 3504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4773542413879392e-05,
+      "loss": 0.4737,
+      "step": 3505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4771874218033198e-05,
+      "loss": 0.47,
+      "step": 3506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4770206022187006e-05,
+      "loss": 0.5087,
+      "step": 3507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4768537826340814e-05,
+      "loss": 0.4853,
+      "step": 3508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.476686963049462e-05,
+      "loss": 0.4696,
+      "step": 3509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4765201434648428e-05,
+      "loss": 0.4759,
+      "step": 3510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4763533238802237e-05,
+      "loss": 0.4771,
+      "step": 3511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4761865042956045e-05,
+      "loss": 0.4831,
+      "step": 3512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4760196847109852e-05,
+      "loss": 0.4729,
+      "step": 3513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.475852865126366e-05,
+      "loss": 0.5037,
+      "step": 3514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4756860455417467e-05,
+      "loss": 0.5033,
+      "step": 3515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4755192259571274e-05,
+      "loss": 0.4738,
+      "step": 3516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4753524063725081e-05,
+      "loss": 0.5223,
+      "step": 3517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4751855867878891e-05,
+      "loss": 0.4562,
+      "step": 3518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4750187672032699e-05,
+      "loss": 0.4941,
+      "step": 3519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4748519476186505e-05,
+      "loss": 0.4738,
+      "step": 3520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.4746851280340313e-05,
+      "loss": 0.5007,
+      "step": 3521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.474518308449412e-05,
+      "loss": 0.465,
+      "step": 3522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.474351488864793e-05,
+      "loss": 0.4796,
+      "step": 3523
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4741846692801735e-05,
+      "loss": 0.4674,
+      "step": 3524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4740178496955544e-05,
+      "loss": 0.4688,
+      "step": 3525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4738510301109352e-05,
+      "loss": 0.503,
+      "step": 3526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.4745,
+      "step": 3527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4735173909416966e-05,
+      "loss": 0.4651,
+      "step": 3528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4733505713570774e-05,
+      "loss": 0.4746,
+      "step": 3529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4731837517724584e-05,
+      "loss": 0.4502,
+      "step": 3530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4730169321878388e-05,
+      "loss": 0.5306,
+      "step": 3531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4728501126032198e-05,
+      "loss": 0.4817,
+      "step": 3532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4726832930186006e-05,
+      "loss": 0.4836,
+      "step": 3533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4725164734339812e-05,
+      "loss": 0.4914,
+      "step": 3534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.472349653849362e-05,
+      "loss": 0.4839,
+      "step": 3535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4721828342647428e-05,
+      "loss": 0.4697,
+      "step": 3536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4720160146801237e-05,
+      "loss": 0.4693,
+      "step": 3537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4718491950955044e-05,
+      "loss": 0.4841,
+      "step": 3538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4716823755108852e-05,
+      "loss": 0.5399,
+      "step": 3539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.471515555926266e-05,
+      "loss": 0.4681,
+      "step": 3540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4713487363416466e-05,
+      "loss": 0.437,
+      "step": 3541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4711819167570273e-05,
+      "loss": 0.4695,
+      "step": 3542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4710150971724081e-05,
+      "loss": 0.49,
+      "step": 3543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4708482775877891e-05,
+      "loss": 0.413,
+      "step": 3544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4706814580031697e-05,
+      "loss": 0.5045,
+      "step": 3545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4705146384185505e-05,
+      "loss": 0.4798,
+      "step": 3546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4703478188339313e-05,
+      "loss": 0.457,
+      "step": 3547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4701809992493119e-05,
+      "loss": 0.4861,
+      "step": 3548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4700141796646927e-05,
+      "loss": 0.4985,
+      "step": 3549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4698473600800735e-05,
+      "loss": 0.4524,
+      "step": 3550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4696805404954544e-05,
+      "loss": 0.4789,
+      "step": 3551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.469513720910835e-05,
+      "loss": 0.5019,
+      "step": 3552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4693469013262159e-05,
+      "loss": 0.4573,
+      "step": 3553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4691800817415966e-05,
+      "loss": 0.504,
+      "step": 3554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4690132621569773e-05,
+      "loss": 0.4683,
+      "step": 3555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.468846442572358e-05,
+      "loss": 0.4498,
+      "step": 3556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4686796229877388e-05,
+      "loss": 0.5041,
+      "step": 3557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4685128034031198e-05,
+      "loss": 0.4823,
+      "step": 3558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4683459838185004e-05,
+      "loss": 0.4655,
+      "step": 3559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4681791642338812e-05,
+      "loss": 0.4991,
+      "step": 3560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.468012344649262e-05,
+      "loss": 0.4704,
+      "step": 3561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4678455250646426e-05,
+      "loss": 0.4839,
+      "step": 3562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4676787054800234e-05,
+      "loss": 0.512,
+      "step": 3563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.4675118858954042e-05,
+      "loss": 0.4853,
+      "step": 3564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4673450663107851e-05,
+      "loss": 0.442,
+      "step": 3565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4671782467261658e-05,
+      "loss": 0.4584,
+      "step": 3566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4670114271415466e-05,
+      "loss": 0.478,
+      "step": 3567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4668446075569273e-05,
+      "loss": 0.4733,
+      "step": 3568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.466677787972308e-05,
+      "loss": 0.4609,
+      "step": 3569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4665109683876888e-05,
+      "loss": 0.491,
+      "step": 3570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4663441488030697e-05,
+      "loss": 0.4936,
+      "step": 3571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4661773292184505e-05,
+      "loss": 0.5003,
+      "step": 3572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4660105096338311e-05,
+      "loss": 0.4504,
+      "step": 3573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4658436900492119e-05,
+      "loss": 0.482,
+      "step": 3574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4656768704645927e-05,
+      "loss": 0.4898,
+      "step": 3575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4655100508799733e-05,
+      "loss": 0.4659,
+      "step": 3576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4653432312953541e-05,
+      "loss": 0.512,
+      "step": 3577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.465176411710735e-05,
+      "loss": 0.4817,
+      "step": 3578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4650095921261159e-05,
+      "loss": 0.4548,
+      "step": 3579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4648427725414965e-05,
+      "loss": 0.5008,
+      "step": 3580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4646759529568773e-05,
+      "loss": 0.4932,
+      "step": 3581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.464509133372258e-05,
+      "loss": 0.51,
+      "step": 3582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4643423137876387e-05,
+      "loss": 0.4621,
+      "step": 3583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4641754942030195e-05,
+      "loss": 0.4771,
+      "step": 3584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4640086746184004e-05,
+      "loss": 0.4638,
+      "step": 3585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4638418550337812e-05,
+      "loss": 0.4432,
+      "step": 3586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4636750354491618e-05,
+      "loss": 0.5198,
+      "step": 3587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4635082158645426e-05,
+      "loss": 0.4705,
+      "step": 3588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4633413962799234e-05,
+      "loss": 0.4931,
+      "step": 3589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.463174576695304e-05,
+      "loss": 0.4873,
+      "step": 3590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4630077571106848e-05,
+      "loss": 0.4893,
+      "step": 3591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4628409375260658e-05,
+      "loss": 0.4822,
+      "step": 3592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4626741179414466e-05,
+      "loss": 0.5154,
+      "step": 3593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4625072983568272e-05,
+      "loss": 0.4852,
+      "step": 3594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.462340478772208e-05,
+      "loss": 0.4873,
+      "step": 3595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4621736591875888e-05,
+      "loss": 0.4956,
+      "step": 3596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4620068396029694e-05,
+      "loss": 0.5371,
+      "step": 3597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4618400200183502e-05,
+      "loss": 0.4863,
+      "step": 3598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4616732004337311e-05,
+      "loss": 0.4749,
+      "step": 3599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4615063808491119e-05,
+      "loss": 0.4517,
+      "step": 3600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4613395612644925e-05,
+      "loss": 0.474,
+      "step": 3601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4611727416798733e-05,
+      "loss": 0.5314,
+      "step": 3602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4610059220952541e-05,
+      "loss": 0.4513,
+      "step": 3603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4608391025106347e-05,
+      "loss": 0.5275,
+      "step": 3604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.4606722829260155e-05,
+      "loss": 0.4804,
+      "step": 3605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4605054633413965e-05,
+      "loss": 0.5106,
+      "step": 3606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4603386437567773e-05,
+      "loss": 0.4782,
+      "step": 3607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4601718241721579e-05,
+      "loss": 0.5664,
+      "step": 3608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4600050045875387e-05,
+      "loss": 0.4651,
+      "step": 3609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4598381850029195e-05,
+      "loss": 0.4918,
+      "step": 3610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4596713654183e-05,
+      "loss": 0.4896,
+      "step": 3611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.459504545833681e-05,
+      "loss": 0.4742,
+      "step": 3612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4593377262490618e-05,
+      "loss": 0.4683,
+      "step": 3613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4591709066644426e-05,
+      "loss": 0.4901,
+      "step": 3614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4590040870798232e-05,
+      "loss": 0.5078,
+      "step": 3615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.458837267495204e-05,
+      "loss": 0.4907,
+      "step": 3616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4586704479105848e-05,
+      "loss": 0.5103,
+      "step": 3617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4585036283259654e-05,
+      "loss": 0.4844,
+      "step": 3618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4583368087413464e-05,
+      "loss": 0.4959,
+      "step": 3619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4581699891567272e-05,
+      "loss": 0.4846,
+      "step": 3620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.458003169572108e-05,
+      "loss": 0.4839,
+      "step": 3621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4578363499874886e-05,
+      "loss": 0.484,
+      "step": 3622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4576695304028694e-05,
+      "loss": 0.4711,
+      "step": 3623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4575027108182502e-05,
+      "loss": 0.484,
+      "step": 3624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4573358912336308e-05,
+      "loss": 0.5051,
+      "step": 3625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4571690716490117e-05,
+      "loss": 0.4632,
+      "step": 3626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4570022520643925e-05,
+      "loss": 0.49,
+      "step": 3627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4568354324797733e-05,
+      "loss": 0.4909,
+      "step": 3628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.456668612895154e-05,
+      "loss": 0.4876,
+      "step": 3629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4565017933105347e-05,
+      "loss": 0.5128,
+      "step": 3630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4563349737259155e-05,
+      "loss": 0.4746,
+      "step": 3631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4561681541412961e-05,
+      "loss": 0.4827,
+      "step": 3632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4560013345566771e-05,
+      "loss": 0.5254,
+      "step": 3633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4558345149720579e-05,
+      "loss": 0.4956,
+      "step": 3634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4556676953874387e-05,
+      "loss": 0.4691,
+      "step": 3635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4555008758028193e-05,
+      "loss": 0.4714,
+      "step": 3636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4553340562182e-05,
+      "loss": 0.4903,
+      "step": 3637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.455167236633581e-05,
+      "loss": 0.4648,
+      "step": 3638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4550004170489618e-05,
+      "loss": 0.4873,
+      "step": 3639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4548335974643424e-05,
+      "loss": 0.4995,
+      "step": 3640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4546667778797232e-05,
+      "loss": 0.4659,
+      "step": 3641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.454499958295104e-05,
+      "loss": 0.4661,
+      "step": 3642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4543331387104846e-05,
+      "loss": 0.5073,
+      "step": 3643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4541663191258654e-05,
+      "loss": 0.4546,
+      "step": 3644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4539994995412464e-05,
+      "loss": 0.5123,
+      "step": 3645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.4538326799566272e-05,
+      "loss": 0.5171,
+      "step": 3646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4536658603720078e-05,
+      "loss": 0.4729,
+      "step": 3647
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4534990407873886e-05,
+      "loss": 0.4745,
+      "step": 3648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4533322212027694e-05,
+      "loss": 0.4835,
+      "step": 3649
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.45316540161815e-05,
+      "loss": 0.4879,
+      "step": 3650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4529985820335308e-05,
+      "loss": 0.5206,
+      "step": 3651
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4528317624489117e-05,
+      "loss": 0.4439,
+      "step": 3652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4526649428642925e-05,
+      "loss": 0.4661,
+      "step": 3653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4524981232796731e-05,
+      "loss": 0.457,
+      "step": 3654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.452331303695054e-05,
+      "loss": 0.4746,
+      "step": 3655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4521644841104347e-05,
+      "loss": 0.4686,
+      "step": 3656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4519976645258153e-05,
+      "loss": 0.5411,
+      "step": 3657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4518308449411961e-05,
+      "loss": 0.475,
+      "step": 3658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4516640253565771e-05,
+      "loss": 0.4854,
+      "step": 3659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4514972057719579e-05,
+      "loss": 0.4888,
+      "step": 3660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4513303861873385e-05,
+      "loss": 0.4541,
+      "step": 3661
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4511635666027193e-05,
+      "loss": 0.5128,
+      "step": 3662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4509967470181e-05,
+      "loss": 0.5174,
+      "step": 3663
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4508299274334807e-05,
+      "loss": 0.4905,
+      "step": 3664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4506631078488615e-05,
+      "loss": 0.4641,
+      "step": 3665
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4504962882642424e-05,
+      "loss": 0.4916,
+      "step": 3666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4503294686796232e-05,
+      "loss": 0.4884,
+      "step": 3667
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4501626490950038e-05,
+      "loss": 0.5317,
+      "step": 3668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4499958295103846e-05,
+      "loss": 0.4866,
+      "step": 3669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4498290099257654e-05,
+      "loss": 0.4907,
+      "step": 3670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.449662190341146e-05,
+      "loss": 0.5132,
+      "step": 3671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4494953707565268e-05,
+      "loss": 0.4985,
+      "step": 3672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4493285511719078e-05,
+      "loss": 0.4902,
+      "step": 3673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4491617315872886e-05,
+      "loss": 0.48,
+      "step": 3674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4489949120026692e-05,
+      "loss": 0.4847,
+      "step": 3675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.44882809241805e-05,
+      "loss": 0.4922,
+      "step": 3676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4486612728334308e-05,
+      "loss": 0.4805,
+      "step": 3677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4484944532488114e-05,
+      "loss": 0.4588,
+      "step": 3678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4483276336641924e-05,
+      "loss": 0.4567,
+      "step": 3679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4481608140795731e-05,
+      "loss": 0.4765,
+      "step": 3680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.447993994494954e-05,
+      "loss": 0.4688,
+      "step": 3681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4478271749103345e-05,
+      "loss": 0.4911,
+      "step": 3682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4476603553257153e-05,
+      "loss": 0.4922,
+      "step": 3683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4474935357410961e-05,
+      "loss": 0.4919,
+      "step": 3684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4473267161564767e-05,
+      "loss": 0.5015,
+      "step": 3685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4471598965718577e-05,
+      "loss": 0.4775,
+      "step": 3686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4469930769872385e-05,
+      "loss": 0.4864,
+      "step": 3687
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4468262574026193e-05,
+      "loss": 0.5133,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4466594378179999e-05,
+      "loss": 0.4866,
+      "step": 3689
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4464926182333807e-05,
+      "loss": 0.5089,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4463257986487615e-05,
+      "loss": 0.5438,
+      "step": 3691
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4461589790641421e-05,
+      "loss": 0.4583,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.445992159479523e-05,
+      "loss": 0.4524,
+      "step": 3693
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4458253398949038e-05,
+      "loss": 0.4477,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4456585203102846e-05,
+      "loss": 0.4577,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4454917007256653e-05,
+      "loss": 0.485,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.445324881141046e-05,
+      "loss": 0.4757,
+      "step": 3697
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4451580615564268e-05,
+      "loss": 0.5109,
+      "step": 3698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4449912419718074e-05,
+      "loss": 0.4609,
+      "step": 3699
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4448244223871884e-05,
+      "loss": 0.5047,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4446576028025692e-05,
+      "loss": 0.4391,
+      "step": 3701
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.44449078321795e-05,
+      "loss": 0.5166,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4443239636333306e-05,
+      "loss": 0.499,
+      "step": 3703
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4441571440487114e-05,
+      "loss": 0.4991,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4439903244640922e-05,
+      "loss": 0.4845,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4438235048794728e-05,
+      "loss": 0.4878,
+      "step": 3706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4436566852948538e-05,
+      "loss": 0.4671,
+      "step": 3707
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4434898657102345e-05,
+      "loss": 0.469,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4433230461256153e-05,
+      "loss": 0.5158,
+      "step": 3709
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.443156226540996e-05,
+      "loss": 0.4748,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4429894069563767e-05,
+      "loss": 0.4988,
+      "step": 3711
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4428225873717577e-05,
+      "loss": 0.473,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4426557677871382e-05,
+      "loss": 0.4941,
+      "step": 3713
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4424889482025191e-05,
+      "loss": 0.5006,
+      "step": 3714
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4423221286178999e-05,
+      "loss": 0.4886,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4421553090332807e-05,
+      "loss": 0.4899,
+      "step": 3716
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4419884894486613e-05,
+      "loss": 0.4684,
+      "step": 3717
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4418216698640421e-05,
+      "loss": 0.4403,
+      "step": 3718
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.441654850279423e-05,
+      "loss": 0.4745,
+      "step": 3719
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4414880306948035e-05,
+      "loss": 0.5194,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4413212111101845e-05,
+      "loss": 0.5086,
+      "step": 3721
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4411543915255652e-05,
+      "loss": 0.4527,
+      "step": 3722
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.440987571940946e-05,
+      "loss": 0.4888,
+      "step": 3723
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4408207523563267e-05,
+      "loss": 0.4779,
+      "step": 3724
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4406539327717074e-05,
+      "loss": 0.5136,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4404871131870884e-05,
+      "loss": 0.4902,
+      "step": 3726
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.440320293602469e-05,
+      "loss": 0.486,
+      "step": 3727
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4401534740178498e-05,
+      "loss": 0.4766,
+      "step": 3728
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.4399866544332306e-05,
+      "loss": 0.4652,
+      "step": 3729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4398198348486114e-05,
+      "loss": 0.5249,
+      "step": 3730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.439653015263992e-05,
+      "loss": 0.5133,
+      "step": 3731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4394861956793728e-05,
+      "loss": 0.5224,
+      "step": 3732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4393193760947538e-05,
+      "loss": 0.4879,
+      "step": 3733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4391525565101344e-05,
+      "loss": 0.5107,
+      "step": 3734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4389857369255152e-05,
+      "loss": 0.5033,
+      "step": 3735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.438818917340896e-05,
+      "loss": 0.4996,
+      "step": 3736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4386520977562767e-05,
+      "loss": 0.5173,
+      "step": 3737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4384852781716574e-05,
+      "loss": 0.5189,
+      "step": 3738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4383184585870381e-05,
+      "loss": 0.4911,
+      "step": 3739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4381516390024191e-05,
+      "loss": 0.4787,
+      "step": 3740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4379848194177997e-05,
+      "loss": 0.4888,
+      "step": 3741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4378179998331805e-05,
+      "loss": 0.4838,
+      "step": 3742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4376511802485613e-05,
+      "loss": 0.4585,
+      "step": 3743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4374843606639421e-05,
+      "loss": 0.4753,
+      "step": 3744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4373175410793227e-05,
+      "loss": 0.4977,
+      "step": 3745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4371507214947035e-05,
+      "loss": 0.4957,
+      "step": 3746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4369839019100845e-05,
+      "loss": 0.5005,
+      "step": 3747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.436817082325465e-05,
+      "loss": 0.4765,
+      "step": 3748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4366502627408459e-05,
+      "loss": 0.4633,
+      "step": 3749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4364834431562267e-05,
+      "loss": 0.4635,
+      "step": 3750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4363166235716074e-05,
+      "loss": 0.4741,
+      "step": 3751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.436149803986988e-05,
+      "loss": 0.5138,
+      "step": 3752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.435982984402369e-05,
+      "loss": 0.4839,
+      "step": 3753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4358161648177498e-05,
+      "loss": 0.4815,
+      "step": 3754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4356493452331304e-05,
+      "loss": 0.4893,
+      "step": 3755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4354825256485112e-05,
+      "loss": 0.4917,
+      "step": 3756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.435315706063892e-05,
+      "loss": 0.5019,
+      "step": 3757
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4351488864792728e-05,
+      "loss": 0.4674,
+      "step": 3758
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4349820668946534e-05,
+      "loss": 0.5079,
+      "step": 3759
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4348152473100344e-05,
+      "loss": 0.4841,
+      "step": 3760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4346484277254152e-05,
+      "loss": 0.4874,
+      "step": 3761
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.434481608140796e-05,
+      "loss": 0.4897,
+      "step": 3762
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4343147885561766e-05,
+      "loss": 0.4623,
+      "step": 3763
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4341479689715574e-05,
+      "loss": 0.4984,
+      "step": 3764
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4339811493869381e-05,
+      "loss": 0.4698,
+      "step": 3765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4338143298023188e-05,
+      "loss": 0.4728,
+      "step": 3766
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4336475102176997e-05,
+      "loss": 0.4802,
+      "step": 3767
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4334806906330805e-05,
+      "loss": 0.51,
+      "step": 3768
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.4333138710484613e-05,
+      "loss": 0.4833,
+      "step": 3769
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.433147051463842e-05,
+      "loss": 0.4606,
+      "step": 3770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4329802318792227e-05,
+      "loss": 0.5117,
+      "step": 3771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4328134122946035e-05,
+      "loss": 0.4779,
+      "step": 3772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4326465927099841e-05,
+      "loss": 0.4463,
+      "step": 3773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.432479773125365e-05,
+      "loss": 0.5069,
+      "step": 3774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4323129535407459e-05,
+      "loss": 0.4773,
+      "step": 3775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4321461339561267e-05,
+      "loss": 0.5109,
+      "step": 3776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4319793143715073e-05,
+      "loss": 0.4904,
+      "step": 3777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.431812494786888e-05,
+      "loss": 0.5241,
+      "step": 3778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.431645675202269e-05,
+      "loss": 0.4587,
+      "step": 3779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4314788556176495e-05,
+      "loss": 0.4776,
+      "step": 3780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4313120360330304e-05,
+      "loss": 0.4608,
+      "step": 3781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4311452164484112e-05,
+      "loss": 0.5019,
+      "step": 3782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.430978396863792e-05,
+      "loss": 0.4929,
+      "step": 3783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4308115772791726e-05,
+      "loss": 0.4889,
+      "step": 3784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4306447576945534e-05,
+      "loss": 0.4935,
+      "step": 3785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4304779381099344e-05,
+      "loss": 0.4358,
+      "step": 3786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4303111185253148e-05,
+      "loss": 0.4803,
+      "step": 3787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4301442989406958e-05,
+      "loss": 0.4582,
+      "step": 3788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4299774793560766e-05,
+      "loss": 0.4678,
+      "step": 3789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4298106597714574e-05,
+      "loss": 0.4717,
+      "step": 3790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.429643840186838e-05,
+      "loss": 0.4872,
+      "step": 3791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4294770206022188e-05,
+      "loss": 0.4556,
+      "step": 3792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4293102010175997e-05,
+      "loss": 0.5123,
+      "step": 3793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4291433814329803e-05,
+      "loss": 0.4615,
+      "step": 3794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4289765618483611e-05,
+      "loss": 0.4279,
+      "step": 3795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.428809742263742e-05,
+      "loss": 0.4722,
+      "step": 3796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4286429226791227e-05,
+      "loss": 0.5435,
+      "step": 3797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4284761030945033e-05,
+      "loss": 0.511,
+      "step": 3798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4283092835098841e-05,
+      "loss": 0.4613,
+      "step": 3799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.428142463925265e-05,
+      "loss": 0.4838,
+      "step": 3800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4279756443406457e-05,
+      "loss": 0.5109,
+      "step": 3801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4278088247560265e-05,
+      "loss": 0.4433,
+      "step": 3802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4276420051714073e-05,
+      "loss": 0.5287,
+      "step": 3803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.427475185586788e-05,
+      "loss": 0.4798,
+      "step": 3804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4273083660021687e-05,
+      "loss": 0.4802,
+      "step": 3805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4271415464175495e-05,
+      "loss": 0.4711,
+      "step": 3806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4269747268329304e-05,
+      "loss": 0.4649,
+      "step": 3807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.426807907248311e-05,
+      "loss": 0.481,
+      "step": 3808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4266410876636918e-05,
+      "loss": 0.4521,
+      "step": 3809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4264742680790726e-05,
+      "loss": 0.4764,
+      "step": 3810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4263074484944534e-05,
+      "loss": 0.4997,
+      "step": 3811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.426140628909834e-05,
+      "loss": 0.5007,
+      "step": 3812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4259738093252148e-05,
+      "loss": 0.4792,
+      "step": 3813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4258069897405958e-05,
+      "loss": 0.4946,
+      "step": 3814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4256401701559764e-05,
+      "loss": 0.513,
+      "step": 3815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4254733505713572e-05,
+      "loss": 0.4597,
+      "step": 3816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.425306530986738e-05,
+      "loss": 0.46,
+      "step": 3817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4251397114021188e-05,
+      "loss": 0.4791,
+      "step": 3818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4249728918174994e-05,
+      "loss": 0.4915,
+      "step": 3819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4248060722328802e-05,
+      "loss": 0.4983,
+      "step": 3820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4246392526482611e-05,
+      "loss": 0.4496,
+      "step": 3821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4244724330636417e-05,
+      "loss": 0.4934,
+      "step": 3822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4243056134790225e-05,
+      "loss": 0.4829,
+      "step": 3823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4241387938944033e-05,
+      "loss": 0.4828,
+      "step": 3824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4239719743097841e-05,
+      "loss": 0.4814,
+      "step": 3825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4238051547251647e-05,
+      "loss": 0.512,
+      "step": 3826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4236383351405457e-05,
+      "loss": 0.5038,
+      "step": 3827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4234715155559265e-05,
+      "loss": 0.4945,
+      "step": 3828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4233046959713071e-05,
+      "loss": 0.4739,
+      "step": 3829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4231378763866879e-05,
+      "loss": 0.4573,
+      "step": 3830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4229710568020687e-05,
+      "loss": 0.4395,
+      "step": 3831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4228042372174495e-05,
+      "loss": 0.4447,
+      "step": 3832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4226374176328301e-05,
+      "loss": 0.4735,
+      "step": 3833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.422470598048211e-05,
+      "loss": 0.4938,
+      "step": 3834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4223037784635918e-05,
+      "loss": 0.4809,
+      "step": 3835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4221369588789725e-05,
+      "loss": 0.5077,
+      "step": 3836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4219701392943532e-05,
+      "loss": 0.4915,
+      "step": 3837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.421803319709734e-05,
+      "loss": 0.5104,
+      "step": 3838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4216365001251148e-05,
+      "loss": 0.466,
+      "step": 3839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4214696805404954e-05,
+      "loss": 0.4921,
+      "step": 3840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4213028609558764e-05,
+      "loss": 0.4807,
+      "step": 3841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4211360413712572e-05,
+      "loss": 0.4706,
+      "step": 3842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4209692217866378e-05,
+      "loss": 0.4927,
+      "step": 3843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4208024022020186e-05,
+      "loss": 0.4642,
+      "step": 3844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4206355826173994e-05,
+      "loss": 0.4458,
+      "step": 3845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4204687630327802e-05,
+      "loss": 0.4665,
+      "step": 3846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4203019434481608e-05,
+      "loss": 0.4717,
+      "step": 3847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4201351238635417e-05,
+      "loss": 0.4908,
+      "step": 3848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4199683042789225e-05,
+      "loss": 0.482,
+      "step": 3849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4198014846943032e-05,
+      "loss": 0.4889,
+      "step": 3850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.419634665109684e-05,
+      "loss": 0.5088,
+      "step": 3851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4194678455250647e-05,
+      "loss": 0.4608,
+      "step": 3852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4193010259404457e-05,
+      "loss": 0.4645,
+      "step": 3853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4191342063558261e-05,
+      "loss": 0.4685,
+      "step": 3854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4189673867712071e-05,
+      "loss": 0.465,
+      "step": 3855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4188005671865879e-05,
+      "loss": 0.4899,
+      "step": 3856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4186337476019685e-05,
+      "loss": 0.4615,
+      "step": 3857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4184669280173493e-05,
+      "loss": 0.5416,
+      "step": 3858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.41830010843273e-05,
+      "loss": 0.4663,
+      "step": 3859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.418133288848111e-05,
+      "loss": 0.467,
+      "step": 3860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4179664692634915e-05,
+      "loss": 0.4954,
+      "step": 3861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4177996496788724e-05,
+      "loss": 0.4678,
+      "step": 3862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4176328300942532e-05,
+      "loss": 0.5426,
+      "step": 3863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4174660105096339e-05,
+      "loss": 0.4851,
+      "step": 3864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4172991909250146e-05,
+      "loss": 0.5195,
+      "step": 3865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4171323713403954e-05,
+      "loss": 0.5123,
+      "step": 3866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4169655517557764e-05,
+      "loss": 0.4857,
+      "step": 3867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.416798732171157e-05,
+      "loss": 0.4697,
+      "step": 3868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4166319125865378e-05,
+      "loss": 0.4774,
+      "step": 3869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4164650930019186e-05,
+      "loss": 0.4536,
+      "step": 3870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4162982734172992e-05,
+      "loss": 0.4868,
+      "step": 3871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.41613145383268e-05,
+      "loss": 0.5081,
+      "step": 3872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4159646342480608e-05,
+      "loss": 0.4615,
+      "step": 3873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4157978146634417e-05,
+      "loss": 0.4927,
+      "step": 3874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4156309950788224e-05,
+      "loss": 0.4637,
+      "step": 3875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4154641754942032e-05,
+      "loss": 0.4515,
+      "step": 3876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.415297355909584e-05,
+      "loss": 0.4655,
+      "step": 3877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4151305363249647e-05,
+      "loss": 0.5037,
+      "step": 3878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4149637167403453e-05,
+      "loss": 0.4768,
+      "step": 3879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4147968971557261e-05,
+      "loss": 0.4791,
+      "step": 3880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4146300775711071e-05,
+      "loss": 0.4646,
+      "step": 3881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4144632579864877e-05,
+      "loss": 0.453,
+      "step": 3882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4142964384018685e-05,
+      "loss": 0.5008,
+      "step": 3883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4141296188172493e-05,
+      "loss": 0.53,
+      "step": 3884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.41396279923263e-05,
+      "loss": 0.4848,
+      "step": 3885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4137959796480107e-05,
+      "loss": 0.445,
+      "step": 3886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4136291600633915e-05,
+      "loss": 0.5073,
+      "step": 3887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4134623404787724e-05,
+      "loss": 0.4845,
+      "step": 3888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.413295520894153e-05,
+      "loss": 0.5149,
+      "step": 3889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4131287013095339e-05,
+      "loss": 0.5224,
+      "step": 3890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4129618817249146e-05,
+      "loss": 0.4656,
+      "step": 3891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.4127950621402954e-05,
+      "loss": 0.4822,
+      "step": 3892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.412628242555676e-05,
+      "loss": 0.4487,
+      "step": 3893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.412461422971057e-05,
+      "loss": 0.4795,
+      "step": 3894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4122946033864378e-05,
+      "loss": 0.5113,
+      "step": 3895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4121277838018184e-05,
+      "loss": 0.4874,
+      "step": 3896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4119609642171992e-05,
+      "loss": 0.4714,
+      "step": 3897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.41179414463258e-05,
+      "loss": 0.4963,
+      "step": 3898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4116273250479608e-05,
+      "loss": 0.5063,
+      "step": 3899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4114605054633414e-05,
+      "loss": 0.4694,
+      "step": 3900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4112936858787224e-05,
+      "loss": 0.4565,
+      "step": 3901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4111268662941032e-05,
+      "loss": 0.5028,
+      "step": 3902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4109600467094838e-05,
+      "loss": 0.4677,
+      "step": 3903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4107932271248646e-05,
+      "loss": 0.4948,
+      "step": 3904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4106264075402453e-05,
+      "loss": 0.4889,
+      "step": 3905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4104595879556261e-05,
+      "loss": 0.4878,
+      "step": 3906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4102927683710068e-05,
+      "loss": 0.4526,
+      "step": 3907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4101259487863877e-05,
+      "loss": 0.4719,
+      "step": 3908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4099591292017685e-05,
+      "loss": 0.488,
+      "step": 3909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4097923096171491e-05,
+      "loss": 0.5092,
+      "step": 3910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4096254900325299e-05,
+      "loss": 0.5115,
+      "step": 3911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4094586704479107e-05,
+      "loss": 0.489,
+      "step": 3912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4092918508632915e-05,
+      "loss": 0.4758,
+      "step": 3913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4091250312786721e-05,
+      "loss": 0.5244,
+      "step": 3914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.408958211694053e-05,
+      "loss": 0.483,
+      "step": 3915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4087913921094339e-05,
+      "loss": 0.453,
+      "step": 3916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4086245725248145e-05,
+      "loss": 0.5174,
+      "step": 3917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4084577529401953e-05,
+      "loss": 0.527,
+      "step": 3918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.408290933355576e-05,
+      "loss": 0.4675,
+      "step": 3919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4081241137709568e-05,
+      "loss": 0.4546,
+      "step": 3920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4079572941863375e-05,
+      "loss": 0.5344,
+      "step": 3921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4077904746017184e-05,
+      "loss": 0.498,
+      "step": 3922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4076236550170992e-05,
+      "loss": 0.503,
+      "step": 3923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4074568354324798e-05,
+      "loss": 0.5128,
+      "step": 3924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4072900158478606e-05,
+      "loss": 0.4469,
+      "step": 3925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4071231962632414e-05,
+      "loss": 0.4835,
+      "step": 3926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4069563766786224e-05,
+      "loss": 0.4643,
+      "step": 3927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4067895570940028e-05,
+      "loss": 0.5074,
+      "step": 3928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4066227375093838e-05,
+      "loss": 0.4467,
+      "step": 3929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4064559179247646e-05,
+      "loss": 0.4929,
+      "step": 3930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4062890983401452e-05,
+      "loss": 0.4779,
+      "step": 3931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.406122278755526e-05,
+      "loss": 0.4598,
+      "step": 3932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4059554591709068e-05,
+      "loss": 0.4804,
+      "step": 3933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4057886395862877e-05,
+      "loss": 0.5135,
+      "step": 3934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4056218200016683e-05,
+      "loss": 0.4975,
+      "step": 3935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4054550004170491e-05,
+      "loss": 0.499,
+      "step": 3936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4052881808324299e-05,
+      "loss": 0.4824,
+      "step": 3937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4051213612478105e-05,
+      "loss": 0.5222,
+      "step": 3938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4049545416631913e-05,
+      "loss": 0.4724,
+      "step": 3939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4047877220785721e-05,
+      "loss": 0.4508,
+      "step": 3940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.404620902493953e-05,
+      "loss": 0.4824,
+      "step": 3941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4044540829093337e-05,
+      "loss": 0.4915,
+      "step": 3942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4042872633247145e-05,
+      "loss": 0.5138,
+      "step": 3943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4041204437400953e-05,
+      "loss": 0.4761,
+      "step": 3944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4039536241554759e-05,
+      "loss": 0.4836,
+      "step": 3945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4037868045708567e-05,
+      "loss": 0.5173,
+      "step": 3946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4036199849862375e-05,
+      "loss": 0.4866,
+      "step": 3947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4034531654016184e-05,
+      "loss": 0.4628,
+      "step": 3948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.403286345816999e-05,
+      "loss": 0.4608,
+      "step": 3949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4031195262323798e-05,
+      "loss": 0.4928,
+      "step": 3950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4029527066477606e-05,
+      "loss": 0.456,
+      "step": 3951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4027858870631412e-05,
+      "loss": 0.4723,
+      "step": 3952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.402619067478522e-05,
+      "loss": 0.4709,
+      "step": 3953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4024522478939028e-05,
+      "loss": 0.4939,
+      "step": 3954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4022854283092838e-05,
+      "loss": 0.5096,
+      "step": 3955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4021186087246644e-05,
+      "loss": 0.5137,
+      "step": 3956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4019517891400452e-05,
+      "loss": 0.4709,
+      "step": 3957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.401784969555426e-05,
+      "loss": 0.5038,
+      "step": 3958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4016181499708066e-05,
+      "loss": 0.5092,
+      "step": 3959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4014513303861874e-05,
+      "loss": 0.4716,
+      "step": 3960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4012845108015682e-05,
+      "loss": 0.49,
+      "step": 3961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4011176912169491e-05,
+      "loss": 0.4551,
+      "step": 3962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4009508716323297e-05,
+      "loss": 0.492,
+      "step": 3963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4007840520477105e-05,
+      "loss": 0.4983,
+      "step": 3964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4006172324630913e-05,
+      "loss": 0.4814,
+      "step": 3965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.400450412878472e-05,
+      "loss": 0.453,
+      "step": 3966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4002835932938527e-05,
+      "loss": 0.4629,
+      "step": 3967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.4001167737092337e-05,
+      "loss": 0.5318,
+      "step": 3968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3999499541246145e-05,
+      "loss": 0.5086,
+      "step": 3969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3997831345399951e-05,
+      "loss": 0.4794,
+      "step": 3970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3996163149553759e-05,
+      "loss": 0.4699,
+      "step": 3971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3994494953707567e-05,
+      "loss": 0.5183,
+      "step": 3972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3992826757861373e-05,
+      "loss": 0.4805,
+      "step": 3973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.399115856201518e-05,
+      "loss": 0.4823,
+      "step": 3974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.398949036616899e-05,
+      "loss": 0.473,
+      "step": 3975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3987822170322798e-05,
+      "loss": 0.4951,
+      "step": 3976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3986153974476604e-05,
+      "loss": 0.4826,
+      "step": 3977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3984485778630412e-05,
+      "loss": 0.4983,
+      "step": 3978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.398281758278422e-05,
+      "loss": 0.4899,
+      "step": 3979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3981149386938026e-05,
+      "loss": 0.5084,
+      "step": 3980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3979481191091834e-05,
+      "loss": 0.4493,
+      "step": 3981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3977812995245644e-05,
+      "loss": 0.5059,
+      "step": 3982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3976144799399452e-05,
+      "loss": 0.4328,
+      "step": 3983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3974476603553258e-05,
+      "loss": 0.4624,
+      "step": 3984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3972808407707066e-05,
+      "loss": 0.4608,
+      "step": 3985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3971140211860874e-05,
+      "loss": 0.4632,
+      "step": 3986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.396947201601468e-05,
+      "loss": 0.5191,
+      "step": 3987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3967803820168488e-05,
+      "loss": 0.476,
+      "step": 3988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3966135624322297e-05,
+      "loss": 0.4834,
+      "step": 3989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3964467428476105e-05,
+      "loss": 0.4874,
+      "step": 3990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3962799232629911e-05,
+      "loss": 0.4654,
+      "step": 3991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.396113103678372e-05,
+      "loss": 0.4853,
+      "step": 3992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3959462840937527e-05,
+      "loss": 0.4707,
+      "step": 3993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3957794645091337e-05,
+      "loss": 0.4551,
+      "step": 3994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3956126449245141e-05,
+      "loss": 0.4866,
+      "step": 3995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3954458253398951e-05,
+      "loss": 0.5192,
+      "step": 3996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3952790057552759e-05,
+      "loss": 0.5021,
+      "step": 3997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3951121861706565e-05,
+      "loss": 0.4882,
+      "step": 3998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3949453665860373e-05,
+      "loss": 0.4414,
+      "step": 3999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.394778547001418e-05,
+      "loss": 0.4806,
+      "step": 4000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.394611727416799e-05,
+      "loss": 0.4977,
+      "step": 4001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3944449078321795e-05,
+      "loss": 0.5148,
+      "step": 4002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3942780882475604e-05,
+      "loss": 0.5283,
+      "step": 4003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3941112686629412e-05,
+      "loss": 0.4915,
+      "step": 4004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3939444490783218e-05,
+      "loss": 0.4599,
+      "step": 4005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3937776294937026e-05,
+      "loss": 0.4709,
+      "step": 4006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3936108099090834e-05,
+      "loss": 0.4634,
+      "step": 4007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3934439903244644e-05,
+      "loss": 0.4912,
+      "step": 4008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.393277170739845e-05,
+      "loss": 0.4868,
+      "step": 4009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3931103511552258e-05,
+      "loss": 0.4695,
+      "step": 4010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3929435315706066e-05,
+      "loss": 0.5024,
+      "step": 4011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3927767119859872e-05,
+      "loss": 0.4521,
+      "step": 4012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.392609892401368e-05,
+      "loss": 0.5095,
+      "step": 4013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3924430728167488e-05,
+      "loss": 0.5008,
+      "step": 4014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3922762532321297e-05,
+      "loss": 0.4993,
+      "step": 4015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3921094336475104e-05,
+      "loss": 0.4749,
+      "step": 4016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.3919426140628911e-05,
+      "loss": 0.4894,
+      "step": 4017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.391775794478272e-05,
+      "loss": 0.4897,
+      "step": 4018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3916089748936525e-05,
+      "loss": 0.5213,
+      "step": 4019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3914421553090333e-05,
+      "loss": 0.4659,
+      "step": 4020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3912753357244141e-05,
+      "loss": 0.5035,
+      "step": 4021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3911085161397951e-05,
+      "loss": 0.4793,
+      "step": 4022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3909416965551757e-05,
+      "loss": 0.4812,
+      "step": 4023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3907748769705565e-05,
+      "loss": 0.468,
+      "step": 4024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3906080573859373e-05,
+      "loss": 0.4799,
+      "step": 4025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3904412378013179e-05,
+      "loss": 0.5172,
+      "step": 4026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3902744182166987e-05,
+      "loss": 0.4809,
+      "step": 4027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3901075986320795e-05,
+      "loss": 0.5187,
+      "step": 4028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3899407790474604e-05,
+      "loss": 0.5058,
+      "step": 4029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389773959462841e-05,
+      "loss": 0.495,
+      "step": 4030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3896071398782218e-05,
+      "loss": 0.5234,
+      "step": 4031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3894403202936026e-05,
+      "loss": 0.4903,
+      "step": 4032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3892735007089833e-05,
+      "loss": 0.4676,
+      "step": 4033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.389106681124364e-05,
+      "loss": 0.442,
+      "step": 4034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.388939861539745e-05,
+      "loss": 0.4751,
+      "step": 4035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3887730419551258e-05,
+      "loss": 0.4559,
+      "step": 4036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3886062223705064e-05,
+      "loss": 0.4746,
+      "step": 4037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3884394027858872e-05,
+      "loss": 0.5272,
+      "step": 4038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.388272583201268e-05,
+      "loss": 0.4479,
+      "step": 4039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3881057636166486e-05,
+      "loss": 0.4898,
+      "step": 4040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3879389440320294e-05,
+      "loss": 0.48,
+      "step": 4041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3877721244474104e-05,
+      "loss": 0.4988,
+      "step": 4042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3876053048627911e-05,
+      "loss": 0.4976,
+      "step": 4043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3874384852781718e-05,
+      "loss": 0.4916,
+      "step": 4044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3872716656935525e-05,
+      "loss": 0.4651,
+      "step": 4045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3871048461089333e-05,
+      "loss": 0.4972,
+      "step": 4046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.386938026524314e-05,
+      "loss": 0.4807,
+      "step": 4047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3867712069396947e-05,
+      "loss": 0.4875,
+      "step": 4048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3866043873550757e-05,
+      "loss": 0.4811,
+      "step": 4049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3864375677704565e-05,
+      "loss": 0.4716,
+      "step": 4050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3862707481858371e-05,
+      "loss": 0.4698,
+      "step": 4051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3861039286012179e-05,
+      "loss": 0.4417,
+      "step": 4052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3859371090165987e-05,
+      "loss": 0.5096,
+      "step": 4053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3857702894319793e-05,
+      "loss": 0.4841,
+      "step": 4054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3856034698473601e-05,
+      "loss": 0.4476,
+      "step": 4055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.385436650262741e-05,
+      "loss": 0.4851,
+      "step": 4056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3852698306781218e-05,
+      "loss": 0.4848,
+      "step": 4057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3851030110935025e-05,
+      "loss": 0.4511,
+      "step": 4058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3849361915088833e-05,
+      "loss": 0.4787,
+      "step": 4059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.384769371924264e-05,
+      "loss": 0.5006,
+      "step": 4060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3846025523396447e-05,
+      "loss": 0.4723,
+      "step": 4061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3844357327550254e-05,
+      "loss": 0.4689,
+      "step": 4062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3842689131704064e-05,
+      "loss": 0.4648,
+      "step": 4063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3841020935857872e-05,
+      "loss": 0.4797,
+      "step": 4064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3839352740011678e-05,
+      "loss": 0.4842,
+      "step": 4065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3837684544165486e-05,
+      "loss": 0.4647,
+      "step": 4066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3836016348319294e-05,
+      "loss": 0.4595,
+      "step": 4067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.38343481524731e-05,
+      "loss": 0.4696,
+      "step": 4068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3832679956626908e-05,
+      "loss": 0.4694,
+      "step": 4069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3831011760780718e-05,
+      "loss": 0.482,
+      "step": 4070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3829343564934525e-05,
+      "loss": 0.5065,
+      "step": 4071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3827675369088332e-05,
+      "loss": 0.467,
+      "step": 4072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.382600717324214e-05,
+      "loss": 0.5075,
+      "step": 4073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3824338977395947e-05,
+      "loss": 0.4936,
+      "step": 4074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3822670781549754e-05,
+      "loss": 0.459,
+      "step": 4075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3821002585703562e-05,
+      "loss": 0.4738,
+      "step": 4076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3819334389857371e-05,
+      "loss": 0.5274,
+      "step": 4077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3817666194011179e-05,
+      "loss": 0.4807,
+      "step": 4078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3815997998164985e-05,
+      "loss": 0.4596,
+      "step": 4079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3814329802318793e-05,
+      "loss": 0.4655,
+      "step": 4080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3812661606472601e-05,
+      "loss": 0.5014,
+      "step": 4081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3810993410626407e-05,
+      "loss": 0.4535,
+      "step": 4082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3809325214780217e-05,
+      "loss": 0.4849,
+      "step": 4083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3807657018934025e-05,
+      "loss": 0.4722,
+      "step": 4084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3805988823087832e-05,
+      "loss": 0.4567,
+      "step": 4085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3804320627241639e-05,
+      "loss": 0.4264,
+      "step": 4086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3802652431395447e-05,
+      "loss": 0.4662,
+      "step": 4087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3800984235549254e-05,
+      "loss": 0.5011,
+      "step": 4088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.379931603970306e-05,
+      "loss": 0.4916,
+      "step": 4089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.379764784385687e-05,
+      "loss": 0.5477,
+      "step": 4090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3795979648010678e-05,
+      "loss": 0.5178,
+      "step": 4091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3794311452164486e-05,
+      "loss": 0.5126,
+      "step": 4092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3792643256318292e-05,
+      "loss": 0.5046,
+      "step": 4093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.37909750604721e-05,
+      "loss": 0.4769,
+      "step": 4094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3789306864625908e-05,
+      "loss": 0.4951,
+      "step": 4095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3787638668779714e-05,
+      "loss": 0.4897,
+      "step": 4096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3785970472933524e-05,
+      "loss": 0.5003,
+      "step": 4097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3784302277087332e-05,
+      "loss": 0.4754,
+      "step": 4098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.378263408124114e-05,
+      "loss": 0.4897,
+      "step": 4099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3780965885394946e-05,
+      "loss": 0.503,
+      "step": 4100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3779297689548754e-05,
+      "loss": 0.4448,
+      "step": 4101
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3777629493702561e-05,
+      "loss": 0.4697,
+      "step": 4102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3775961297856368e-05,
+      "loss": 0.5125,
+      "step": 4103
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3774293102010177e-05,
+      "loss": 0.4937,
+      "step": 4104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3772624906163985e-05,
+      "loss": 0.4528,
+      "step": 4105
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3770956710317793e-05,
+      "loss": 0.5057,
+      "step": 4106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.37692885144716e-05,
+      "loss": 0.4442,
+      "step": 4107
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3767620318625407e-05,
+      "loss": 0.5009,
+      "step": 4108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3765952122779217e-05,
+      "loss": 0.4886,
+      "step": 4109
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3764283926933021e-05,
+      "loss": 0.517,
+      "step": 4110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.376261573108683e-05,
+      "loss": 0.4972,
+      "step": 4111
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3760947535240639e-05,
+      "loss": 0.4854,
+      "step": 4112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3759279339394447e-05,
+      "loss": 0.4848,
+      "step": 4113
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3757611143548253e-05,
+      "loss": 0.4925,
+      "step": 4114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.375594294770206e-05,
+      "loss": 0.5218,
+      "step": 4115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.375427475185587e-05,
+      "loss": 0.5368,
+      "step": 4116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3752606556009678e-05,
+      "loss": 0.4926,
+      "step": 4117
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3750938360163484e-05,
+      "loss": 0.4635,
+      "step": 4118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3749270164317292e-05,
+      "loss": 0.4843,
+      "step": 4119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.37476019684711e-05,
+      "loss": 0.4798,
+      "step": 4120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3745933772624906e-05,
+      "loss": 0.4472,
+      "step": 4121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3744265576778714e-05,
+      "loss": 0.4121,
+      "step": 4122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3742597380932524e-05,
+      "loss": 0.4013,
+      "step": 4123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3740929185086332e-05,
+      "loss": 0.4116,
+      "step": 4124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3739260989240138e-05,
+      "loss": 0.3978,
+      "step": 4125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3737592793393946e-05,
+      "loss": 0.405,
+      "step": 4126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3735924597547754e-05,
+      "loss": 0.4401,
+      "step": 4127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.373425640170156e-05,
+      "loss": 0.4094,
+      "step": 4128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3732588205855368e-05,
+      "loss": 0.4132,
+      "step": 4129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3730920010009177e-05,
+      "loss": 0.4026,
+      "step": 4130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3729251814162985e-05,
+      "loss": 0.4039,
+      "step": 4131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3727583618316791e-05,
+      "loss": 0.411,
+      "step": 4132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.37259154224706e-05,
+      "loss": 0.3908,
+      "step": 4133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3724247226624407e-05,
+      "loss": 0.4182,
+      "step": 4134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3722579030778213e-05,
+      "loss": 0.3747,
+      "step": 4135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3720910834932021e-05,
+      "loss": 0.4266,
+      "step": 4136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.371924263908583e-05,
+      "loss": 0.3833,
+      "step": 4137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3717574443239639e-05,
+      "loss": 0.3986,
+      "step": 4138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3715906247393445e-05,
+      "loss": 0.3971,
+      "step": 4139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3714238051547253e-05,
+      "loss": 0.4007,
+      "step": 4140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.371256985570106e-05,
+      "loss": 0.3857,
+      "step": 4141
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3710901659854867e-05,
+      "loss": 0.402,
+      "step": 4142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3709233464008675e-05,
+      "loss": 0.3845,
+      "step": 4143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3707565268162484e-05,
+      "loss": 0.4077,
+      "step": 4144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3705897072316292e-05,
+      "loss": 0.4253,
+      "step": 4145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3704228876470098e-05,
+      "loss": 0.3682,
+      "step": 4146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3702560680623906e-05,
+      "loss": 0.4238,
+      "step": 4147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3700892484777714e-05,
+      "loss": 0.3929,
+      "step": 4148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.369922428893152e-05,
+      "loss": 0.4106,
+      "step": 4149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.369755609308533e-05,
+      "loss": 0.425,
+      "step": 4150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3695887897239138e-05,
+      "loss": 0.4107,
+      "step": 4151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3694219701392946e-05,
+      "loss": 0.3926,
+      "step": 4152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3692551505546752e-05,
+      "loss": 0.4074,
+      "step": 4153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.369088330970056e-05,
+      "loss": 0.4087,
+      "step": 4154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3689215113854368e-05,
+      "loss": 0.3516,
+      "step": 4155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3687546918008174e-05,
+      "loss": 0.394,
+      "step": 4156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3685878722161983e-05,
+      "loss": 0.4035,
+      "step": 4157
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 0.3966,
+      "step": 4158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.36825423304696e-05,
+      "loss": 0.3936,
+      "step": 4159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3680874134623405e-05,
+      "loss": 0.3867,
+      "step": 4160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3679205938777213e-05,
+      "loss": 0.3995,
+      "step": 4161
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3677537742931021e-05,
+      "loss": 0.4031,
+      "step": 4162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3675869547084827e-05,
+      "loss": 0.4302,
+      "step": 4163
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3674201351238637e-05,
+      "loss": 0.3991,
+      "step": 4164
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3672533155392445e-05,
+      "loss": 0.3874,
+      "step": 4165
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3670864959546253e-05,
+      "loss": 0.4048,
+      "step": 4166
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3669196763700059e-05,
+      "loss": 0.4082,
+      "step": 4167
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3667528567853867e-05,
+      "loss": 0.4036,
+      "step": 4168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3665860372007675e-05,
+      "loss": 0.401,
+      "step": 4169
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3664192176161481e-05,
+      "loss": 0.4233,
+      "step": 4170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.366252398031529e-05,
+      "loss": 0.42,
+      "step": 4171
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3660855784469098e-05,
+      "loss": 0.3782,
+      "step": 4172
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3659187588622906e-05,
+      "loss": 0.4094,
+      "step": 4173
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3657519392776712e-05,
+      "loss": 0.3952,
+      "step": 4174
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.365585119693052e-05,
+      "loss": 0.3732,
+      "step": 4175
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3654183001084328e-05,
+      "loss": 0.3933,
+      "step": 4176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3652514805238134e-05,
+      "loss": 0.4279,
+      "step": 4177
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3650846609391944e-05,
+      "loss": 0.4211,
+      "step": 4178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3649178413545752e-05,
+      "loss": 0.3995,
+      "step": 4179
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.364751021769956e-05,
+      "loss": 0.4193,
+      "step": 4180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3645842021853366e-05,
+      "loss": 0.3901,
+      "step": 4181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.3644173826007174e-05,
+      "loss": 0.3655,
+      "step": 4182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3642505630160983e-05,
+      "loss": 0.4132,
+      "step": 4183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3640837434314788e-05,
+      "loss": 0.3697,
+      "step": 4184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3639169238468597e-05,
+      "loss": 0.3992,
+      "step": 4185
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3637501042622405e-05,
+      "loss": 0.3677,
+      "step": 4186
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3635832846776213e-05,
+      "loss": 0.3885,
+      "step": 4187
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.363416465093002e-05,
+      "loss": 0.439,
+      "step": 4188
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3632496455083827e-05,
+      "loss": 0.3951,
+      "step": 4189
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3630828259237637e-05,
+      "loss": 0.3902,
+      "step": 4190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3629160063391443e-05,
+      "loss": 0.3834,
+      "step": 4191
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3627491867545251e-05,
+      "loss": 0.4118,
+      "step": 4192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3625823671699059e-05,
+      "loss": 0.4206,
+      "step": 4193
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3624155475852867e-05,
+      "loss": 0.396,
+      "step": 4194
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3622487280006673e-05,
+      "loss": 0.4109,
+      "step": 4195
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3620819084160481e-05,
+      "loss": 0.3908,
+      "step": 4196
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.361915088831429e-05,
+      "loss": 0.4219,
+      "step": 4197
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3617482692468097e-05,
+      "loss": 0.3976,
+      "step": 4198
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3615814496621905e-05,
+      "loss": 0.3607,
+      "step": 4199
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3614146300775712e-05,
+      "loss": 0.3936,
+      "step": 4200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.361247810492952e-05,
+      "loss": 0.421,
+      "step": 4201
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3610809909083326e-05,
+      "loss": 0.3828,
+      "step": 4202
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3609141713237134e-05,
+      "loss": 0.3889,
+      "step": 4203
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3607473517390944e-05,
+      "loss": 0.4142,
+      "step": 4204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.360580532154475e-05,
+      "loss": 0.4225,
+      "step": 4205
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3604137125698558e-05,
+      "loss": 0.4145,
+      "step": 4206
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3602468929852366e-05,
+      "loss": 0.3793,
+      "step": 4207
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3600800734006174e-05,
+      "loss": 0.4312,
+      "step": 4208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.359913253815998e-05,
+      "loss": 0.4041,
+      "step": 4209
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3597464342313788e-05,
+      "loss": 0.3866,
+      "step": 4210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3595796146467597e-05,
+      "loss": 0.4373,
+      "step": 4211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3594127950621404e-05,
+      "loss": 0.3811,
+      "step": 4212
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3592459754775212e-05,
+      "loss": 0.3837,
+      "step": 4213
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.359079155892902e-05,
+      "loss": 0.4063,
+      "step": 4214
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3589123363082827e-05,
+      "loss": 0.379,
+      "step": 4215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3587455167236634e-05,
+      "loss": 0.4156,
+      "step": 4216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3585786971390441e-05,
+      "loss": 0.4118,
+      "step": 4217
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3584118775544251e-05,
+      "loss": 0.4264,
+      "step": 4218
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3582450579698057e-05,
+      "loss": 0.4015,
+      "step": 4219
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3580782383851865e-05,
+      "loss": 0.4174,
+      "step": 4220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3579114188005673e-05,
+      "loss": 0.4112,
+      "step": 4221
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.357744599215948e-05,
+      "loss": 0.3972,
+      "step": 4222
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3575777796313287e-05,
+      "loss": 0.4091,
+      "step": 4223
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3574109600467097e-05,
+      "loss": 0.3825,
+      "step": 4224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3572441404620904e-05,
+      "loss": 0.3806,
+      "step": 4225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.357077320877471e-05,
+      "loss": 0.4022,
+      "step": 4226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3569105012928519e-05,
+      "loss": 0.4086,
+      "step": 4227
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3567436817082326e-05,
+      "loss": 0.4048,
+      "step": 4228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3565768621236134e-05,
+      "loss": 0.402,
+      "step": 4229
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.356410042538994e-05,
+      "loss": 0.3835,
+      "step": 4230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.356243222954375e-05,
+      "loss": 0.3819,
+      "step": 4231
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3560764033697558e-05,
+      "loss": 0.4052,
+      "step": 4232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3559095837851366e-05,
+      "loss": 0.4139,
+      "step": 4233
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3557427642005172e-05,
+      "loss": 0.4015,
+      "step": 4234
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.355575944615898e-05,
+      "loss": 0.3753,
+      "step": 4235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3554091250312788e-05,
+      "loss": 0.3845,
+      "step": 4236
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3552423054466594e-05,
+      "loss": 0.3927,
+      "step": 4237
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3550754858620404e-05,
+      "loss": 0.3976,
+      "step": 4238
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3549086662774212e-05,
+      "loss": 0.4154,
+      "step": 4239
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.354741846692802e-05,
+      "loss": 0.4074,
+      "step": 4240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3545750271081826e-05,
+      "loss": 0.4068,
+      "step": 4241
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3544082075235633e-05,
+      "loss": 0.3727,
+      "step": 4242
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3542413879389441e-05,
+      "loss": 0.3878,
+      "step": 4243
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3540745683543248e-05,
+      "loss": 0.3969,
+      "step": 4244
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3539077487697057e-05,
+      "loss": 0.4174,
+      "step": 4245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3537409291850865e-05,
+      "loss": 0.4427,
+      "step": 4246
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3535741096004673e-05,
+      "loss": 0.4165,
+      "step": 4247
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3534072900158479e-05,
+      "loss": 0.3648,
+      "step": 4248
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3532404704312287e-05,
+      "loss": 0.3733,
+      "step": 4249
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3530736508466097e-05,
+      "loss": 0.4175,
+      "step": 4250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3529068312619901e-05,
+      "loss": 0.3884,
+      "step": 4251
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.352740011677371e-05,
+      "loss": 0.4313,
+      "step": 4252
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3525731920927519e-05,
+      "loss": 0.3959,
+      "step": 4253
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3524063725081326e-05,
+      "loss": 0.4151,
+      "step": 4254
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3522395529235133e-05,
+      "loss": 0.4168,
+      "step": 4255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.352072733338894e-05,
+      "loss": 0.3944,
+      "step": 4256
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.351905913754275e-05,
+      "loss": 0.3946,
+      "step": 4257
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3517390941696555e-05,
+      "loss": 0.4024,
+      "step": 4258
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3515722745850364e-05,
+      "loss": 0.3894,
+      "step": 4259
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3514054550004172e-05,
+      "loss": 0.3952,
+      "step": 4260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.351238635415798e-05,
+      "loss": 0.3932,
+      "step": 4261
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3510718158311786e-05,
+      "loss": 0.4187,
+      "step": 4262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3509049962465594e-05,
+      "loss": 0.4082,
+      "step": 4263
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.3507381766619404e-05,
+      "loss": 0.4379,
+      "step": 4264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.350571357077321e-05,
+      "loss": 0.3787,
+      "step": 4265
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3504045374927018e-05,
+      "loss": 0.3932,
+      "step": 4266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3502377179080826e-05,
+      "loss": 0.4299,
+      "step": 4267
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3500708983234633e-05,
+      "loss": 0.4059,
+      "step": 4268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.349904078738844e-05,
+      "loss": 0.3924,
+      "step": 4269
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3497372591542248e-05,
+      "loss": 0.3909,
+      "step": 4270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3495704395696057e-05,
+      "loss": 0.3868,
+      "step": 4271
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3494036199849863e-05,
+      "loss": 0.4011,
+      "step": 4272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3492368004003671e-05,
+      "loss": 0.397,
+      "step": 4273
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3490699808157479e-05,
+      "loss": 0.4045,
+      "step": 4274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3489031612311287e-05,
+      "loss": 0.4052,
+      "step": 4275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3487363416465093e-05,
+      "loss": 0.4069,
+      "step": 4276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3485695220618901e-05,
+      "loss": 0.4136,
+      "step": 4277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.348402702477271e-05,
+      "loss": 0.3983,
+      "step": 4278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3482358828926517e-05,
+      "loss": 0.3802,
+      "step": 4279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3480690633080325e-05,
+      "loss": 0.3924,
+      "step": 4280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3479022437234133e-05,
+      "loss": 0.4087,
+      "step": 4281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.347735424138794e-05,
+      "loss": 0.396,
+      "step": 4282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3475686045541747e-05,
+      "loss": 0.4508,
+      "step": 4283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3474017849695555e-05,
+      "loss": 0.3787,
+      "step": 4284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3472349653849364e-05,
+      "loss": 0.4088,
+      "step": 4285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.347068145800317e-05,
+      "loss": 0.395,
+      "step": 4286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3469013262156978e-05,
+      "loss": 0.3991,
+      "step": 4287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3467345066310786e-05,
+      "loss": 0.39,
+      "step": 4288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3465676870464594e-05,
+      "loss": 0.4083,
+      "step": 4289
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.34640086746184e-05,
+      "loss": 0.4141,
+      "step": 4290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.346234047877221e-05,
+      "loss": 0.4257,
+      "step": 4291
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3460672282926018e-05,
+      "loss": 0.3951,
+      "step": 4292
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3459004087079824e-05,
+      "loss": 0.3851,
+      "step": 4293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3457335891233632e-05,
+      "loss": 0.3998,
+      "step": 4294
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.345566769538744e-05,
+      "loss": 0.4188,
+      "step": 4295
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3453999499541248e-05,
+      "loss": 0.4236,
+      "step": 4296
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3452331303695054e-05,
+      "loss": 0.4067,
+      "step": 4297
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3450663107848863e-05,
+      "loss": 0.4221,
+      "step": 4298
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3448994912002671e-05,
+      "loss": 0.3858,
+      "step": 4299
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3447326716156477e-05,
+      "loss": 0.3966,
+      "step": 4300
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3445658520310285e-05,
+      "loss": 0.3958,
+      "step": 4301
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3443990324464093e-05,
+      "loss": 0.4004,
+      "step": 4302
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3442322128617901e-05,
+      "loss": 0.4034,
+      "step": 4303
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3440653932771707e-05,
+      "loss": 0.3701,
+      "step": 4304
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.3438985736925517e-05,
+      "loss": 0.4069,
+      "step": 4305
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3437317541079325e-05,
+      "loss": 0.3545,
+      "step": 4306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3435649345233131e-05,
+      "loss": 0.3961,
+      "step": 4307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3433981149386939e-05,
+      "loss": 0.4018,
+      "step": 4308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3432312953540747e-05,
+      "loss": 0.3811,
+      "step": 4309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3430644757694555e-05,
+      "loss": 0.4054,
+      "step": 4310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.342897656184836e-05,
+      "loss": 0.3884,
+      "step": 4311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.342730836600217e-05,
+      "loss": 0.3979,
+      "step": 4312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3425640170155978e-05,
+      "loss": 0.414,
+      "step": 4313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3423971974309784e-05,
+      "loss": 0.4175,
+      "step": 4314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3422303778463592e-05,
+      "loss": 0.4248,
+      "step": 4315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.34206355826174e-05,
+      "loss": 0.3664,
+      "step": 4316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3418967386771208e-05,
+      "loss": 0.424,
+      "step": 4317
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3417299190925014e-05,
+      "loss": 0.3867,
+      "step": 4318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3415630995078824e-05,
+      "loss": 0.3908,
+      "step": 4319
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3413962799232632e-05,
+      "loss": 0.3984,
+      "step": 4320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3412294603386438e-05,
+      "loss": 0.3732,
+      "step": 4321
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3410626407540246e-05,
+      "loss": 0.3793,
+      "step": 4322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3408958211694054e-05,
+      "loss": 0.3865,
+      "step": 4323
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3407290015847863e-05,
+      "loss": 0.4071,
+      "step": 4324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3405621820001668e-05,
+      "loss": 0.3919,
+      "step": 4325
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3403953624155477e-05,
+      "loss": 0.3897,
+      "step": 4326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3402285428309285e-05,
+      "loss": 0.4205,
+      "step": 4327
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3400617232463091e-05,
+      "loss": 0.372,
+      "step": 4328
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.33989490366169e-05,
+      "loss": 0.4065,
+      "step": 4329
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3397280840770707e-05,
+      "loss": 0.3942,
+      "step": 4330
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3395612644924517e-05,
+      "loss": 0.3636,
+      "step": 4331
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3393944449078321e-05,
+      "loss": 0.366,
+      "step": 4332
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3392276253232131e-05,
+      "loss": 0.3754,
+      "step": 4333
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3390608057385939e-05,
+      "loss": 0.3976,
+      "step": 4334
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3388939861539745e-05,
+      "loss": 0.4036,
+      "step": 4335
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3387271665693553e-05,
+      "loss": 0.3954,
+      "step": 4336
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.338560346984736e-05,
+      "loss": 0.3903,
+      "step": 4337
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.338393527400117e-05,
+      "loss": 0.3858,
+      "step": 4338
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3382267078154977e-05,
+      "loss": 0.3909,
+      "step": 4339
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3380598882308784e-05,
+      "loss": 0.4141,
+      "step": 4340
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3378930686462592e-05,
+      "loss": 0.4127,
+      "step": 4341
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3377262490616398e-05,
+      "loss": 0.3893,
+      "step": 4342
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3375594294770206e-05,
+      "loss": 0.3828,
+      "step": 4343
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3373926098924014e-05,
+      "loss": 0.4202,
+      "step": 4344
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3372257903077824e-05,
+      "loss": 0.3975,
+      "step": 4345
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.337058970723163e-05,
+      "loss": 0.4023,
+      "step": 4346
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.3368921511385438e-05,
+      "loss": 0.4074,
+      "step": 4347
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3367253315539246e-05,
+      "loss": 0.4298,
+      "step": 4348
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3365585119693054e-05,
+      "loss": 0.3824,
+      "step": 4349
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.336391692384686e-05,
+      "loss": 0.4331,
+      "step": 4350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3362248728000668e-05,
+      "loss": 0.3791,
+      "step": 4351
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3360580532154477e-05,
+      "loss": 0.4351,
+      "step": 4352
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3358912336308284e-05,
+      "loss": 0.4175,
+      "step": 4353
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3357244140462091e-05,
+      "loss": 0.3991,
+      "step": 4354
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.33555759446159e-05,
+      "loss": 0.4213,
+      "step": 4355
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3353907748769707e-05,
+      "loss": 0.4225,
+      "step": 4356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3352239552923513e-05,
+      "loss": 0.4119,
+      "step": 4357
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3350571357077321e-05,
+      "loss": 0.4261,
+      "step": 4358
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3348903161231131e-05,
+      "loss": 0.4008,
+      "step": 4359
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3347234965384937e-05,
+      "loss": 0.4243,
+      "step": 4360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3345566769538745e-05,
+      "loss": 0.4306,
+      "step": 4361
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3343898573692553e-05,
+      "loss": 0.4118,
+      "step": 4362
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.334223037784636e-05,
+      "loss": 0.3993,
+      "step": 4363
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3340562182000167e-05,
+      "loss": 0.3933,
+      "step": 4364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3338893986153976e-05,
+      "loss": 0.418,
+      "step": 4365
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3337225790307784e-05,
+      "loss": 0.392,
+      "step": 4366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.333555759446159e-05,
+      "loss": 0.4028,
+      "step": 4367
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3333889398615398e-05,
+      "loss": 0.4205,
+      "step": 4368
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3332221202769206e-05,
+      "loss": 0.4033,
+      "step": 4369
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3330553006923014e-05,
+      "loss": 0.3756,
+      "step": 4370
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.332888481107682e-05,
+      "loss": 0.3927,
+      "step": 4371
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.332721661523063e-05,
+      "loss": 0.4119,
+      "step": 4372
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3325548419384438e-05,
+      "loss": 0.4063,
+      "step": 4373
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3323880223538244e-05,
+      "loss": 0.36,
+      "step": 4374
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3322212027692052e-05,
+      "loss": 0.4079,
+      "step": 4375
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.332054383184586e-05,
+      "loss": 0.3996,
+      "step": 4376
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3318875635999668e-05,
+      "loss": 0.3762,
+      "step": 4377
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3317207440153474e-05,
+      "loss": 0.3864,
+      "step": 4378
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3315539244307284e-05,
+      "loss": 0.4442,
+      "step": 4379
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3313871048461091e-05,
+      "loss": 0.4381,
+      "step": 4380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3312202852614898e-05,
+      "loss": 0.4083,
+      "step": 4381
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3310534656768705e-05,
+      "loss": 0.4068,
+      "step": 4382
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3308866460922513e-05,
+      "loss": 0.4262,
+      "step": 4383
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3307198265076321e-05,
+      "loss": 0.3814,
+      "step": 4384
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3305530069230127e-05,
+      "loss": 0.411,
+      "step": 4385
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3303861873383937e-05,
+      "loss": 0.3839,
+      "step": 4386
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3302193677537745e-05,
+      "loss": 0.3827,
+      "step": 4387
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.3300525481691551e-05,
+      "loss": 0.3942,
+      "step": 4388
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3298857285845359e-05,
+      "loss": 0.4021,
+      "step": 4389
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3297189089999167e-05,
+      "loss": 0.4087,
+      "step": 4390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3295520894152976e-05,
+      "loss": 0.4019,
+      "step": 4391
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3293852698306781e-05,
+      "loss": 0.4145,
+      "step": 4392
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.329218450246059e-05,
+      "loss": 0.38,
+      "step": 4393
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3290516306614398e-05,
+      "loss": 0.4357,
+      "step": 4394
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3288848110768205e-05,
+      "loss": 0.393,
+      "step": 4395
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3287179914922013e-05,
+      "loss": 0.3986,
+      "step": 4396
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.328551171907582e-05,
+      "loss": 0.41,
+      "step": 4397
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.328384352322963e-05,
+      "loss": 0.4219,
+      "step": 4398
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3282175327383434e-05,
+      "loss": 0.3794,
+      "step": 4399
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3280507131537244e-05,
+      "loss": 0.3817,
+      "step": 4400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3278838935691052e-05,
+      "loss": 0.4192,
+      "step": 4401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3277170739844858e-05,
+      "loss": 0.3717,
+      "step": 4402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3275502543998666e-05,
+      "loss": 0.4114,
+      "step": 4403
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3273834348152474e-05,
+      "loss": 0.3795,
+      "step": 4404
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3272166152306284e-05,
+      "loss": 0.3882,
+      "step": 4405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.327049795646009e-05,
+      "loss": 0.4,
+      "step": 4406
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3268829760613898e-05,
+      "loss": 0.4025,
+      "step": 4407
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3267161564767705e-05,
+      "loss": 0.3872,
+      "step": 4408
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3265493368921512e-05,
+      "loss": 0.3969,
+      "step": 4409
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.326382517307532e-05,
+      "loss": 0.4292,
+      "step": 4410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3262156977229127e-05,
+      "loss": 0.4161,
+      "step": 4411
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3260488781382937e-05,
+      "loss": 0.4046,
+      "step": 4412
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3258820585536743e-05,
+      "loss": 0.4021,
+      "step": 4413
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3257152389690551e-05,
+      "loss": 0.4148,
+      "step": 4414
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3255484193844359e-05,
+      "loss": 0.422,
+      "step": 4415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3253815997998165e-05,
+      "loss": 0.4061,
+      "step": 4416
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3252147802151973e-05,
+      "loss": 0.3701,
+      "step": 4417
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3250479606305781e-05,
+      "loss": 0.3805,
+      "step": 4418
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.324881141045959e-05,
+      "loss": 0.4076,
+      "step": 4419
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3247143214613397e-05,
+      "loss": 0.3858,
+      "step": 4420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3245475018767205e-05,
+      "loss": 0.3642,
+      "step": 4421
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3243806822921013e-05,
+      "loss": 0.3929,
+      "step": 4422
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3242138627074819e-05,
+      "loss": 0.4078,
+      "step": 4423
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3240470431228627e-05,
+      "loss": 0.4077,
+      "step": 4424
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3238802235382434e-05,
+      "loss": 0.4043,
+      "step": 4425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3237134039536244e-05,
+      "loss": 0.4153,
+      "step": 4426
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.323546584369005e-05,
+      "loss": 0.4171,
+      "step": 4427
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3233797647843858e-05,
+      "loss": 0.4056,
+      "step": 4428
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3232129451997666e-05,
+      "loss": 0.4097,
+      "step": 4429
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3230461256151472e-05,
+      "loss": 0.4141,
+      "step": 4430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.322879306030528e-05,
+      "loss": 0.4024,
+      "step": 4431
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3227124864459088e-05,
+      "loss": 0.412,
+      "step": 4432
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3225456668612898e-05,
+      "loss": 0.3853,
+      "step": 4433
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3223788472766704e-05,
+      "loss": 0.3775,
+      "step": 4434
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3222120276920512e-05,
+      "loss": 0.4285,
+      "step": 4435
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.322045208107432e-05,
+      "loss": 0.4046,
+      "step": 4436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3218783885228126e-05,
+      "loss": 0.4099,
+      "step": 4437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3217115689381934e-05,
+      "loss": 0.3641,
+      "step": 4438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3215447493535743e-05,
+      "loss": 0.427,
+      "step": 4439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3213779297689551e-05,
+      "loss": 0.4316,
+      "step": 4440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3212111101843357e-05,
+      "loss": 0.3647,
+      "step": 4441
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3210442905997165e-05,
+      "loss": 0.4109,
+      "step": 4442
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3208774710150973e-05,
+      "loss": 0.3968,
+      "step": 4443
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.320710651430478e-05,
+      "loss": 0.4097,
+      "step": 4444
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3205438318458587e-05,
+      "loss": 0.4123,
+      "step": 4445
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3203770122612397e-05,
+      "loss": 0.4045,
+      "step": 4446
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3202101926766205e-05,
+      "loss": 0.3976,
+      "step": 4447
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.320043373092001e-05,
+      "loss": 0.3883,
+      "step": 4448
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3198765535073819e-05,
+      "loss": 0.4164,
+      "step": 4449
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3197097339227627e-05,
+      "loss": 0.376,
+      "step": 4450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3195429143381433e-05,
+      "loss": 0.4112,
+      "step": 4451
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.319376094753524e-05,
+      "loss": 0.4035,
+      "step": 4452
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.319209275168905e-05,
+      "loss": 0.406,
+      "step": 4453
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3190424555842858e-05,
+      "loss": 0.4081,
+      "step": 4454
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3188756359996664e-05,
+      "loss": 0.3715,
+      "step": 4455
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3187088164150472e-05,
+      "loss": 0.3993,
+      "step": 4456
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.318541996830428e-05,
+      "loss": 0.4405,
+      "step": 4457
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3183751772458086e-05,
+      "loss": 0.3694,
+      "step": 4458
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3182083576611894e-05,
+      "loss": 0.4236,
+      "step": 4459
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3180415380765704e-05,
+      "loss": 0.3907,
+      "step": 4460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3178747184919512e-05,
+      "loss": 0.3802,
+      "step": 4461
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3177078989073318e-05,
+      "loss": 0.3988,
+      "step": 4462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3175410793227126e-05,
+      "loss": 0.3836,
+      "step": 4463
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3173742597380934e-05,
+      "loss": 0.3751,
+      "step": 4464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.317207440153474e-05,
+      "loss": 0.3815,
+      "step": 4465
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3170406205688548e-05,
+      "loss": 0.4118,
+      "step": 4466
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3168738009842357e-05,
+      "loss": 0.3879,
+      "step": 4467
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3167069813996165e-05,
+      "loss": 0.4083,
+      "step": 4468
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3165401618149971e-05,
+      "loss": 0.4042,
+      "step": 4469
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.316373342230378e-05,
+      "loss": 0.409,
+      "step": 4470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3162065226457587e-05,
+      "loss": 0.4037,
+      "step": 4471
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3160397030611397e-05,
+      "loss": 0.4057,
+      "step": 4472
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3158728834765201e-05,
+      "loss": 0.4022,
+      "step": 4473
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.315706063891901e-05,
+      "loss": 0.4184,
+      "step": 4474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3155392443072819e-05,
+      "loss": 0.4308,
+      "step": 4475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3153724247226625e-05,
+      "loss": 0.3755,
+      "step": 4476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3152056051380433e-05,
+      "loss": 0.41,
+      "step": 4477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.315038785553424e-05,
+      "loss": 0.4204,
+      "step": 4478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.314871965968805e-05,
+      "loss": 0.3894,
+      "step": 4479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3147051463841856e-05,
+      "loss": 0.404,
+      "step": 4480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3145383267995664e-05,
+      "loss": 0.4183,
+      "step": 4481
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3143715072149472e-05,
+      "loss": 0.4047,
+      "step": 4482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3142046876303278e-05,
+      "loss": 0.393,
+      "step": 4483
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3140378680457086e-05,
+      "loss": 0.365,
+      "step": 4484
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3138710484610894e-05,
+      "loss": 0.4052,
+      "step": 4485
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3137042288764704e-05,
+      "loss": 0.3898,
+      "step": 4486
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.313537409291851e-05,
+      "loss": 0.4229,
+      "step": 4487
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3133705897072318e-05,
+      "loss": 0.3654,
+      "step": 4488
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3132037701226126e-05,
+      "loss": 0.4109,
+      "step": 4489
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3130369505379932e-05,
+      "loss": 0.4174,
+      "step": 4490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.312870130953374e-05,
+      "loss": 0.4095,
+      "step": 4491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3127033113687548e-05,
+      "loss": 0.3937,
+      "step": 4492
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3125364917841357e-05,
+      "loss": 0.3521,
+      "step": 4493
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3123696721995163e-05,
+      "loss": 0.4066,
+      "step": 4494
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3122028526148971e-05,
+      "loss": 0.3887,
+      "step": 4495
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.312036033030278e-05,
+      "loss": 0.4211,
+      "step": 4496
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3118692134456585e-05,
+      "loss": 0.386,
+      "step": 4497
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3117023938610393e-05,
+      "loss": 0.3861,
+      "step": 4498
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3115355742764201e-05,
+      "loss": 0.4152,
+      "step": 4499
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.311368754691801e-05,
+      "loss": 0.3966,
+      "step": 4500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3112019351071817e-05,
+      "loss": 0.3976,
+      "step": 4501
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3110351155225625e-05,
+      "loss": 0.4281,
+      "step": 4502
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3108682959379433e-05,
+      "loss": 0.4139,
+      "step": 4503
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3107014763533239e-05,
+      "loss": 0.3784,
+      "step": 4504
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3105346567687047e-05,
+      "loss": 0.4082,
+      "step": 4505
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3103678371840856e-05,
+      "loss": 0.4026,
+      "step": 4506
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3102010175994664e-05,
+      "loss": 0.3769,
+      "step": 4507
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.310034198014847e-05,
+      "loss": 0.4204,
+      "step": 4508
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3098673784302278e-05,
+      "loss": 0.4305,
+      "step": 4509
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3097005588456086e-05,
+      "loss": 0.4096,
+      "step": 4510
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3095337392609892e-05,
+      "loss": 0.4392,
+      "step": 4511
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.30936691967637e-05,
+      "loss": 0.3871,
+      "step": 4512
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.309200100091751e-05,
+      "loss": 0.3905,
+      "step": 4513
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3090332805071318e-05,
+      "loss": 0.3953,
+      "step": 4514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3088664609225124e-05,
+      "loss": 0.3905,
+      "step": 4515
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3086996413378932e-05,
+      "loss": 0.3971,
+      "step": 4516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.308532821753274e-05,
+      "loss": 0.3907,
+      "step": 4517
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3083660021686546e-05,
+      "loss": 0.3956,
+      "step": 4518
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3081991825840354e-05,
+      "loss": 0.4119,
+      "step": 4519
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3080323629994163e-05,
+      "loss": 0.3884,
+      "step": 4520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3078655434147971e-05,
+      "loss": 0.3884,
+      "step": 4521
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3076987238301777e-05,
+      "loss": 0.401,
+      "step": 4522
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3075319042455585e-05,
+      "loss": 0.4183,
+      "step": 4523
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3073650846609393e-05,
+      "loss": 0.3996,
+      "step": 4524
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.30719826507632e-05,
+      "loss": 0.3904,
+      "step": 4525
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3070314454917007e-05,
+      "loss": 0.3848,
+      "step": 4526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3068646259070817e-05,
+      "loss": 0.4326,
+      "step": 4527
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3066978063224625e-05,
+      "loss": 0.3747,
+      "step": 4528
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3065309867378431e-05,
+      "loss": 0.4159,
+      "step": 4529
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3063641671532239e-05,
+      "loss": 0.4064,
+      "step": 4530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3061973475686047e-05,
+      "loss": 0.4148,
+      "step": 4531
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3060305279839853e-05,
+      "loss": 0.4151,
+      "step": 4532
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3058637083993661e-05,
+      "loss": 0.3971,
+      "step": 4533
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.305696888814747e-05,
+      "loss": 0.3579,
+      "step": 4534
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3055300692301278e-05,
+      "loss": 0.4251,
+      "step": 4535
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3053632496455085e-05,
+      "loss": 0.3696,
+      "step": 4536
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3051964300608892e-05,
+      "loss": 0.3816,
+      "step": 4537
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.30502961047627e-05,
+      "loss": 0.3929,
+      "step": 4538
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3048627908916506e-05,
+      "loss": 0.3426,
+      "step": 4539
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3046959713070314e-05,
+      "loss": 0.38,
+      "step": 4540
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3045291517224124e-05,
+      "loss": 0.3777,
+      "step": 4541
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3043623321377932e-05,
+      "loss": 0.4108,
+      "step": 4542
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3041955125531738e-05,
+      "loss": 0.3908,
+      "step": 4543
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3040286929685546e-05,
+      "loss": 0.4102,
+      "step": 4544
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3038618733839354e-05,
+      "loss": 0.4348,
+      "step": 4545
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.303695053799316e-05,
+      "loss": 0.3844,
+      "step": 4546
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.303528234214697e-05,
+      "loss": 0.4487,
+      "step": 4547
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3033614146300777e-05,
+      "loss": 0.3753,
+      "step": 4548
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3031945950454585e-05,
+      "loss": 0.3986,
+      "step": 4549
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3030277754608392e-05,
+      "loss": 0.4081,
+      "step": 4550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.30286095587622e-05,
+      "loss": 0.4197,
+      "step": 4551
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3026941362916007e-05,
+      "loss": 0.365,
+      "step": 4552
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3025273167069814e-05,
+      "loss": 0.3915,
+      "step": 4553
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3023604971223623e-05,
+      "loss": 0.4056,
+      "step": 4554
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3021936775377431e-05,
+      "loss": 0.3756,
+      "step": 4555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3020268579531239e-05,
+      "loss": 0.4214,
+      "step": 4556
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3018600383685045e-05,
+      "loss": 0.4089,
+      "step": 4557
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3016932187838853e-05,
+      "loss": 0.4157,
+      "step": 4558
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.301526399199266e-05,
+      "loss": 0.3962,
+      "step": 4559
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3013595796146467e-05,
+      "loss": 0.393,
+      "step": 4560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3011927600300277e-05,
+      "loss": 0.4113,
+      "step": 4561
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3010259404454085e-05,
+      "loss": 0.4209,
+      "step": 4562
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3008591208607892e-05,
+      "loss": 0.4275,
+      "step": 4563
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3006923012761699e-05,
+      "loss": 0.4078,
+      "step": 4564
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3005254816915506e-05,
+      "loss": 0.4027,
+      "step": 4565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3003586621069314e-05,
+      "loss": 0.3832,
+      "step": 4566
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300191842522312e-05,
+      "loss": 0.4098,
+      "step": 4567
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.300025022937693e-05,
+      "loss": 0.3803,
+      "step": 4568
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2998582033530738e-05,
+      "loss": 0.4006,
+      "step": 4569
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2996913837684546e-05,
+      "loss": 0.3967,
+      "step": 4570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2995245641838352e-05,
+      "loss": 0.4069,
+      "step": 4571
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.299357744599216e-05,
+      "loss": 0.3846,
+      "step": 4572
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2991909250145968e-05,
+      "loss": 0.3821,
+      "step": 4573
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2990241054299774e-05,
+      "loss": 0.433,
+      "step": 4574
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2988572858453584e-05,
+      "loss": 0.3928,
+      "step": 4575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2986904662607392e-05,
+      "loss": 0.387,
+      "step": 4576
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.29852364667612e-05,
+      "loss": 0.4053,
+      "step": 4577
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2983568270915006e-05,
+      "loss": 0.4109,
+      "step": 4578
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2981900075068814e-05,
+      "loss": 0.3712,
+      "step": 4579
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2980231879222623e-05,
+      "loss": 0.4005,
+      "step": 4580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2978563683376428e-05,
+      "loss": 0.4199,
+      "step": 4581
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2976895487530237e-05,
+      "loss": 0.4082,
+      "step": 4582
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2975227291684045e-05,
+      "loss": 0.3811,
+      "step": 4583
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2973559095837853e-05,
+      "loss": 0.4219,
+      "step": 4584
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2971890899991659e-05,
+      "loss": 0.4078,
+      "step": 4585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2970222704145467e-05,
+      "loss": 0.4308,
+      "step": 4586
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2968554508299277e-05,
+      "loss": 0.4121,
+      "step": 4587
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2966886312453084e-05,
+      "loss": 0.3973,
+      "step": 4588
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.296521811660689e-05,
+      "loss": 0.4129,
+      "step": 4589
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2963549920760699e-05,
+      "loss": 0.3867,
+      "step": 4590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2961881724914506e-05,
+      "loss": 0.3489,
+      "step": 4591
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.2960213529068313e-05,
+      "loss": 0.3841,
+      "step": 4592
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.295854533322212e-05,
+      "loss": 0.397,
+      "step": 4593
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.295687713737593e-05,
+      "loss": 0.411,
+      "step": 4594
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2955208941529738e-05,
+      "loss": 0.4022,
+      "step": 4595
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2953540745683544e-05,
+      "loss": 0.3735,
+      "step": 4596
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2951872549837352e-05,
+      "loss": 0.3671,
+      "step": 4597
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.295020435399116e-05,
+      "loss": 0.4064,
+      "step": 4598
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2948536158144966e-05,
+      "loss": 0.4046,
+      "step": 4599
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2946867962298774e-05,
+      "loss": 0.3742,
+      "step": 4600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2945199766452584e-05,
+      "loss": 0.3768,
+      "step": 4601
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2943531570606392e-05,
+      "loss": 0.4098,
+      "step": 4602
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2941863374760198e-05,
+      "loss": 0.4095,
+      "step": 4603
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2940195178914006e-05,
+      "loss": 0.4053,
+      "step": 4604
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2938526983067813e-05,
+      "loss": 0.3939,
+      "step": 4605
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.293685878722162e-05,
+      "loss": 0.3795,
+      "step": 4606
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2935190591375428e-05,
+      "loss": 0.3835,
+      "step": 4607
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2933522395529237e-05,
+      "loss": 0.4251,
+      "step": 4608
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2931854199683045e-05,
+      "loss": 0.418,
+      "step": 4609
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2930186003836851e-05,
+      "loss": 0.4077,
+      "step": 4610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2928517807990659e-05,
+      "loss": 0.412,
+      "step": 4611
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2926849612144467e-05,
+      "loss": 0.3833,
+      "step": 4612
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2925181416298273e-05,
+      "loss": 0.4041,
+      "step": 4613
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2923513220452081e-05,
+      "loss": 0.4122,
+      "step": 4614
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.292184502460589e-05,
+      "loss": 0.4144,
+      "step": 4615
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2920176828759699e-05,
+      "loss": 0.372,
+      "step": 4616
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2918508632913505e-05,
+      "loss": 0.4242,
+      "step": 4617
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2916840437067313e-05,
+      "loss": 0.417,
+      "step": 4618
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.291517224122112e-05,
+      "loss": 0.4059,
+      "step": 4619
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2913504045374927e-05,
+      "loss": 0.3526,
+      "step": 4620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2911835849528736e-05,
+      "loss": 0.4114,
+      "step": 4621
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2910167653682544e-05,
+      "loss": 0.3951,
+      "step": 4622
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2908499457836352e-05,
+      "loss": 0.3894,
+      "step": 4623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2906831261990158e-05,
+      "loss": 0.4134,
+      "step": 4624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2905163066143966e-05,
+      "loss": 0.3935,
+      "step": 4625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2903494870297774e-05,
+      "loss": 0.4163,
+      "step": 4626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.290182667445158e-05,
+      "loss": 0.3886,
+      "step": 4627
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.290015847860539e-05,
+      "loss": 0.4218,
+      "step": 4628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2898490282759198e-05,
+      "loss": 0.3977,
+      "step": 4629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2896822086913006e-05,
+      "loss": 0.4298,
+      "step": 4630
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2895153891066812e-05,
+      "loss": 0.4153,
+      "step": 4631
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.289348569522062e-05,
+      "loss": 0.408,
+      "step": 4632
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2891817499374428e-05,
+      "loss": 0.4002,
+      "step": 4633
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2890149303528234e-05,
+      "loss": 0.3918,
+      "step": 4634
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.2888481107682043e-05,
+      "loss": 0.4272,
+      "step": 4635
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2886812911835851e-05,
+      "loss": 0.4276,
+      "step": 4636
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2885144715989659e-05,
+      "loss": 0.4318,
+      "step": 4637
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2883476520143465e-05,
+      "loss": 0.3727,
+      "step": 4638
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2881808324297273e-05,
+      "loss": 0.4281,
+      "step": 4639
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2880140128451081e-05,
+      "loss": 0.3816,
+      "step": 4640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2878471932604887e-05,
+      "loss": 0.3957,
+      "step": 4641
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2876803736758697e-05,
+      "loss": 0.4309,
+      "step": 4642
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2875135540912505e-05,
+      "loss": 0.4328,
+      "step": 4643
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2873467345066313e-05,
+      "loss": 0.4301,
+      "step": 4644
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2871799149220119e-05,
+      "loss": 0.4132,
+      "step": 4645
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2870130953373927e-05,
+      "loss": 0.402,
+      "step": 4646
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2868462757527736e-05,
+      "loss": 0.4035,
+      "step": 4647
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.286679456168154e-05,
+      "loss": 0.4021,
+      "step": 4648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.286512636583535e-05,
+      "loss": 0.3956,
+      "step": 4649
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2863458169989158e-05,
+      "loss": 0.4271,
+      "step": 4650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2861789974142966e-05,
+      "loss": 0.4077,
+      "step": 4651
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2860121778296772e-05,
+      "loss": 0.3827,
+      "step": 4652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.285845358245058e-05,
+      "loss": 0.3918,
+      "step": 4653
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.285678538660439e-05,
+      "loss": 0.4292,
+      "step": 4654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2855117190758194e-05,
+      "loss": 0.3828,
+      "step": 4655
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2853448994912004e-05,
+      "loss": 0.3724,
+      "step": 4656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2851780799065812e-05,
+      "loss": 0.4198,
+      "step": 4657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.285011260321962e-05,
+      "loss": 0.4017,
+      "step": 4658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2848444407373426e-05,
+      "loss": 0.436,
+      "step": 4659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2846776211527234e-05,
+      "loss": 0.3903,
+      "step": 4660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2845108015681043e-05,
+      "loss": 0.4105,
+      "step": 4661
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.284343981983485e-05,
+      "loss": 0.3845,
+      "step": 4662
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2841771623988657e-05,
+      "loss": 0.3845,
+      "step": 4663
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2840103428142465e-05,
+      "loss": 0.4256,
+      "step": 4664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2838435232296273e-05,
+      "loss": 0.3973,
+      "step": 4665
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.283676703645008e-05,
+      "loss": 0.3891,
+      "step": 4666
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2835098840603887e-05,
+      "loss": 0.4513,
+      "step": 4667
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2833430644757697e-05,
+      "loss": 0.4087,
+      "step": 4668
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2831762448911503e-05,
+      "loss": 0.4113,
+      "step": 4669
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2830094253065311e-05,
+      "loss": 0.4203,
+      "step": 4670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2828426057219119e-05,
+      "loss": 0.3931,
+      "step": 4671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2826757861372927e-05,
+      "loss": 0.4022,
+      "step": 4672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2825089665526733e-05,
+      "loss": 0.4126,
+      "step": 4673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.282342146968054e-05,
+      "loss": 0.3886,
+      "step": 4674
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.282175327383435e-05,
+      "loss": 0.389,
+      "step": 4675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2820085077988157e-05,
+      "loss": 0.3784,
+      "step": 4676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2818416882141964e-05,
+      "loss": 0.4016,
+      "step": 4677
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2816748686295772e-05,
+      "loss": 0.4019,
+      "step": 4678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.281508049044958e-05,
+      "loss": 0.3795,
+      "step": 4679
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2813412294603386e-05,
+      "loss": 0.3861,
+      "step": 4680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2811744098757194e-05,
+      "loss": 0.3936,
+      "step": 4681
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2810075902911004e-05,
+      "loss": 0.377,
+      "step": 4682
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.280840770706481e-05,
+      "loss": 0.4148,
+      "step": 4683
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2806739511218618e-05,
+      "loss": 0.4089,
+      "step": 4684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2805071315372426e-05,
+      "loss": 0.3888,
+      "step": 4685
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2803403119526234e-05,
+      "loss": 0.406,
+      "step": 4686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.280173492368004e-05,
+      "loss": 0.3778,
+      "step": 4687
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2800066727833848e-05,
+      "loss": 0.3923,
+      "step": 4688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2798398531987657e-05,
+      "loss": 0.4237,
+      "step": 4689
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2796730336141464e-05,
+      "loss": 0.3996,
+      "step": 4690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2795062140295271e-05,
+      "loss": 0.4126,
+      "step": 4691
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.279339394444908e-05,
+      "loss": 0.4119,
+      "step": 4692
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2791725748602887e-05,
+      "loss": 0.4127,
+      "step": 4693
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2790057552756693e-05,
+      "loss": 0.3766,
+      "step": 4694
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2788389356910503e-05,
+      "loss": 0.4383,
+      "step": 4695
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2786721161064311e-05,
+      "loss": 0.388,
+      "step": 4696
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2785052965218117e-05,
+      "loss": 0.4194,
+      "step": 4697
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2783384769371925e-05,
+      "loss": 0.3941,
+      "step": 4698
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2781716573525733e-05,
+      "loss": 0.4297,
+      "step": 4699
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.278004837767954e-05,
+      "loss": 0.4326,
+      "step": 4700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2778380181833347e-05,
+      "loss": 0.4228,
+      "step": 4701
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2776711985987157e-05,
+      "loss": 0.4016,
+      "step": 4702
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2775043790140964e-05,
+      "loss": 0.4164,
+      "step": 4703
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.277337559429477e-05,
+      "loss": 0.3864,
+      "step": 4704
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2771707398448578e-05,
+      "loss": 0.3896,
+      "step": 4705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2770039202602386e-05,
+      "loss": 0.4082,
+      "step": 4706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2768371006756194e-05,
+      "loss": 0.3837,
+      "step": 4707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.276670281091e-05,
+      "loss": 0.4027,
+      "step": 4708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.276503461506381e-05,
+      "loss": 0.3943,
+      "step": 4709
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2763366419217618e-05,
+      "loss": 0.4011,
+      "step": 4710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2761698223371426e-05,
+      "loss": 0.401,
+      "step": 4711
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2760030027525232e-05,
+      "loss": 0.4123,
+      "step": 4712
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.275836183167904e-05,
+      "loss": 0.3827,
+      "step": 4713
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2756693635832848e-05,
+      "loss": 0.3925,
+      "step": 4714
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2755025439986654e-05,
+      "loss": 0.4117,
+      "step": 4715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2753357244140464e-05,
+      "loss": 0.4351,
+      "step": 4716
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2751689048294271e-05,
+      "loss": 0.4058,
+      "step": 4717
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.275002085244808e-05,
+      "loss": 0.4003,
+      "step": 4718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2748352656601886e-05,
+      "loss": 0.4207,
+      "step": 4719
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2746684460755693e-05,
+      "loss": 0.4081,
+      "step": 4720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2745016264909503e-05,
+      "loss": 0.4134,
+      "step": 4721
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2743348069063307e-05,
+      "loss": 0.4129,
+      "step": 4722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2741679873217117e-05,
+      "loss": 0.418,
+      "step": 4723
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2740011677370925e-05,
+      "loss": 0.4087,
+      "step": 4724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2738343481524733e-05,
+      "loss": 0.4212,
+      "step": 4725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2736675285678539e-05,
+      "loss": 0.446,
+      "step": 4726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2735007089832347e-05,
+      "loss": 0.4483,
+      "step": 4727
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2733338893986156e-05,
+      "loss": 0.3708,
+      "step": 4728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2731670698139961e-05,
+      "loss": 0.4084,
+      "step": 4729
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.273000250229377e-05,
+      "loss": 0.3998,
+      "step": 4730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2728334306447578e-05,
+      "loss": 0.4045,
+      "step": 4731
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2726666110601386e-05,
+      "loss": 0.4162,
+      "step": 4732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2724997914755193e-05,
+      "loss": 0.3858,
+      "step": 4733
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2723329718909e-05,
+      "loss": 0.3869,
+      "step": 4734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.272166152306281e-05,
+      "loss": 0.4329,
+      "step": 4735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2719993327216616e-05,
+      "loss": 0.4001,
+      "step": 4736
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2718325131370424e-05,
+      "loss": 0.3895,
+      "step": 4737
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2716656935524232e-05,
+      "loss": 0.4096,
+      "step": 4738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.271498873967804e-05,
+      "loss": 0.3982,
+      "step": 4739
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2713320543831846e-05,
+      "loss": 0.3977,
+      "step": 4740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2711652347985654e-05,
+      "loss": 0.4085,
+      "step": 4741
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2709984152139464e-05,
+      "loss": 0.4032,
+      "step": 4742
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.270831595629327e-05,
+      "loss": 0.3744,
+      "step": 4743
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2706647760447078e-05,
+      "loss": 0.3961,
+      "step": 4744
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2704979564600885e-05,
+      "loss": 0.3826,
+      "step": 4745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2703311368754693e-05,
+      "loss": 0.373,
+      "step": 4746
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.27016431729085e-05,
+      "loss": 0.3878,
+      "step": 4747
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2699974977062307e-05,
+      "loss": 0.4075,
+      "step": 4748
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2698306781216117e-05,
+      "loss": 0.4393,
+      "step": 4749
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2696638585369923e-05,
+      "loss": 0.4236,
+      "step": 4750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2694970389523731e-05,
+      "loss": 0.3958,
+      "step": 4751
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2693302193677539e-05,
+      "loss": 0.4,
+      "step": 4752
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2691633997831347e-05,
+      "loss": 0.3765,
+      "step": 4753
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2689965801985153e-05,
+      "loss": 0.3543,
+      "step": 4754
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2688297606138961e-05,
+      "loss": 0.4077,
+      "step": 4755
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.268662941029277e-05,
+      "loss": 0.3704,
+      "step": 4756
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2684961214446577e-05,
+      "loss": 0.3929,
+      "step": 4757
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2683293018600385e-05,
+      "loss": 0.4236,
+      "step": 4758
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2681624822754193e-05,
+      "loss": 0.3626,
+      "step": 4759
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2679956626908e-05,
+      "loss": 0.4308,
+      "step": 4760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2678288431061807e-05,
+      "loss": 0.3936,
+      "step": 4761
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2676620235215616e-05,
+      "loss": 0.3912,
+      "step": 4762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2674952039369424e-05,
+      "loss": 0.4005,
+      "step": 4763
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.267328384352323e-05,
+      "loss": 0.4019,
+      "step": 4764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2671615647677038e-05,
+      "loss": 0.4122,
+      "step": 4765
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2669947451830846e-05,
+      "loss": 0.3979,
+      "step": 4766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2668279255984654e-05,
+      "loss": 0.3931,
+      "step": 4767
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.266661106013846e-05,
+      "loss": 0.4163,
+      "step": 4768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.266494286429227e-05,
+      "loss": 0.4025,
+      "step": 4769
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2663274668446078e-05,
+      "loss": 0.3954,
+      "step": 4770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2661606472599884e-05,
+      "loss": 0.3933,
+      "step": 4771
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2659938276753692e-05,
+      "loss": 0.4232,
+      "step": 4772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.26582700809075e-05,
+      "loss": 0.3979,
+      "step": 4773
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2656601885061307e-05,
+      "loss": 0.4034,
+      "step": 4774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2654933689215114e-05,
+      "loss": 0.39,
+      "step": 4775
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2653265493368923e-05,
+      "loss": 0.44,
+      "step": 4776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2651597297522731e-05,
+      "loss": 0.3966,
+      "step": 4777
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2649929101676537e-05,
+      "loss": 0.3832,
+      "step": 4778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2648260905830345e-05,
+      "loss": 0.3818,
+      "step": 4779
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2646592709984153e-05,
+      "loss": 0.3879,
+      "step": 4780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2644924514137961e-05,
+      "loss": 0.373,
+      "step": 4781
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2643256318291767e-05,
+      "loss": 0.3927,
+      "step": 4782
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2641588122445577e-05,
+      "loss": 0.4174,
+      "step": 4783
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2639919926599385e-05,
+      "loss": 0.3999,
+      "step": 4784
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.263825173075319e-05,
+      "loss": 0.3845,
+      "step": 4785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2636583534906999e-05,
+      "loss": 0.4252,
+      "step": 4786
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2634915339060807e-05,
+      "loss": 0.3907,
+      "step": 4787
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2633247143214616e-05,
+      "loss": 0.3911,
+      "step": 4788
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 0.3891,
+      "step": 4789
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262991075152223e-05,
+      "loss": 0.4,
+      "step": 4790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2628242555676038e-05,
+      "loss": 0.397,
+      "step": 4791
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2626574359829844e-05,
+      "loss": 0.3922,
+      "step": 4792
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2624906163983652e-05,
+      "loss": 0.4431,
+      "step": 4793
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262323796813746e-05,
+      "loss": 0.3736,
+      "step": 4794
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262156977229127e-05,
+      "loss": 0.3934,
+      "step": 4795
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2619901576445074e-05,
+      "loss": 0.3794,
+      "step": 4796
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2618233380598884e-05,
+      "loss": 0.3988,
+      "step": 4797
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2616565184752692e-05,
+      "loss": 0.3799,
+      "step": 4798
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2614896988906498e-05,
+      "loss": 0.3753,
+      "step": 4799
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2613228793060306e-05,
+      "loss": 0.3885,
+      "step": 4800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2611560597214114e-05,
+      "loss": 0.4125,
+      "step": 4801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2609892401367923e-05,
+      "loss": 0.374,
+      "step": 4802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.260822420552173e-05,
+      "loss": 0.4013,
+      "step": 4803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2606556009675537e-05,
+      "loss": 0.3763,
+      "step": 4804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2604887813829345e-05,
+      "loss": 0.4348,
+      "step": 4805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2603219617983151e-05,
+      "loss": 0.3969,
+      "step": 4806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.260155142213696e-05,
+      "loss": 0.4414,
+      "step": 4807
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2599883226290767e-05,
+      "loss": 0.3987,
+      "step": 4808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2598215030444577e-05,
+      "loss": 0.4117,
+      "step": 4809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2596546834598383e-05,
+      "loss": 0.4119,
+      "step": 4810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.259487863875219e-05,
+      "loss": 0.4157,
+      "step": 4811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2593210442905999e-05,
+      "loss": 0.4008,
+      "step": 4812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2591542247059805e-05,
+      "loss": 0.3845,
+      "step": 4813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2589874051213613e-05,
+      "loss": 0.4026,
+      "step": 4814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.258820585536742e-05,
+      "loss": 0.3933,
+      "step": 4815
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.258653765952123e-05,
+      "loss": 0.3876,
+      "step": 4816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2584869463675036e-05,
+      "loss": 0.4064,
+      "step": 4817
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2583201267828844e-05,
+      "loss": 0.415,
+      "step": 4818
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2581533071982652e-05,
+      "loss": 0.3837,
+      "step": 4819
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2579864876136458e-05,
+      "loss": 0.3929,
+      "step": 4820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2578196680290266e-05,
+      "loss": 0.3932,
+      "step": 4821
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2576528484444074e-05,
+      "loss": 0.4026,
+      "step": 4822
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2574860288597884e-05,
+      "loss": 0.3649,
+      "step": 4823
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.257319209275169e-05,
+      "loss": 0.3966,
+      "step": 4824
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2571523896905498e-05,
+      "loss": 0.3955,
+      "step": 4825
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2569855701059306e-05,
+      "loss": 0.4084,
+      "step": 4826
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2568187505213114e-05,
+      "loss": 0.391,
+      "step": 4827
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.256651930936692e-05,
+      "loss": 0.4189,
+      "step": 4828
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2564851113520728e-05,
+      "loss": 0.3992,
+      "step": 4829
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2563182917674537e-05,
+      "loss": 0.3932,
+      "step": 4830
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2561514721828343e-05,
+      "loss": 0.423,
+      "step": 4831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2559846525982151e-05,
+      "loss": 0.4166,
+      "step": 4832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.255817833013596e-05,
+      "loss": 0.3992,
+      "step": 4833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2556510134289767e-05,
+      "loss": 0.3783,
+      "step": 4834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2554841938443573e-05,
+      "loss": 0.3984,
+      "step": 4835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2553173742597383e-05,
+      "loss": 0.3986,
+      "step": 4836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.255150554675119e-05,
+      "loss": 0.3837,
+      "step": 4837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2549837350904997e-05,
+      "loss": 0.3725,
+      "step": 4838
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2548169155058805e-05,
+      "loss": 0.3703,
+      "step": 4839
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2546500959212613e-05,
+      "loss": 0.4153,
+      "step": 4840
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.254483276336642e-05,
+      "loss": 0.436,
+      "step": 4841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2543164567520227e-05,
+      "loss": 0.4033,
+      "step": 4842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2541496371674036e-05,
+      "loss": 0.4317,
+      "step": 4843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2539828175827844e-05,
+      "loss": 0.3896,
+      "step": 4844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.253815997998165e-05,
+      "loss": 0.4343,
+      "step": 4845
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2536491784135458e-05,
+      "loss": 0.3932,
+      "step": 4846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2534823588289266e-05,
+      "loss": 0.3914,
+      "step": 4847
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2533155392443074e-05,
+      "loss": 0.408,
+      "step": 4848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.253148719659688e-05,
+      "loss": 0.3955,
+      "step": 4849
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.252981900075069e-05,
+      "loss": 0.3819,
+      "step": 4850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2528150804904498e-05,
+      "loss": 0.3971,
+      "step": 4851
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2526482609058304e-05,
+      "loss": 0.4258,
+      "step": 4852
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2524814413212112e-05,
+      "loss": 0.4221,
+      "step": 4853
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.252314621736592e-05,
+      "loss": 0.3879,
+      "step": 4854
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2521478021519728e-05,
+      "loss": 0.4205,
+      "step": 4855
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2519809825673534e-05,
+      "loss": 0.3708,
+      "step": 4856
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2518141629827343e-05,
+      "loss": 0.4118,
+      "step": 4857
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2516473433981151e-05,
+      "loss": 0.3824,
+      "step": 4858
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2514805238134958e-05,
+      "loss": 0.4103,
+      "step": 4859
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2513137042288765e-05,
+      "loss": 0.4349,
+      "step": 4860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2511468846442573e-05,
+      "loss": 0.4186,
+      "step": 4861
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2509800650596383e-05,
+      "loss": 0.4038,
+      "step": 4862
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2508132454750187e-05,
+      "loss": 0.3876,
+      "step": 4863
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2506464258903997e-05,
+      "loss": 0.4058,
+      "step": 4864
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2504796063057805e-05,
+      "loss": 0.4052,
+      "step": 4865
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2503127867211611e-05,
+      "loss": 0.3862,
+      "step": 4866
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2501459671365419e-05,
+      "loss": 0.3885,
+      "step": 4867
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2499791475519227e-05,
+      "loss": 0.3971,
+      "step": 4868
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2498123279673036e-05,
+      "loss": 0.3869,
+      "step": 4869
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2496455083826841e-05,
+      "loss": 0.4184,
+      "step": 4870
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.249478688798065e-05,
+      "loss": 0.3862,
+      "step": 4871
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2493118692134458e-05,
+      "loss": 0.3801,
+      "step": 4872
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2491450496288265e-05,
+      "loss": 0.4191,
+      "step": 4873
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2489782300442072e-05,
+      "loss": 0.3863,
+      "step": 4874
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.248811410459588e-05,
+      "loss": 0.4144,
+      "step": 4875
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.248644590874969e-05,
+      "loss": 0.3958,
+      "step": 4876
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2484777712903496e-05,
+      "loss": 0.4123,
+      "step": 4877
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2483109517057304e-05,
+      "loss": 0.3925,
+      "step": 4878
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2481441321211112e-05,
+      "loss": 0.4093,
+      "step": 4879
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2479773125364918e-05,
+      "loss": 0.4169,
+      "step": 4880
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2478104929518726e-05,
+      "loss": 0.4017,
+      "step": 4881
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2476436733672534e-05,
+      "loss": 0.4278,
+      "step": 4882
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2474768537826343e-05,
+      "loss": 0.3949,
+      "step": 4883
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.247310034198015e-05,
+      "loss": 0.3764,
+      "step": 4884
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2471432146133957e-05,
+      "loss": 0.4048,
+      "step": 4885
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2469763950287765e-05,
+      "loss": 0.3973,
+      "step": 4886
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2468095754441572e-05,
+      "loss": 0.4123,
+      "step": 4887
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.246642755859538e-05,
+      "loss": 0.4073,
+      "step": 4888
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2464759362749187e-05,
+      "loss": 0.395,
+      "step": 4889
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2463091166902997e-05,
+      "loss": 0.4177,
+      "step": 4890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2461422971056803e-05,
+      "loss": 0.3874,
+      "step": 4891
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2459754775210611e-05,
+      "loss": 0.4114,
+      "step": 4892
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2458086579364419e-05,
+      "loss": 0.4128,
+      "step": 4893
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2456418383518225e-05,
+      "loss": 0.3889,
+      "step": 4894
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2454750187672033e-05,
+      "loss": 0.3756,
+      "step": 4895
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2453081991825841e-05,
+      "loss": 0.3709,
+      "step": 4896
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.245141379597965e-05,
+      "loss": 0.4052,
+      "step": 4897
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2449745600133457e-05,
+      "loss": 0.3875,
+      "step": 4898
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2448077404287265e-05,
+      "loss": 0.3903,
+      "step": 4899
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2446409208441072e-05,
+      "loss": 0.3881,
+      "step": 4900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2444741012594879e-05,
+      "loss": 0.4021,
+      "step": 4901
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2443072816748686e-05,
+      "loss": 0.3722,
+      "step": 4902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2441404620902496e-05,
+      "loss": 0.3959,
+      "step": 4903
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2439736425056304e-05,
+      "loss": 0.3845,
+      "step": 4904
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.243806822921011e-05,
+      "loss": 0.4321,
+      "step": 4905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2436400033363918e-05,
+      "loss": 0.359,
+      "step": 4906
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2434731837517726e-05,
+      "loss": 0.3931,
+      "step": 4907
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2433063641671532e-05,
+      "loss": 0.3922,
+      "step": 4908
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.243139544582534e-05,
+      "loss": 0.3929,
+      "step": 4909
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.242972724997915e-05,
+      "loss": 0.4039,
+      "step": 4910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2428059054132957e-05,
+      "loss": 0.3889,
+      "step": 4911
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2426390858286764e-05,
+      "loss": 0.3847,
+      "step": 4912
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2424722662440572e-05,
+      "loss": 0.3935,
+      "step": 4913
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.242305446659438e-05,
+      "loss": 0.3816,
+      "step": 4914
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2421386270748186e-05,
+      "loss": 0.4165,
+      "step": 4915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2419718074901994e-05,
+      "loss": 0.4152,
+      "step": 4916
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2418049879055803e-05,
+      "loss": 0.3855,
+      "step": 4917
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2416381683209611e-05,
+      "loss": 0.3951,
+      "step": 4918
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2414713487363417e-05,
+      "loss": 0.4042,
+      "step": 4919
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2413045291517225e-05,
+      "loss": 0.4031,
+      "step": 4920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2411377095671033e-05,
+      "loss": 0.4111,
+      "step": 4921
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2409708899824839e-05,
+      "loss": 0.4088,
+      "step": 4922
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2408040703978647e-05,
+      "loss": 0.381,
+      "step": 4923
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2406372508132457e-05,
+      "loss": 0.3914,
+      "step": 4924
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2404704312286265e-05,
+      "loss": 0.4081,
+      "step": 4925
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.240303611644007e-05,
+      "loss": 0.3975,
+      "step": 4926
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2401367920593879e-05,
+      "loss": 0.3507,
+      "step": 4927
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2399699724747686e-05,
+      "loss": 0.3986,
+      "step": 4928
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2398031528901493e-05,
+      "loss": 0.3912,
+      "step": 4929
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.23963633330553e-05,
+      "loss": 0.409,
+      "step": 4930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.239469513720911e-05,
+      "loss": 0.3862,
+      "step": 4931
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2393026941362918e-05,
+      "loss": 0.399,
+      "step": 4932
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2391358745516724e-05,
+      "loss": 0.4346,
+      "step": 4933
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2389690549670532e-05,
+      "loss": 0.3874,
+      "step": 4934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.238802235382434e-05,
+      "loss": 0.3797,
+      "step": 4935
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2386354157978146e-05,
+      "loss": 0.3799,
+      "step": 4936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2384685962131954e-05,
+      "loss": 0.3777,
+      "step": 4937
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2383017766285764e-05,
+      "loss": 0.4128,
+      "step": 4938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2381349570439572e-05,
+      "loss": 0.3835,
+      "step": 4939
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2379681374593378e-05,
+      "loss": 0.3961,
+      "step": 4940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2378013178747186e-05,
+      "loss": 0.4253,
+      "step": 4941
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2376344982900993e-05,
+      "loss": 0.3965,
+      "step": 4942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2374676787054803e-05,
+      "loss": 0.3651,
+      "step": 4943
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2373008591208608e-05,
+      "loss": 0.4413,
+      "step": 4944
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2371340395362417e-05,
+      "loss": 0.3957,
+      "step": 4945
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2369672199516225e-05,
+      "loss": 0.3605,
+      "step": 4946
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2368004003670031e-05,
+      "loss": 0.4357,
+      "step": 4947
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2366335807823839e-05,
+      "loss": 0.3698,
+      "step": 4948
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2364667611977647e-05,
+      "loss": 0.4292,
+      "step": 4949
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2362999416131457e-05,
+      "loss": 0.3885,
+      "step": 4950
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2361331220285263e-05,
+      "loss": 0.3866,
+      "step": 4951
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.235966302443907e-05,
+      "loss": 0.3994,
+      "step": 4952
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2357994828592879e-05,
+      "loss": 0.4304,
+      "step": 4953
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2356326632746685e-05,
+      "loss": 0.4103,
+      "step": 4954
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2354658436900493e-05,
+      "loss": 0.4045,
+      "step": 4955
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.23529902410543e-05,
+      "loss": 0.4198,
+      "step": 4956
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.235132204520811e-05,
+      "loss": 0.4182,
+      "step": 4957
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2349653849361916e-05,
+      "loss": 0.3715,
+      "step": 4958
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2347985653515724e-05,
+      "loss": 0.4071,
+      "step": 4959
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2346317457669532e-05,
+      "loss": 0.3928,
+      "step": 4960
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2344649261823338e-05,
+      "loss": 0.4004,
+      "step": 4961
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2342981065977146e-05,
+      "loss": 0.4504,
+      "step": 4962
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2341312870130954e-05,
+      "loss": 0.398,
+      "step": 4963
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2339644674284764e-05,
+      "loss": 0.3746,
+      "step": 4964
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.233797647843857e-05,
+      "loss": 0.365,
+      "step": 4965
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2336308282592378e-05,
+      "loss": 0.4021,
+      "step": 4966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2334640086746186e-05,
+      "loss": 0.3945,
+      "step": 4967
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2332971890899992e-05,
+      "loss": 0.374,
+      "step": 4968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.23313036950538e-05,
+      "loss": 0.3953,
+      "step": 4969
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2329635499207608e-05,
+      "loss": 0.4091,
+      "step": 4970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2327967303361417e-05,
+      "loss": 0.4028,
+      "step": 4971
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2326299107515223e-05,
+      "loss": 0.4145,
+      "step": 4972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2324630911669031e-05,
+      "loss": 0.3869,
+      "step": 4973
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2322962715822839e-05,
+      "loss": 0.3882,
+      "step": 4974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2321294519976645e-05,
+      "loss": 0.408,
+      "step": 4975
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2319626324130453e-05,
+      "loss": 0.375,
+      "step": 4976
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2317958128284263e-05,
+      "loss": 0.4232,
+      "step": 4977
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.231628993243807e-05,
+      "loss": 0.4332,
+      "step": 4978
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2314621736591877e-05,
+      "loss": 0.3965,
+      "step": 4979
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2312953540745685e-05,
+      "loss": 0.3993,
+      "step": 4980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2311285344899493e-05,
+      "loss": 0.3847,
+      "step": 4981
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2309617149053299e-05,
+      "loss": 0.382,
+      "step": 4982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2307948953207107e-05,
+      "loss": 0.3894,
+      "step": 4983
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2306280757360916e-05,
+      "loss": 0.39,
+      "step": 4984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2304612561514724e-05,
+      "loss": 0.4174,
+      "step": 4985
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.230294436566853e-05,
+      "loss": 0.3702,
+      "step": 4986
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2301276169822338e-05,
+      "loss": 0.3984,
+      "step": 4987
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2299607973976146e-05,
+      "loss": 0.4144,
+      "step": 4988
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2297939778129952e-05,
+      "loss": 0.3903,
+      "step": 4989
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.229627158228376e-05,
+      "loss": 0.4495,
+      "step": 4990
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.229460338643757e-05,
+      "loss": 0.4241,
+      "step": 4991
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2292935190591378e-05,
+      "loss": 0.3963,
+      "step": 4992
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2291266994745184e-05,
+      "loss": 0.3874,
+      "step": 4993
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2289598798898992e-05,
+      "loss": 0.3761,
+      "step": 4994
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.22879306030528e-05,
+      "loss": 0.4135,
+      "step": 4995
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2286262407206606e-05,
+      "loss": 0.4014,
+      "step": 4996
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2284594211360414e-05,
+      "loss": 0.4131,
+      "step": 4997
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2282926015514223e-05,
+      "loss": 0.4086,
+      "step": 4998
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2281257819668031e-05,
+      "loss": 0.4111,
+      "step": 4999
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2279589623821837e-05,
+      "loss": 0.4198,
+      "step": 5000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2277921427975645e-05,
+      "loss": 0.4157,
+      "step": 5001
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2276253232129453e-05,
+      "loss": 0.4004,
+      "step": 5002
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.227458503628326e-05,
+      "loss": 0.4121,
+      "step": 5003
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2272916840437067e-05,
+      "loss": 0.3832,
+      "step": 5004
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2271248644590877e-05,
+      "loss": 0.4162,
+      "step": 5005
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2269580448744685e-05,
+      "loss": 0.4234,
+      "step": 5006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2267912252898491e-05,
+      "loss": 0.4153,
+      "step": 5007
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2266244057052299e-05,
+      "loss": 0.4116,
+      "step": 5008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2264575861206107e-05,
+      "loss": 0.4245,
+      "step": 5009
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2262907665359913e-05,
+      "loss": 0.4184,
+      "step": 5010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.226123946951372e-05,
+      "loss": 0.4035,
+      "step": 5011
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.225957127366753e-05,
+      "loss": 0.4007,
+      "step": 5012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2257903077821338e-05,
+      "loss": 0.4063,
+      "step": 5013
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2256234881975144e-05,
+      "loss": 0.3692,
+      "step": 5014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2254566686128952e-05,
+      "loss": 0.3772,
+      "step": 5015
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.225289849028276e-05,
+      "loss": 0.4412,
+      "step": 5016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2251230294436566e-05,
+      "loss": 0.3754,
+      "step": 5017
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2249562098590376e-05,
+      "loss": 0.3981,
+      "step": 5018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2247893902744184e-05,
+      "loss": 0.3932,
+      "step": 5019
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2246225706897992e-05,
+      "loss": 0.3894,
+      "step": 5020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2244557511051798e-05,
+      "loss": 0.4172,
+      "step": 5021
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2242889315205606e-05,
+      "loss": 0.4147,
+      "step": 5022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2241221119359414e-05,
+      "loss": 0.3864,
+      "step": 5023
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.223955292351322e-05,
+      "loss": 0.4076,
+      "step": 5024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.223788472766703e-05,
+      "loss": 0.3848,
+      "step": 5025
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2236216531820837e-05,
+      "loss": 0.4087,
+      "step": 5026
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2234548335974645e-05,
+      "loss": 0.405,
+      "step": 5027
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2232880140128451e-05,
+      "loss": 0.3818,
+      "step": 5028
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.223121194428226e-05,
+      "loss": 0.3722,
+      "step": 5029
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2229543748436067e-05,
+      "loss": 0.4331,
+      "step": 5030
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2227875552589873e-05,
+      "loss": 0.3857,
+      "step": 5031
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2226207356743683e-05,
+      "loss": 0.4007,
+      "step": 5032
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2224539160897491e-05,
+      "loss": 0.4006,
+      "step": 5033
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2222870965051299e-05,
+      "loss": 0.3861,
+      "step": 5034
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2221202769205105e-05,
+      "loss": 0.4071,
+      "step": 5035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2219534573358913e-05,
+      "loss": 0.3774,
+      "step": 5036
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.221786637751272e-05,
+      "loss": 0.3886,
+      "step": 5037
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2216198181666527e-05,
+      "loss": 0.4071,
+      "step": 5038
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2214529985820337e-05,
+      "loss": 0.3959,
+      "step": 5039
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2212861789974144e-05,
+      "loss": 0.3838,
+      "step": 5040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2211193594127952e-05,
+      "loss": 0.3931,
+      "step": 5041
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2209525398281758e-05,
+      "loss": 0.385,
+      "step": 5042
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2207857202435566e-05,
+      "loss": 0.3875,
+      "step": 5043
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2206189006589376e-05,
+      "loss": 0.4184,
+      "step": 5044
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.220452081074318e-05,
+      "loss": 0.3913,
+      "step": 5045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.220285261489699e-05,
+      "loss": 0.419,
+      "step": 5046
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2201184419050798e-05,
+      "loss": 0.4197,
+      "step": 5047
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2199516223204606e-05,
+      "loss": 0.3889,
+      "step": 5048
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2197848027358412e-05,
+      "loss": 0.3886,
+      "step": 5049
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.219617983151222e-05,
+      "loss": 0.4065,
+      "step": 5050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.219451163566603e-05,
+      "loss": 0.4214,
+      "step": 5051
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2192843439819834e-05,
+      "loss": 0.3814,
+      "step": 5052
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2191175243973644e-05,
+      "loss": 0.4303,
+      "step": 5053
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2189507048127451e-05,
+      "loss": 0.4094,
+      "step": 5054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.218783885228126e-05,
+      "loss": 0.4059,
+      "step": 5055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2186170656435066e-05,
+      "loss": 0.3812,
+      "step": 5056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2184502460588873e-05,
+      "loss": 0.3933,
+      "step": 5057
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2182834264742683e-05,
+      "loss": 0.3964,
+      "step": 5058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.218116606889649e-05,
+      "loss": 0.4115,
+      "step": 5059
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2179497873050297e-05,
+      "loss": 0.3964,
+      "step": 5060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2177829677204105e-05,
+      "loss": 0.413,
+      "step": 5061
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2176161481357913e-05,
+      "loss": 0.3858,
+      "step": 5062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2174493285511719e-05,
+      "loss": 0.3887,
+      "step": 5063
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2172825089665527e-05,
+      "loss": 0.429,
+      "step": 5064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2171156893819337e-05,
+      "loss": 0.4095,
+      "step": 5065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2169488697973144e-05,
+      "loss": 0.3971,
+      "step": 5066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.216782050212695e-05,
+      "loss": 0.4366,
+      "step": 5067
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2166152306280758e-05,
+      "loss": 0.3975,
+      "step": 5068
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2164484110434566e-05,
+      "loss": 0.3974,
+      "step": 5069
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2162815914588373e-05,
+      "loss": 0.4065,
+      "step": 5070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.216114771874218e-05,
+      "loss": 0.3903,
+      "step": 5071
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.215947952289599e-05,
+      "loss": 0.4284,
+      "step": 5072
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2157811327049798e-05,
+      "loss": 0.3856,
+      "step": 5073
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2156143131203604e-05,
+      "loss": 0.3988,
+      "step": 5074
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2154474935357412e-05,
+      "loss": 0.3948,
+      "step": 5075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.215280673951122e-05,
+      "loss": 0.3907,
+      "step": 5076
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2151138543665026e-05,
+      "loss": 0.4081,
+      "step": 5077
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2149470347818834e-05,
+      "loss": 0.386,
+      "step": 5078
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2147802151972644e-05,
+      "loss": 0.4139,
+      "step": 5079
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2146133956126451e-05,
+      "loss": 0.3827,
+      "step": 5080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2144465760280258e-05,
+      "loss": 0.4127,
+      "step": 5081
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2142797564434066e-05,
+      "loss": 0.3903,
+      "step": 5082
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2141129368587873e-05,
+      "loss": 0.3871,
+      "step": 5083
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.213946117274168e-05,
+      "loss": 0.407,
+      "step": 5084
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2137792976895487e-05,
+      "loss": 0.4327,
+      "step": 5085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2136124781049297e-05,
+      "loss": 0.3981,
+      "step": 5086
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2134456585203105e-05,
+      "loss": 0.3932,
+      "step": 5087
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2132788389356911e-05,
+      "loss": 0.3888,
+      "step": 5088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2131120193510719e-05,
+      "loss": 0.4097,
+      "step": 5089
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2129451997664527e-05,
+      "loss": 0.4337,
+      "step": 5090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2127783801818333e-05,
+      "loss": 0.411,
+      "step": 5091
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2126115605972143e-05,
+      "loss": 0.4242,
+      "step": 5092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.212444741012595e-05,
+      "loss": 0.4062,
+      "step": 5093
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2122779214279758e-05,
+      "loss": 0.3928,
+      "step": 5094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2121111018433565e-05,
+      "loss": 0.3734,
+      "step": 5095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2119442822587373e-05,
+      "loss": 0.3936,
+      "step": 5096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.211777462674118e-05,
+      "loss": 0.3865,
+      "step": 5097
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2116106430894987e-05,
+      "loss": 0.3999,
+      "step": 5098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2114438235048796e-05,
+      "loss": 0.4137,
+      "step": 5099
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2112770039202604e-05,
+      "loss": 0.3886,
+      "step": 5100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2111101843356412e-05,
+      "loss": 0.3921,
+      "step": 5101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2109433647510218e-05,
+      "loss": 0.4024,
+      "step": 5102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2107765451664026e-05,
+      "loss": 0.3897,
+      "step": 5103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2106097255817834e-05,
+      "loss": 0.393,
+      "step": 5104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.210442905997164e-05,
+      "loss": 0.3824,
+      "step": 5105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.210276086412545e-05,
+      "loss": 0.428,
+      "step": 5106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2101092668279258e-05,
+      "loss": 0.3929,
+      "step": 5107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2099424472433065e-05,
+      "loss": 0.3795,
+      "step": 5108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2097756276586872e-05,
+      "loss": 0.4053,
+      "step": 5109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.209608808074068e-05,
+      "loss": 0.3719,
+      "step": 5110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2094419884894487e-05,
+      "loss": 0.3996,
+      "step": 5111
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2092751689048294e-05,
+      "loss": 0.4149,
+      "step": 5112
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2091083493202103e-05,
+      "loss": 0.3885,
+      "step": 5113
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2089415297355911e-05,
+      "loss": 0.4445,
+      "step": 5114
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2087747101509719e-05,
+      "loss": 0.4161,
+      "step": 5115
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2086078905663525e-05,
+      "loss": 0.4063,
+      "step": 5116
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2084410709817333e-05,
+      "loss": 0.4212,
+      "step": 5117
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2082742513971143e-05,
+      "loss": 0.4227,
+      "step": 5118
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2081074318124947e-05,
+      "loss": 0.4123,
+      "step": 5119
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2079406122278757e-05,
+      "loss": 0.4027,
+      "step": 5120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2077737926432565e-05,
+      "loss": 0.3809,
+      "step": 5121
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2076069730586373e-05,
+      "loss": 0.4133,
+      "step": 5122
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2074401534740179e-05,
+      "loss": 0.3883,
+      "step": 5123
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2072733338893987e-05,
+      "loss": 0.4059,
+      "step": 5124
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2071065143047796e-05,
+      "loss": 0.3936,
+      "step": 5125
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.20693969472016e-05,
+      "loss": 0.3957,
+      "step": 5126
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.206772875135541e-05,
+      "loss": 0.3843,
+      "step": 5127
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2066060555509218e-05,
+      "loss": 0.4177,
+      "step": 5128
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2064392359663026e-05,
+      "loss": 0.4165,
+      "step": 5129
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2062724163816832e-05,
+      "loss": 0.3731,
+      "step": 5130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.206105596797064e-05,
+      "loss": 0.4009,
+      "step": 5131
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.205938777212445e-05,
+      "loss": 0.3873,
+      "step": 5132
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2057719576278256e-05,
+      "loss": 0.4078,
+      "step": 5133
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2056051380432064e-05,
+      "loss": 0.3753,
+      "step": 5134
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2054383184585872e-05,
+      "loss": 0.398,
+      "step": 5135
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.205271498873968e-05,
+      "loss": 0.3952,
+      "step": 5136
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2051046792893486e-05,
+      "loss": 0.3999,
+      "step": 5137
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2049378597047294e-05,
+      "loss": 0.3853,
+      "step": 5138
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2047710401201103e-05,
+      "loss": 0.399,
+      "step": 5139
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.204604220535491e-05,
+      "loss": 0.4235,
+      "step": 5140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2044374009508717e-05,
+      "loss": 0.3712,
+      "step": 5141
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2042705813662525e-05,
+      "loss": 0.3879,
+      "step": 5142
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2041037617816333e-05,
+      "loss": 0.4061,
+      "step": 5143
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.203936942197014e-05,
+      "loss": 0.4141,
+      "step": 5144
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2037701226123947e-05,
+      "loss": 0.3804,
+      "step": 5145
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2036033030277757e-05,
+      "loss": 0.4484,
+      "step": 5146
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2034364834431563e-05,
+      "loss": 0.3967,
+      "step": 5147
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.203269663858537e-05,
+      "loss": 0.3832,
+      "step": 5148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2031028442739179e-05,
+      "loss": 0.3895,
+      "step": 5149
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2029360246892987e-05,
+      "loss": 0.4133,
+      "step": 5150
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2027692051046793e-05,
+      "loss": 0.3596,
+      "step": 5151
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.20260238552006e-05,
+      "loss": 0.3728,
+      "step": 5152
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.202435565935441e-05,
+      "loss": 0.4123,
+      "step": 5153
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2022687463508216e-05,
+      "loss": 0.3967,
+      "step": 5154
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2021019267662024e-05,
+      "loss": 0.3999,
+      "step": 5155
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2019351071815832e-05,
+      "loss": 0.3699,
+      "step": 5156
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.201768287596964e-05,
+      "loss": 0.4163,
+      "step": 5157
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2016014680123446e-05,
+      "loss": 0.3919,
+      "step": 5158
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2014346484277256e-05,
+      "loss": 0.4157,
+      "step": 5159
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2012678288431064e-05,
+      "loss": 0.4034,
+      "step": 5160
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.201101009258487e-05,
+      "loss": 0.3898,
+      "step": 5161
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2009341896738678e-05,
+      "loss": 0.3913,
+      "step": 5162
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2007673700892486e-05,
+      "loss": 0.4286,
+      "step": 5163
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2006005505046294e-05,
+      "loss": 0.3946,
+      "step": 5164
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.20043373092001e-05,
+      "loss": 0.3844,
+      "step": 5165
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.200266911335391e-05,
+      "loss": 0.3873,
+      "step": 5166
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2001000917507717e-05,
+      "loss": 0.3904,
+      "step": 5167
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1999332721661523e-05,
+      "loss": 0.4276,
+      "step": 5168
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1997664525815331e-05,
+      "loss": 0.409,
+      "step": 5169
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.199599632996914e-05,
+      "loss": 0.3659,
+      "step": 5170
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1994328134122947e-05,
+      "loss": 0.4421,
+      "step": 5171
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1992659938276753e-05,
+      "loss": 0.3937,
+      "step": 5172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1990991742430563e-05,
+      "loss": 0.4315,
+      "step": 5173
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.198932354658437e-05,
+      "loss": 0.3922,
+      "step": 5174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1987655350738177e-05,
+      "loss": 0.4147,
+      "step": 5175
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1985987154891985e-05,
+      "loss": 0.3746,
+      "step": 5176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1984318959045793e-05,
+      "loss": 0.4455,
+      "step": 5177
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.19826507631996e-05,
+      "loss": 0.4275,
+      "step": 5178
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1980982567353407e-05,
+      "loss": 0.4065,
+      "step": 5179
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1979314371507216e-05,
+      "loss": 0.4371,
+      "step": 5180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1977646175661024e-05,
+      "loss": 0.3866,
+      "step": 5181
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1975977979814832e-05,
+      "loss": 0.4107,
+      "step": 5182
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1974309783968638e-05,
+      "loss": 0.4384,
+      "step": 5183
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1972641588122446e-05,
+      "loss": 0.3955,
+      "step": 5184
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1970973392276254e-05,
+      "loss": 0.4037,
+      "step": 5185
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.196930519643006e-05,
+      "loss": 0.4017,
+      "step": 5186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.196763700058387e-05,
+      "loss": 0.3749,
+      "step": 5187
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1965968804737678e-05,
+      "loss": 0.402,
+      "step": 5188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1964300608891486e-05,
+      "loss": 0.4211,
+      "step": 5189
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1962632413045292e-05,
+      "loss": 0.4298,
+      "step": 5190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.19609642171991e-05,
+      "loss": 0.4006,
+      "step": 5191
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.195929602135291e-05,
+      "loss": 0.4098,
+      "step": 5192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1957627825506714e-05,
+      "loss": 0.4255,
+      "step": 5193
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1955959629660523e-05,
+      "loss": 0.4135,
+      "step": 5194
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1954291433814331e-05,
+      "loss": 0.3794,
+      "step": 5195
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.195262323796814e-05,
+      "loss": 0.4021,
+      "step": 5196
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1950955042121945e-05,
+      "loss": 0.3728,
+      "step": 5197
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1949286846275753e-05,
+      "loss": 0.3962,
+      "step": 5198
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1947618650429563e-05,
+      "loss": 0.3726,
+      "step": 5199
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1945950454583367e-05,
+      "loss": 0.4281,
+      "step": 5200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1944282258737177e-05,
+      "loss": 0.3933,
+      "step": 5201
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1942614062890985e-05,
+      "loss": 0.3898,
+      "step": 5202
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1940945867044793e-05,
+      "loss": 0.417,
+      "step": 5203
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1939277671198599e-05,
+      "loss": 0.3921,
+      "step": 5204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1937609475352407e-05,
+      "loss": 0.4145,
+      "step": 5205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1935941279506216e-05,
+      "loss": 0.406,
+      "step": 5206
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1934273083660023e-05,
+      "loss": 0.394,
+      "step": 5207
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.193260488781383e-05,
+      "loss": 0.38,
+      "step": 5208
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1930936691967638e-05,
+      "loss": 0.4174,
+      "step": 5209
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1929268496121446e-05,
+      "loss": 0.3997,
+      "step": 5210
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1927600300275252e-05,
+      "loss": 0.4513,
+      "step": 5211
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.192593210442906e-05,
+      "loss": 0.3804,
+      "step": 5212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.192426390858287e-05,
+      "loss": 0.4074,
+      "step": 5213
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1922595712736676e-05,
+      "loss": 0.404,
+      "step": 5214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1920927516890484e-05,
+      "loss": 0.3926,
+      "step": 5215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1919259321044292e-05,
+      "loss": 0.414,
+      "step": 5216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.19175911251981e-05,
+      "loss": 0.3764,
+      "step": 5217
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1915922929351906e-05,
+      "loss": 0.3835,
+      "step": 5218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1914254733505714e-05,
+      "loss": 0.4272,
+      "step": 5219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1912586537659523e-05,
+      "loss": 0.4071,
+      "step": 5220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.191091834181333e-05,
+      "loss": 0.3978,
+      "step": 5221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1909250145967138e-05,
+      "loss": 0.415,
+      "step": 5222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1907581950120945e-05,
+      "loss": 0.4138,
+      "step": 5223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1905913754274753e-05,
+      "loss": 0.417,
+      "step": 5224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.190424555842856e-05,
+      "loss": 0.3943,
+      "step": 5225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1902577362582367e-05,
+      "loss": 0.4175,
+      "step": 5226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1900909166736177e-05,
+      "loss": 0.3928,
+      "step": 5227
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1899240970889983e-05,
+      "loss": 0.3978,
+      "step": 5228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1897572775043791e-05,
+      "loss": 0.4077,
+      "step": 5229
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1895904579197599e-05,
+      "loss": 0.4164,
+      "step": 5230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1894236383351407e-05,
+      "loss": 0.4079,
+      "step": 5231
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1892568187505213e-05,
+      "loss": 0.3908,
+      "step": 5232
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1890899991659023e-05,
+      "loss": 0.4061,
+      "step": 5233
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.188923179581283e-05,
+      "loss": 0.4263,
+      "step": 5234
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1887563599966637e-05,
+      "loss": 0.423,
+      "step": 5235
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1885895404120445e-05,
+      "loss": 0.3685,
+      "step": 5236
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1884227208274252e-05,
+      "loss": 0.4185,
+      "step": 5237
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.188255901242806e-05,
+      "loss": 0.4093,
+      "step": 5238
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1880890816581866e-05,
+      "loss": 0.4255,
+      "step": 5239
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1879222620735676e-05,
+      "loss": 0.4125,
+      "step": 5240
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1877554424889484e-05,
+      "loss": 0.4204,
+      "step": 5241
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.187588622904329e-05,
+      "loss": 0.3844,
+      "step": 5242
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1874218033197098e-05,
+      "loss": 0.3976,
+      "step": 5243
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1872549837350906e-05,
+      "loss": 0.4379,
+      "step": 5244
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1870881641504714e-05,
+      "loss": 0.363,
+      "step": 5245
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.186921344565852e-05,
+      "loss": 0.3825,
+      "step": 5246
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.186754524981233e-05,
+      "loss": 0.3979,
+      "step": 5247
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1865877053966137e-05,
+      "loss": 0.4069,
+      "step": 5248
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1864208858119944e-05,
+      "loss": 0.3977,
+      "step": 5249
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1862540662273752e-05,
+      "loss": 0.3758,
+      "step": 5250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.186087246642756e-05,
+      "loss": 0.3889,
+      "step": 5251
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1859204270581367e-05,
+      "loss": 0.3822,
+      "step": 5252
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1857536074735174e-05,
+      "loss": 0.3938,
+      "step": 5253
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1855867878888983e-05,
+      "loss": 0.3808,
+      "step": 5254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1854199683042791e-05,
+      "loss": 0.4045,
+      "step": 5255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1852531487196597e-05,
+      "loss": 0.407,
+      "step": 5256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1850863291350405e-05,
+      "loss": 0.3954,
+      "step": 5257
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1849195095504213e-05,
+      "loss": 0.3882,
+      "step": 5258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1847526899658023e-05,
+      "loss": 0.4138,
+      "step": 5259
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1845858703811827e-05,
+      "loss": 0.3794,
+      "step": 5260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1844190507965637e-05,
+      "loss": 0.384,
+      "step": 5261
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1842522312119445e-05,
+      "loss": 0.3853,
+      "step": 5262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.184085411627325e-05,
+      "loss": 0.4275,
+      "step": 5263
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1839185920427059e-05,
+      "loss": 0.4116,
+      "step": 5264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1837517724580866e-05,
+      "loss": 0.3889,
+      "step": 5265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1835849528734676e-05,
+      "loss": 0.3833,
+      "step": 5266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.183418133288848e-05,
+      "loss": 0.3892,
+      "step": 5267
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.183251313704229e-05,
+      "loss": 0.3946,
+      "step": 5268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1830844941196098e-05,
+      "loss": 0.4071,
+      "step": 5269
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1829176745349904e-05,
+      "loss": 0.4158,
+      "step": 5270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1827508549503712e-05,
+      "loss": 0.3879,
+      "step": 5271
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.182584035365752e-05,
+      "loss": 0.4003,
+      "step": 5272
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.182417215781133e-05,
+      "loss": 0.3898,
+      "step": 5273
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1822503961965136e-05,
+      "loss": 0.4109,
+      "step": 5274
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1820835766118944e-05,
+      "loss": 0.3918,
+      "step": 5275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1819167570272752e-05,
+      "loss": 0.4154,
+      "step": 5276
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1817499374426558e-05,
+      "loss": 0.3866,
+      "step": 5277
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1815831178580366e-05,
+      "loss": 0.4176,
+      "step": 5278
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1814162982734174e-05,
+      "loss": 0.38,
+      "step": 5279
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1812494786887983e-05,
+      "loss": 0.3991,
+      "step": 5280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.181082659104179e-05,
+      "loss": 0.4153,
+      "step": 5281
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1809158395195597e-05,
+      "loss": 0.3776,
+      "step": 5282
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1807490199349405e-05,
+      "loss": 0.4346,
+      "step": 5283
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1805822003503211e-05,
+      "loss": 0.4158,
+      "step": 5284
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1804153807657019e-05,
+      "loss": 0.4031,
+      "step": 5285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1802485611810827e-05,
+      "loss": 0.4241,
+      "step": 5286
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1800817415964637e-05,
+      "loss": 0.3953,
+      "step": 5287
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1799149220118443e-05,
+      "loss": 0.4021,
+      "step": 5288
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.179748102427225e-05,
+      "loss": 0.4183,
+      "step": 5289
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1795812828426059e-05,
+      "loss": 0.3946,
+      "step": 5290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1794144632579865e-05,
+      "loss": 0.4033,
+      "step": 5291
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.1792476436733673e-05,
+      "loss": 0.4278,
+      "step": 5292
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.179080824088748e-05,
+      "loss": 0.3877,
+      "step": 5293
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.178914004504129e-05,
+      "loss": 0.3915,
+      "step": 5294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1787471849195096e-05,
+      "loss": 0.4065,
+      "step": 5295
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1785803653348904e-05,
+      "loss": 0.3766,
+      "step": 5296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1784135457502712e-05,
+      "loss": 0.4157,
+      "step": 5297
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1782467261656518e-05,
+      "loss": 0.439,
+      "step": 5298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1780799065810326e-05,
+      "loss": 0.4023,
+      "step": 5299
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1779130869964136e-05,
+      "loss": 0.3969,
+      "step": 5300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1777462674117944e-05,
+      "loss": 0.4051,
+      "step": 5301
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.177579447827175e-05,
+      "loss": 0.4154,
+      "step": 5302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1774126282425558e-05,
+      "loss": 0.4334,
+      "step": 5303
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1772458086579366e-05,
+      "loss": 0.385,
+      "step": 5304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1770789890733173e-05,
+      "loss": 0.4213,
+      "step": 5305
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.176912169488698e-05,
+      "loss": 0.3908,
+      "step": 5306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.176745349904079e-05,
+      "loss": 0.3982,
+      "step": 5307
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1765785303194597e-05,
+      "loss": 0.3769,
+      "step": 5308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1764117107348403e-05,
+      "loss": 0.3821,
+      "step": 5309
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1762448911502211e-05,
+      "loss": 0.409,
+      "step": 5310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1760780715656019e-05,
+      "loss": 0.4129,
+      "step": 5311
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1759112519809827e-05,
+      "loss": 0.3996,
+      "step": 5312
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1757444323963633e-05,
+      "loss": 0.4046,
+      "step": 5313
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1755776128117443e-05,
+      "loss": 0.3835,
+      "step": 5314
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.175410793227125e-05,
+      "loss": 0.3943,
+      "step": 5315
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1752439736425057e-05,
+      "loss": 0.4091,
+      "step": 5316
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1750771540578865e-05,
+      "loss": 0.3758,
+      "step": 5317
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1749103344732673e-05,
+      "loss": 0.4074,
+      "step": 5318
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.174743514888648e-05,
+      "loss": 0.4112,
+      "step": 5319
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1745766953040287e-05,
+      "loss": 0.3829,
+      "step": 5320
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1744098757194096e-05,
+      "loss": 0.3972,
+      "step": 5321
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1742430561347904e-05,
+      "loss": 0.3914,
+      "step": 5322
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.174076236550171e-05,
+      "loss": 0.3927,
+      "step": 5323
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1739094169655518e-05,
+      "loss": 0.4211,
+      "step": 5324
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1737425973809326e-05,
+      "loss": 0.4092,
+      "step": 5325
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1735757777963134e-05,
+      "loss": 0.429,
+      "step": 5326
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.173408958211694e-05,
+      "loss": 0.4103,
+      "step": 5327
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.173242138627075e-05,
+      "loss": 0.4179,
+      "step": 5328
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1730753190424558e-05,
+      "loss": 0.3657,
+      "step": 5329
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1729084994578364e-05,
+      "loss": 0.436,
+      "step": 5330
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1727416798732172e-05,
+      "loss": 0.3681,
+      "step": 5331
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.172574860288598e-05,
+      "loss": 0.4078,
+      "step": 5332
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.172408040703979e-05,
+      "loss": 0.4019,
+      "step": 5333
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1722412211193594e-05,
+      "loss": 0.3883,
+      "step": 5334
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1720744015347403e-05,
+      "loss": 0.3936,
+      "step": 5335
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.1719075819501211e-05,
+      "loss": 0.4228,
+      "step": 5336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1717407623655017e-05,
+      "loss": 0.397,
+      "step": 5337
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1715739427808825e-05,
+      "loss": 0.4199,
+      "step": 5338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1714071231962633e-05,
+      "loss": 0.412,
+      "step": 5339
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1712403036116443e-05,
+      "loss": 0.4291,
+      "step": 5340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1710734840270247e-05,
+      "loss": 0.4116,
+      "step": 5341
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1709066644424057e-05,
+      "loss": 0.3875,
+      "step": 5342
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1707398448577865e-05,
+      "loss": 0.4197,
+      "step": 5343
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1705730252731671e-05,
+      "loss": 0.4003,
+      "step": 5344
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1704062056885479e-05,
+      "loss": 0.4036,
+      "step": 5345
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1702393861039287e-05,
+      "loss": 0.4318,
+      "step": 5346
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1700725665193096e-05,
+      "loss": 0.3902,
+      "step": 5347
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1699057469346902e-05,
+      "loss": 0.4142,
+      "step": 5348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.169738927350071e-05,
+      "loss": 0.4309,
+      "step": 5349
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1695721077654518e-05,
+      "loss": 0.401,
+      "step": 5350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1694052881808324e-05,
+      "loss": 0.414,
+      "step": 5351
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1692384685962132e-05,
+      "loss": 0.4149,
+      "step": 5352
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.169071649011594e-05,
+      "loss": 0.4088,
+      "step": 5353
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.168904829426975e-05,
+      "loss": 0.3925,
+      "step": 5354
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1687380098423556e-05,
+      "loss": 0.3878,
+      "step": 5355
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1685711902577364e-05,
+      "loss": 0.3663,
+      "step": 5356
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1684043706731172e-05,
+      "loss": 0.3843,
+      "step": 5357
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1682375510884978e-05,
+      "loss": 0.3756,
+      "step": 5358
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1680707315038786e-05,
+      "loss": 0.3876,
+      "step": 5359
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1679039119192594e-05,
+      "loss": 0.4025,
+      "step": 5360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1677370923346403e-05,
+      "loss": 0.3862,
+      "step": 5361
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.167570272750021e-05,
+      "loss": 0.3908,
+      "step": 5362
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1674034531654017e-05,
+      "loss": 0.3949,
+      "step": 5363
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1672366335807825e-05,
+      "loss": 0.4186,
+      "step": 5364
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1670698139961631e-05,
+      "loss": 0.4056,
+      "step": 5365
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.166902994411544e-05,
+      "loss": 0.4236,
+      "step": 5366
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1667361748269247e-05,
+      "loss": 0.4116,
+      "step": 5367
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1665693552423057e-05,
+      "loss": 0.4221,
+      "step": 5368
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1664025356576863e-05,
+      "loss": 0.391,
+      "step": 5369
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1662357160730671e-05,
+      "loss": 0.4033,
+      "step": 5370
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1660688964884479e-05,
+      "loss": 0.4277,
+      "step": 5371
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1659020769038285e-05,
+      "loss": 0.4314,
+      "step": 5372
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1657352573192093e-05,
+      "loss": 0.4025,
+      "step": 5373
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1655684377345902e-05,
+      "loss": 0.4067,
+      "step": 5374
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.165401618149971e-05,
+      "loss": 0.4168,
+      "step": 5375
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1652347985653517e-05,
+      "loss": 0.3807,
+      "step": 5376
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.1650679789807324e-05,
+      "loss": 0.3692,
+      "step": 5377
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1649011593961132e-05,
+      "loss": 0.413,
+      "step": 5378
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1647343398114939e-05,
+      "loss": 0.4144,
+      "step": 5379
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1645675202268746e-05,
+      "loss": 0.3844,
+      "step": 5380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1644007006422556e-05,
+      "loss": 0.3715,
+      "step": 5381
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1642338810576364e-05,
+      "loss": 0.4117,
+      "step": 5382
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.164067061473017e-05,
+      "loss": 0.4053,
+      "step": 5383
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1639002418883978e-05,
+      "loss": 0.3963,
+      "step": 5384
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1637334223037786e-05,
+      "loss": 0.4115,
+      "step": 5385
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1635666027191592e-05,
+      "loss": 0.389,
+      "step": 5386
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.16339978313454e-05,
+      "loss": 0.4028,
+      "step": 5387
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.163232963549921e-05,
+      "loss": 0.4092,
+      "step": 5388
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1630661439653017e-05,
+      "loss": 0.4082,
+      "step": 5389
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1628993243806824e-05,
+      "loss": 0.3991,
+      "step": 5390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1627325047960631e-05,
+      "loss": 0.3947,
+      "step": 5391
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.162565685211444e-05,
+      "loss": 0.4067,
+      "step": 5392
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1623988656268246e-05,
+      "loss": 0.383,
+      "step": 5393
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1622320460422053e-05,
+      "loss": 0.3932,
+      "step": 5394
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1620652264575863e-05,
+      "loss": 0.356,
+      "step": 5395
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1618984068729671e-05,
+      "loss": 0.403,
+      "step": 5396
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1617315872883477e-05,
+      "loss": 0.391,
+      "step": 5397
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1615647677037285e-05,
+      "loss": 0.3614,
+      "step": 5398
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1613979481191093e-05,
+      "loss": 0.3987,
+      "step": 5399
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1612311285344899e-05,
+      "loss": 0.3855,
+      "step": 5400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1610643089498707e-05,
+      "loss": 0.4436,
+      "step": 5401
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1608974893652517e-05,
+      "loss": 0.4063,
+      "step": 5402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1607306697806324e-05,
+      "loss": 0.4091,
+      "step": 5403
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.160563850196013e-05,
+      "loss": 0.4132,
+      "step": 5404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1603970306113938e-05,
+      "loss": 0.4005,
+      "step": 5405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1602302110267746e-05,
+      "loss": 0.3974,
+      "step": 5406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1600633914421553e-05,
+      "loss": 0.4335,
+      "step": 5407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.159896571857536e-05,
+      "loss": 0.4215,
+      "step": 5408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.159729752272917e-05,
+      "loss": 0.4293,
+      "step": 5409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1595629326882978e-05,
+      "loss": 0.3907,
+      "step": 5410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1593961131036784e-05,
+      "loss": 0.4123,
+      "step": 5411
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1592292935190592e-05,
+      "loss": 0.3976,
+      "step": 5412
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.15906247393444e-05,
+      "loss": 0.4057,
+      "step": 5413
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1588956543498206e-05,
+      "loss": 0.3969,
+      "step": 5414
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1587288347652016e-05,
+      "loss": 0.381,
+      "step": 5415
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1585620151805824e-05,
+      "loss": 0.4241,
+      "step": 5416
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1583951955959631e-05,
+      "loss": 0.4089,
+      "step": 5417
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1582283760113438e-05,
+      "loss": 0.4037,
+      "step": 5418
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1580615564267246e-05,
+      "loss": 0.3689,
+      "step": 5419
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 0.3746,
+      "step": 5420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1577279172574863e-05,
+      "loss": 0.4174,
+      "step": 5421
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.157561097672867e-05,
+      "loss": 0.4135,
+      "step": 5422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1573942780882477e-05,
+      "loss": 0.3998,
+      "step": 5423
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1572274585036285e-05,
+      "loss": 0.4103,
+      "step": 5424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1570606389190091e-05,
+      "loss": 0.3833,
+      "step": 5425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1568938193343899e-05,
+      "loss": 0.395,
+      "step": 5426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1567269997497707e-05,
+      "loss": 0.3898,
+      "step": 5427
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1565601801651516e-05,
+      "loss": 0.4084,
+      "step": 5428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1563933605805323e-05,
+      "loss": 0.3823,
+      "step": 5429
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.156226540995913e-05,
+      "loss": 0.4171,
+      "step": 5430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1560597214112938e-05,
+      "loss": 0.3944,
+      "step": 5431
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1558929018266745e-05,
+      "loss": 0.443,
+      "step": 5432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1557260822420553e-05,
+      "loss": 0.4103,
+      "step": 5433
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.155559262657436e-05,
+      "loss": 0.3728,
+      "step": 5434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.155392443072817e-05,
+      "loss": 0.4135,
+      "step": 5435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1552256234881976e-05,
+      "loss": 0.4323,
+      "step": 5436
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1550588039035784e-05,
+      "loss": 0.4289,
+      "step": 5437
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1548919843189592e-05,
+      "loss": 0.4041,
+      "step": 5438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1547251647343398e-05,
+      "loss": 0.387,
+      "step": 5439
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1545583451497206e-05,
+      "loss": 0.3905,
+      "step": 5440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1543915255651014e-05,
+      "loss": 0.4122,
+      "step": 5441
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1542247059804824e-05,
+      "loss": 0.3802,
+      "step": 5442
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.154057886395863e-05,
+      "loss": 0.3968,
+      "step": 5443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1538910668112438e-05,
+      "loss": 0.4315,
+      "step": 5444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1537242472266245e-05,
+      "loss": 0.4258,
+      "step": 5445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1535574276420052e-05,
+      "loss": 0.3994,
+      "step": 5446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.153390608057386e-05,
+      "loss": 0.3847,
+      "step": 5447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.153223788472767e-05,
+      "loss": 0.4103,
+      "step": 5448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1530569688881477e-05,
+      "loss": 0.3937,
+      "step": 5449
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1528901493035283e-05,
+      "loss": 0.435,
+      "step": 5450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1527233297189091e-05,
+      "loss": 0.4229,
+      "step": 5451
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1525565101342899e-05,
+      "loss": 0.375,
+      "step": 5452
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1523896905496705e-05,
+      "loss": 0.3966,
+      "step": 5453
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1522228709650513e-05,
+      "loss": 0.384,
+      "step": 5454
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1520560513804323e-05,
+      "loss": 0.4099,
+      "step": 5455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.151889231795813e-05,
+      "loss": 0.4225,
+      "step": 5456
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1517224122111937e-05,
+      "loss": 0.3912,
+      "step": 5457
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1515555926265745e-05,
+      "loss": 0.4014,
+      "step": 5458
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1513887730419553e-05,
+      "loss": 0.3772,
+      "step": 5459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1512219534573359e-05,
+      "loss": 0.3862,
+      "step": 5460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1510551338727167e-05,
+      "loss": 0.422,
+      "step": 5461
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1508883142880976e-05,
+      "loss": 0.4152,
+      "step": 5462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1507214947034784e-05,
+      "loss": 0.4278,
+      "step": 5463
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.150554675118859e-05,
+      "loss": 0.4261,
+      "step": 5464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1503878555342398e-05,
+      "loss": 0.3632,
+      "step": 5465
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1502210359496206e-05,
+      "loss": 0.4183,
+      "step": 5466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1500542163650012e-05,
+      "loss": 0.4149,
+      "step": 5467
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.149887396780382e-05,
+      "loss": 0.3892,
+      "step": 5468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.149720577195763e-05,
+      "loss": 0.4,
+      "step": 5469
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1495537576111438e-05,
+      "loss": 0.4403,
+      "step": 5470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1493869380265244e-05,
+      "loss": 0.395,
+      "step": 5471
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1492201184419052e-05,
+      "loss": 0.4075,
+      "step": 5472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.149053298857286e-05,
+      "loss": 0.4073,
+      "step": 5473
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1488864792726666e-05,
+      "loss": 0.3971,
+      "step": 5474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1487196596880474e-05,
+      "loss": 0.3688,
+      "step": 5475
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1485528401034283e-05,
+      "loss": 0.3846,
+      "step": 5476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1483860205188091e-05,
+      "loss": 0.3906,
+      "step": 5477
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1482192009341897e-05,
+      "loss": 0.4085,
+      "step": 5478
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1480523813495705e-05,
+      "loss": 0.4187,
+      "step": 5479
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1478855617649513e-05,
+      "loss": 0.4066,
+      "step": 5480
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.147718742180332e-05,
+      "loss": 0.4263,
+      "step": 5481
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1475519225957127e-05,
+      "loss": 0.4249,
+      "step": 5482
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1473851030110937e-05,
+      "loss": 0.3991,
+      "step": 5483
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1472182834264745e-05,
+      "loss": 0.3736,
+      "step": 5484
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.147051463841855e-05,
+      "loss": 0.3855,
+      "step": 5485
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1468846442572359e-05,
+      "loss": 0.3996,
+      "step": 5486
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1467178246726167e-05,
+      "loss": 0.3841,
+      "step": 5487
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1465510050879973e-05,
+      "loss": 0.3993,
+      "step": 5488
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1463841855033782e-05,
+      "loss": 0.386,
+      "step": 5489
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.146217365918759e-05,
+      "loss": 0.3763,
+      "step": 5490
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1460505463341398e-05,
+      "loss": 0.3912,
+      "step": 5491
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1458837267495204e-05,
+      "loss": 0.4016,
+      "step": 5492
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1457169071649012e-05,
+      "loss": 0.4243,
+      "step": 5493
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.145550087580282e-05,
+      "loss": 0.3981,
+      "step": 5494
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1453832679956626e-05,
+      "loss": 0.4091,
+      "step": 5495
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1452164484110436e-05,
+      "loss": 0.3817,
+      "step": 5496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1450496288264244e-05,
+      "loss": 0.3744,
+      "step": 5497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1448828092418052e-05,
+      "loss": 0.3732,
+      "step": 5498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1447159896571858e-05,
+      "loss": 0.3676,
+      "step": 5499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1445491700725666e-05,
+      "loss": 0.3915,
+      "step": 5500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1443823504879474e-05,
+      "loss": 0.3841,
+      "step": 5501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.144215530903328e-05,
+      "loss": 0.4211,
+      "step": 5502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.144048711318709e-05,
+      "loss": 0.4077,
+      "step": 5503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1438818917340897e-05,
+      "loss": 0.401,
+      "step": 5504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1437150721494705e-05,
+      "loss": 0.3658,
+      "step": 5505
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1435482525648511e-05,
+      "loss": 0.4278,
+      "step": 5506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.143381432980232e-05,
+      "loss": 0.4276,
+      "step": 5507
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1432146133956127e-05,
+      "loss": 0.42,
+      "step": 5508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1430477938109933e-05,
+      "loss": 0.4085,
+      "step": 5509
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1428809742263743e-05,
+      "loss": 0.4041,
+      "step": 5510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.142714154641755e-05,
+      "loss": 0.4317,
+      "step": 5511
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1425473350571359e-05,
+      "loss": 0.4078,
+      "step": 5512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1423805154725165e-05,
+      "loss": 0.4037,
+      "step": 5513
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1422136958878973e-05,
+      "loss": 0.39,
+      "step": 5514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1420468763032782e-05,
+      "loss": 0.3762,
+      "step": 5515
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1418800567186587e-05,
+      "loss": 0.3871,
+      "step": 5516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1417132371340396e-05,
+      "loss": 0.3969,
+      "step": 5517
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1415464175494204e-05,
+      "loss": 0.4155,
+      "step": 5518
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1413795979648012e-05,
+      "loss": 0.4275,
+      "step": 5519
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1412127783801818e-05,
+      "loss": 0.3944,
+      "step": 5520
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1410459587955626e-05,
+      "loss": 0.3973,
+      "step": 5521
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1408791392109436e-05,
+      "loss": 0.401,
+      "step": 5522
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.140712319626324e-05,
+      "loss": 0.3888,
+      "step": 5523
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.140545500041705e-05,
+      "loss": 0.4022,
+      "step": 5524
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1403786804570858e-05,
+      "loss": 0.4183,
+      "step": 5525
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1402118608724666e-05,
+      "loss": 0.3963,
+      "step": 5526
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1400450412878472e-05,
+      "loss": 0.375,
+      "step": 5527
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.139878221703228e-05,
+      "loss": 0.3913,
+      "step": 5528
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.139711402118609e-05,
+      "loss": 0.4095,
+      "step": 5529
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1395445825339896e-05,
+      "loss": 0.3936,
+      "step": 5530
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1393777629493703e-05,
+      "loss": 0.3916,
+      "step": 5531
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1392109433647511e-05,
+      "loss": 0.4197,
+      "step": 5532
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.139044123780132e-05,
+      "loss": 0.4281,
+      "step": 5533
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1388773041955125e-05,
+      "loss": 0.424,
+      "step": 5534
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1387104846108933e-05,
+      "loss": 0.424,
+      "step": 5535
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1385436650262743e-05,
+      "loss": 0.3908,
+      "step": 5536
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.138376845441655e-05,
+      "loss": 0.3971,
+      "step": 5537
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1382100258570357e-05,
+      "loss": 0.4452,
+      "step": 5538
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1380432062724165e-05,
+      "loss": 0.4174,
+      "step": 5539
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1378763866877973e-05,
+      "loss": 0.4252,
+      "step": 5540
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1377095671031779e-05,
+      "loss": 0.3971,
+      "step": 5541
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1375427475185587e-05,
+      "loss": 0.4252,
+      "step": 5542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1373759279339396e-05,
+      "loss": 0.4039,
+      "step": 5543
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1372091083493204e-05,
+      "loss": 0.3941,
+      "step": 5544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.137042288764701e-05,
+      "loss": 0.3943,
+      "step": 5545
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1368754691800818e-05,
+      "loss": 0.414,
+      "step": 5546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1367086495954626e-05,
+      "loss": 0.3856,
+      "step": 5547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1365418300108432e-05,
+      "loss": 0.408,
+      "step": 5548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.136375010426224e-05,
+      "loss": 0.3954,
+      "step": 5549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.136208190841605e-05,
+      "loss": 0.4061,
+      "step": 5550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1360413712569858e-05,
+      "loss": 0.4032,
+      "step": 5551
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1358745516723664e-05,
+      "loss": 0.4561,
+      "step": 5552
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1357077320877472e-05,
+      "loss": 0.3957,
+      "step": 5553
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.135540912503128e-05,
+      "loss": 0.4161,
+      "step": 5554
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1353740929185086e-05,
+      "loss": 0.384,
+      "step": 5555
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1352072733338896e-05,
+      "loss": 0.3867,
+      "step": 5556
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1350404537492703e-05,
+      "loss": 0.3885,
+      "step": 5557
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1348736341646511e-05,
+      "loss": 0.3951,
+      "step": 5558
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1347068145800318e-05,
+      "loss": 0.3856,
+      "step": 5559
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1345399949954125e-05,
+      "loss": 0.4057,
+      "step": 5560
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1343731754107933e-05,
+      "loss": 0.3896,
+      "step": 5561
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.134206355826174e-05,
+      "loss": 0.3727,
+      "step": 5562
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1340395362415549e-05,
+      "loss": 0.3813,
+      "step": 5563
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1338727166569357e-05,
+      "loss": 0.3945,
+      "step": 5564
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1337058970723165e-05,
+      "loss": 0.4245,
+      "step": 5565
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1335390774876971e-05,
+      "loss": 0.4305,
+      "step": 5566
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1333722579030779e-05,
+      "loss": 0.4021,
+      "step": 5567
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1332054383184587e-05,
+      "loss": 0.4156,
+      "step": 5568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1330386187338393e-05,
+      "loss": 0.4056,
+      "step": 5569
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1328717991492203e-05,
+      "loss": 0.3755,
+      "step": 5570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.132704979564601e-05,
+      "loss": 0.4183,
+      "step": 5571
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1325381599799818e-05,
+      "loss": 0.4127,
+      "step": 5572
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1323713403953625e-05,
+      "loss": 0.4256,
+      "step": 5573
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1322045208107432e-05,
+      "loss": 0.3748,
+      "step": 5574
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.132037701226124e-05,
+      "loss": 0.3895,
+      "step": 5575
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1318708816415047e-05,
+      "loss": 0.4221,
+      "step": 5576
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1317040620568856e-05,
+      "loss": 0.3865,
+      "step": 5577
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1315372424722664e-05,
+      "loss": 0.3554,
+      "step": 5578
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1313704228876472e-05,
+      "loss": 0.4135,
+      "step": 5579
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1312036033030278e-05,
+      "loss": 0.4133,
+      "step": 5580
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1310367837184086e-05,
+      "loss": 0.4022,
+      "step": 5581
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1308699641337894e-05,
+      "loss": 0.3662,
+      "step": 5582
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.13070314454917e-05,
+      "loss": 0.4229,
+      "step": 5583
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.130536324964551e-05,
+      "loss": 0.4022,
+      "step": 5584
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1303695053799317e-05,
+      "loss": 0.3978,
+      "step": 5585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1302026857953125e-05,
+      "loss": 0.386,
+      "step": 5586
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1300358662106932e-05,
+      "loss": 0.387,
+      "step": 5587
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.129869046626074e-05,
+      "loss": 0.3902,
+      "step": 5588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1297022270414549e-05,
+      "loss": 0.418,
+      "step": 5589
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1295354074568354e-05,
+      "loss": 0.4185,
+      "step": 5590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1293685878722163e-05,
+      "loss": 0.4191,
+      "step": 5591
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1292017682875971e-05,
+      "loss": 0.3747,
+      "step": 5592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1290349487029779e-05,
+      "loss": 0.423,
+      "step": 5593
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1288681291183585e-05,
+      "loss": 0.3813,
+      "step": 5594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1287013095337393e-05,
+      "loss": 0.4079,
+      "step": 5595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1285344899491203e-05,
+      "loss": 0.4054,
+      "step": 5596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1283676703645007e-05,
+      "loss": 0.4052,
+      "step": 5597
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1282008507798817e-05,
+      "loss": 0.4,
+      "step": 5598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1280340311952625e-05,
+      "loss": 0.4008,
+      "step": 5599
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1278672116106432e-05,
+      "loss": 0.3747,
+      "step": 5600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1277003920260239e-05,
+      "loss": 0.4016,
+      "step": 5601
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1275335724414046e-05,
+      "loss": 0.407,
+      "step": 5602
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1273667528567856e-05,
+      "loss": 0.3865,
+      "step": 5603
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1271999332721662e-05,
+      "loss": 0.382,
+      "step": 5604
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.127033113687547e-05,
+      "loss": 0.4277,
+      "step": 5605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1268662941029278e-05,
+      "loss": 0.3915,
+      "step": 5606
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1266994745183086e-05,
+      "loss": 0.3896,
+      "step": 5607
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1265326549336892e-05,
+      "loss": 0.4046,
+      "step": 5608
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.12636583534907e-05,
+      "loss": 0.3986,
+      "step": 5609
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.126199015764451e-05,
+      "loss": 0.3707,
+      "step": 5610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1260321961798316e-05,
+      "loss": 0.3969,
+      "step": 5611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1258653765952124e-05,
+      "loss": 0.4024,
+      "step": 5612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1256985570105932e-05,
+      "loss": 0.4026,
+      "step": 5613
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.125531737425974e-05,
+      "loss": 0.3982,
+      "step": 5614
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1253649178413546e-05,
+      "loss": 0.3962,
+      "step": 5615
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1251980982567354e-05,
+      "loss": 0.3786,
+      "step": 5616
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1250312786721163e-05,
+      "loss": 0.43,
+      "step": 5617
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.124864459087497e-05,
+      "loss": 0.3859,
+      "step": 5618
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1246976395028777e-05,
+      "loss": 0.4219,
+      "step": 5619
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1245308199182585e-05,
+      "loss": 0.3852,
+      "step": 5620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1243640003336393e-05,
+      "loss": 0.3903,
+      "step": 5621
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.12419718074902e-05,
+      "loss": 0.3874,
+      "step": 5622
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1240303611644007e-05,
+      "loss": 0.378,
+      "step": 5623
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1238635415797817e-05,
+      "loss": 0.4197,
+      "step": 5624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1236967219951623e-05,
+      "loss": 0.3876,
+      "step": 5625
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.123529902410543e-05,
+      "loss": 0.3837,
+      "step": 5626
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1233630828259239e-05,
+      "loss": 0.3946,
+      "step": 5627
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1231962632413046e-05,
+      "loss": 0.3997,
+      "step": 5628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1230294436566853e-05,
+      "loss": 0.4181,
+      "step": 5629
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1228626240720662e-05,
+      "loss": 0.4237,
+      "step": 5630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.122695804487447e-05,
+      "loss": 0.3835,
+      "step": 5631
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1225289849028276e-05,
+      "loss": 0.4203,
+      "step": 5632
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1223621653182084e-05,
+      "loss": 0.4308,
+      "step": 5633
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1221953457335892e-05,
+      "loss": 0.3669,
+      "step": 5634
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.12202852614897e-05,
+      "loss": 0.3796,
+      "step": 5635
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1218617065643506e-05,
+      "loss": 0.3949,
+      "step": 5636
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1216948869797316e-05,
+      "loss": 0.3801,
+      "step": 5637
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1215280673951124e-05,
+      "loss": 0.4311,
+      "step": 5638
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.121361247810493e-05,
+      "loss": 0.3936,
+      "step": 5639
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1211944282258738e-05,
+      "loss": 0.3718,
+      "step": 5640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1210276086412546e-05,
+      "loss": 0.4019,
+      "step": 5641
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1208607890566354e-05,
+      "loss": 0.3944,
+      "step": 5642
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.120693969472016e-05,
+      "loss": 0.4041,
+      "step": 5643
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.120527149887397e-05,
+      "loss": 0.4343,
+      "step": 5644
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1203603303027777e-05,
+      "loss": 0.3889,
+      "step": 5645
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1201935107181583e-05,
+      "loss": 0.3732,
+      "step": 5646
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1200266911335391e-05,
+      "loss": 0.3875,
+      "step": 5647
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1198598715489199e-05,
+      "loss": 0.4042,
+      "step": 5648
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1196930519643007e-05,
+      "loss": 0.3978,
+      "step": 5649
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1195262323796813e-05,
+      "loss": 0.3906,
+      "step": 5650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1193594127950623e-05,
+      "loss": 0.3864,
+      "step": 5651
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.119192593210443e-05,
+      "loss": 0.4009,
+      "step": 5652
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1190257736258237e-05,
+      "loss": 0.3725,
+      "step": 5653
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1188589540412045e-05,
+      "loss": 0.4175,
+      "step": 5654
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1186921344565853e-05,
+      "loss": 0.4362,
+      "step": 5655
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1185253148719662e-05,
+      "loss": 0.3723,
+      "step": 5656
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1183584952873467e-05,
+      "loss": 0.3988,
+      "step": 5657
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1181916757027276e-05,
+      "loss": 0.3962,
+      "step": 5658
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1180248561181084e-05,
+      "loss": 0.4037,
+      "step": 5659
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1178580365334892e-05,
+      "loss": 0.395,
+      "step": 5660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1176912169488698e-05,
+      "loss": 0.4173,
+      "step": 5661
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1175243973642506e-05,
+      "loss": 0.4104,
+      "step": 5662
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1173575777796316e-05,
+      "loss": 0.3848,
+      "step": 5663
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.117190758195012e-05,
+      "loss": 0.4125,
+      "step": 5664
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.117023938610393e-05,
+      "loss": 0.4082,
+      "step": 5665
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1168571190257738e-05,
+      "loss": 0.4132,
+      "step": 5666
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1166902994411546e-05,
+      "loss": 0.3989,
+      "step": 5667
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1165234798565352e-05,
+      "loss": 0.4028,
+      "step": 5668
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.116356660271916e-05,
+      "loss": 0.4218,
+      "step": 5669
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.116189840687297e-05,
+      "loss": 0.3652,
+      "step": 5670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1160230211026775e-05,
+      "loss": 0.4078,
+      "step": 5671
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1158562015180583e-05,
+      "loss": 0.3885,
+      "step": 5672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1156893819334391e-05,
+      "loss": 0.3664,
+      "step": 5673
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1155225623488199e-05,
+      "loss": 0.3927,
+      "step": 5674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1153557427642005e-05,
+      "loss": 0.3855,
+      "step": 5675
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1151889231795813e-05,
+      "loss": 0.4203,
+      "step": 5676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1150221035949623e-05,
+      "loss": 0.3849,
+      "step": 5677
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1148552840103429e-05,
+      "loss": 0.3849,
+      "step": 5678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1146884644257237e-05,
+      "loss": 0.3973,
+      "step": 5679
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1145216448411045e-05,
+      "loss": 0.4346,
+      "step": 5680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1143548252564853e-05,
+      "loss": 0.4428,
+      "step": 5681
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1141880056718659e-05,
+      "loss": 0.3709,
+      "step": 5682
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1140211860872467e-05,
+      "loss": 0.4125,
+      "step": 5683
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1138543665026276e-05,
+      "loss": 0.4135,
+      "step": 5684
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1136875469180082e-05,
+      "loss": 0.4082,
+      "step": 5685
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.113520727333389e-05,
+      "loss": 0.3902,
+      "step": 5686
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1133539077487698e-05,
+      "loss": 0.3747,
+      "step": 5687
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1131870881641506e-05,
+      "loss": 0.3981,
+      "step": 5688
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1130202685795312e-05,
+      "loss": 0.425,
+      "step": 5689
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.112853448994912e-05,
+      "loss": 0.3844,
+      "step": 5690
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.112686629410293e-05,
+      "loss": 0.429,
+      "step": 5691
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1125198098256736e-05,
+      "loss": 0.3943,
+      "step": 5692
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1123529902410544e-05,
+      "loss": 0.41,
+      "step": 5693
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1121861706564352e-05,
+      "loss": 0.4191,
+      "step": 5694
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.112019351071816e-05,
+      "loss": 0.4087,
+      "step": 5695
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1118525314871966e-05,
+      "loss": 0.4078,
+      "step": 5696
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1116857119025774e-05,
+      "loss": 0.4013,
+      "step": 5697
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1115188923179583e-05,
+      "loss": 0.422,
+      "step": 5698
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.111352072733339e-05,
+      "loss": 0.3812,
+      "step": 5699
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1111852531487197e-05,
+      "loss": 0.4144,
+      "step": 5700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1110184335641005e-05,
+      "loss": 0.3719,
+      "step": 5701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1108516139794813e-05,
+      "loss": 0.4114,
+      "step": 5702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.110684794394862e-05,
+      "loss": 0.4086,
+      "step": 5703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1105179748102429e-05,
+      "loss": 0.4099,
+      "step": 5704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1103511552256237e-05,
+      "loss": 0.4013,
+      "step": 5705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1101843356410043e-05,
+      "loss": 0.4008,
+      "step": 5706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1100175160563851e-05,
+      "loss": 0.4024,
+      "step": 5707
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1098506964717659e-05,
+      "loss": 0.415,
+      "step": 5708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1096838768871467e-05,
+      "loss": 0.4163,
+      "step": 5709
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1095170573025273e-05,
+      "loss": 0.4053,
+      "step": 5710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1093502377179082e-05,
+      "loss": 0.4283,
+      "step": 5711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.109183418133289e-05,
+      "loss": 0.3871,
+      "step": 5712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1090165985486697e-05,
+      "loss": 0.4614,
+      "step": 5713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1088497789640504e-05,
+      "loss": 0.4048,
+      "step": 5714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1086829593794312e-05,
+      "loss": 0.397,
+      "step": 5715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.108516139794812e-05,
+      "loss": 0.4082,
+      "step": 5716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1083493202101926e-05,
+      "loss": 0.3917,
+      "step": 5717
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1081825006255736e-05,
+      "loss": 0.3919,
+      "step": 5718
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1080156810409544e-05,
+      "loss": 0.4053,
+      "step": 5719
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.107848861456335e-05,
+      "loss": 0.3931,
+      "step": 5720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1076820418717158e-05,
+      "loss": 0.4465,
+      "step": 5721
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1075152222870966e-05,
+      "loss": 0.4096,
+      "step": 5722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1073484027024774e-05,
+      "loss": 0.4333,
+      "step": 5723
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.107181583117858e-05,
+      "loss": 0.3693,
+      "step": 5724
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.107014763533239e-05,
+      "loss": 0.4239,
+      "step": 5725
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1068479439486197e-05,
+      "loss": 0.4012,
+      "step": 5726
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1066811243640004e-05,
+      "loss": 0.3734,
+      "step": 5727
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1065143047793811e-05,
+      "loss": 0.386,
+      "step": 5728
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.106347485194762e-05,
+      "loss": 0.4052,
+      "step": 5729
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1061806656101429e-05,
+      "loss": 0.42,
+      "step": 5730
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1060138460255233e-05,
+      "loss": 0.3928,
+      "step": 5731
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1058470264409043e-05,
+      "loss": 0.3423,
+      "step": 5732
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1056802068562851e-05,
+      "loss": 0.3951,
+      "step": 5733
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1055133872716657e-05,
+      "loss": 0.408,
+      "step": 5734
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1053465676870465e-05,
+      "loss": 0.403,
+      "step": 5735
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1051797481024273e-05,
+      "loss": 0.4378,
+      "step": 5736
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1050129285178082e-05,
+      "loss": 0.4254,
+      "step": 5737
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1048461089331887e-05,
+      "loss": 0.4165,
+      "step": 5738
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1046792893485697e-05,
+      "loss": 0.4061,
+      "step": 5739
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1045124697639504e-05,
+      "loss": 0.3891,
+      "step": 5740
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.104345650179331e-05,
+      "loss": 0.4058,
+      "step": 5741
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1041788305947118e-05,
+      "loss": 0.3952,
+      "step": 5742
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1040120110100926e-05,
+      "loss": 0.3903,
+      "step": 5743
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1038451914254736e-05,
+      "loss": 0.4211,
+      "step": 5744
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1036783718408542e-05,
+      "loss": 0.4161,
+      "step": 5745
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.103511552256235e-05,
+      "loss": 0.3533,
+      "step": 5746
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1033447326716158e-05,
+      "loss": 0.418,
+      "step": 5747
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1031779130869964e-05,
+      "loss": 0.3976,
+      "step": 5748
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1030110935023772e-05,
+      "loss": 0.3964,
+      "step": 5749
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.102844273917758e-05,
+      "loss": 0.4149,
+      "step": 5750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.102677454333139e-05,
+      "loss": 0.3837,
+      "step": 5751
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1025106347485196e-05,
+      "loss": 0.3936,
+      "step": 5752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1023438151639004e-05,
+      "loss": 0.4111,
+      "step": 5753
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1021769955792811e-05,
+      "loss": 0.4074,
+      "step": 5754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1020101759946618e-05,
+      "loss": 0.3463,
+      "step": 5755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1018433564100426e-05,
+      "loss": 0.3675,
+      "step": 5756
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1016765368254233e-05,
+      "loss": 0.4152,
+      "step": 5757
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1015097172408043e-05,
+      "loss": 0.3739,
+      "step": 5758
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.101342897656185e-05,
+      "loss": 0.39,
+      "step": 5759
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1011760780715657e-05,
+      "loss": 0.3641,
+      "step": 5760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1010092584869465e-05,
+      "loss": 0.3968,
+      "step": 5761
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1008424389023271e-05,
+      "loss": 0.4237,
+      "step": 5762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1006756193177079e-05,
+      "loss": 0.4249,
+      "step": 5763
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1005087997330887e-05,
+      "loss": 0.3952,
+      "step": 5764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1003419801484697e-05,
+      "loss": 0.4076,
+      "step": 5765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1001751605638503e-05,
+      "loss": 0.4354,
+      "step": 5766
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.100008340979231e-05,
+      "loss": 0.3886,
+      "step": 5767
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0998415213946118e-05,
+      "loss": 0.4139,
+      "step": 5768
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0996747018099925e-05,
+      "loss": 0.3939,
+      "step": 5769
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0995078822253733e-05,
+      "loss": 0.3943,
+      "step": 5770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0993410626407542e-05,
+      "loss": 0.4082,
+      "step": 5771
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.099174243056135e-05,
+      "loss": 0.3995,
+      "step": 5772
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0990074234715156e-05,
+      "loss": 0.4236,
+      "step": 5773
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0988406038868964e-05,
+      "loss": 0.3933,
+      "step": 5774
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0986737843022772e-05,
+      "loss": 0.4021,
+      "step": 5775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.098506964717658e-05,
+      "loss": 0.3898,
+      "step": 5776
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0983401451330386e-05,
+      "loss": 0.3929,
+      "step": 5777
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0981733255484196e-05,
+      "loss": 0.3783,
+      "step": 5778
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0980065059638004e-05,
+      "loss": 0.4146,
+      "step": 5779
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.097839686379181e-05,
+      "loss": 0.3784,
+      "step": 5780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0976728667945618e-05,
+      "loss": 0.4209,
+      "step": 5781
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0975060472099426e-05,
+      "loss": 0.3909,
+      "step": 5782
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0973392276253233e-05,
+      "loss": 0.4223,
+      "step": 5783
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.097172408040704e-05,
+      "loss": 0.399,
+      "step": 5784
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.097005588456085e-05,
+      "loss": 0.3996,
+      "step": 5785
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0968387688714657e-05,
+      "loss": 0.4008,
+      "step": 5786
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0966719492868463e-05,
+      "loss": 0.4084,
+      "step": 5787
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0965051297022271e-05,
+      "loss": 0.3949,
+      "step": 5788
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0963383101176079e-05,
+      "loss": 0.3882,
+      "step": 5789
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0961714905329887e-05,
+      "loss": 0.4086,
+      "step": 5790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0960046709483693e-05,
+      "loss": 0.3741,
+      "step": 5791
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0958378513637503e-05,
+      "loss": 0.3873,
+      "step": 5792
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.095671031779131e-05,
+      "loss": 0.393,
+      "step": 5793
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0955042121945117e-05,
+      "loss": 0.4259,
+      "step": 5794
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0953373926098925e-05,
+      "loss": 0.3809,
+      "step": 5795
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0951705730252733e-05,
+      "loss": 0.4158,
+      "step": 5796
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.095003753440654e-05,
+      "loss": 0.384,
+      "step": 5797
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0948369338560347e-05,
+      "loss": 0.4026,
+      "step": 5798
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0946701142714156e-05,
+      "loss": 0.4061,
+      "step": 5799
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0945032946867964e-05,
+      "loss": 0.4076,
+      "step": 5800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.094336475102177e-05,
+      "loss": 0.4248,
+      "step": 5801
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0941696555175578e-05,
+      "loss": 0.3877,
+      "step": 5802
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0940028359329386e-05,
+      "loss": 0.3715,
+      "step": 5803
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0938360163483196e-05,
+      "loss": 0.3766,
+      "step": 5804
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0936691967637e-05,
+      "loss": 0.3928,
+      "step": 5805
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.093502377179081e-05,
+      "loss": 0.3815,
+      "step": 5806
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0933355575944618e-05,
+      "loss": 0.3915,
+      "step": 5807
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0931687380098424e-05,
+      "loss": 0.4274,
+      "step": 5808
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0930019184252232e-05,
+      "loss": 0.3888,
+      "step": 5809
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.092835098840604e-05,
+      "loss": 0.3756,
+      "step": 5810
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0926682792559849e-05,
+      "loss": 0.397,
+      "step": 5811
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0925014596713655e-05,
+      "loss": 0.3796,
+      "step": 5812
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0923346400867463e-05,
+      "loss": 0.3933,
+      "step": 5813
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0921678205021271e-05,
+      "loss": 0.3704,
+      "step": 5814
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0920010009175077e-05,
+      "loss": 0.4263,
+      "step": 5815
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0918341813328885e-05,
+      "loss": 0.3999,
+      "step": 5816
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0916673617482693e-05,
+      "loss": 0.4111,
+      "step": 5817
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0915005421636503e-05,
+      "loss": 0.3854,
+      "step": 5818
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0913337225790309e-05,
+      "loss": 0.3876,
+      "step": 5819
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0911669029944117e-05,
+      "loss": 0.3944,
+      "step": 5820
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0910000834097925e-05,
+      "loss": 0.3825,
+      "step": 5821
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.090833263825173e-05,
+      "loss": 0.3683,
+      "step": 5822
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0906664442405539e-05,
+      "loss": 0.4107,
+      "step": 5823
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0904996246559347e-05,
+      "loss": 0.3975,
+      "step": 5824
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0903328050713156e-05,
+      "loss": 0.4106,
+      "step": 5825
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0901659854866962e-05,
+      "loss": 0.4128,
+      "step": 5826
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.089999165902077e-05,
+      "loss": 0.3924,
+      "step": 5827
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0898323463174578e-05,
+      "loss": 0.4148,
+      "step": 5828
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0896655267328384e-05,
+      "loss": 0.4091,
+      "step": 5829
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0894987071482192e-05,
+      "loss": 0.3812,
+      "step": 5830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0893318875636e-05,
+      "loss": 0.411,
+      "step": 5831
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.089165067978981e-05,
+      "loss": 0.4076,
+      "step": 5832
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0889982483943616e-05,
+      "loss": 0.4009,
+      "step": 5833
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0888314288097424e-05,
+      "loss": 0.4022,
+      "step": 5834
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0886646092251232e-05,
+      "loss": 0.3686,
+      "step": 5835
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0884977896405038e-05,
+      "loss": 0.4316,
+      "step": 5836
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0883309700558846e-05,
+      "loss": 0.4249,
+      "step": 5837
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0881641504712654e-05,
+      "loss": 0.4027,
+      "step": 5838
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0879973308866463e-05,
+      "loss": 0.4058,
+      "step": 5839
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.087830511302027e-05,
+      "loss": 0.3778,
+      "step": 5840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0876636917174077e-05,
+      "loss": 0.4111,
+      "step": 5841
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0874968721327885e-05,
+      "loss": 0.3828,
+      "step": 5842
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0873300525481691e-05,
+      "loss": 0.4153,
+      "step": 5843
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.08716323296355e-05,
+      "loss": 0.3991,
+      "step": 5844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0869964133789309e-05,
+      "loss": 0.4187,
+      "step": 5845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0868295937943117e-05,
+      "loss": 0.3978,
+      "step": 5846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0866627742096923e-05,
+      "loss": 0.4154,
+      "step": 5847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.086495954625073e-05,
+      "loss": 0.3819,
+      "step": 5848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0863291350404539e-05,
+      "loss": 0.4256,
+      "step": 5849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0861623154558345e-05,
+      "loss": 0.4141,
+      "step": 5850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0859954958712153e-05,
+      "loss": 0.3903,
+      "step": 5851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0858286762865962e-05,
+      "loss": 0.3938,
+      "step": 5852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.085661856701977e-05,
+      "loss": 0.4093,
+      "step": 5853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0854950371173576e-05,
+      "loss": 0.4234,
+      "step": 5854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0853282175327384e-05,
+      "loss": 0.3637,
+      "step": 5855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0851613979481192e-05,
+      "loss": 0.3648,
+      "step": 5856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0849945783634998e-05,
+      "loss": 0.3649,
+      "step": 5857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0848277587788806e-05,
+      "loss": 0.4169,
+      "step": 5858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0846609391942616e-05,
+      "loss": 0.3531,
+      "step": 5859
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0844941196096424e-05,
+      "loss": 0.399,
+      "step": 5860
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.084327300025023e-05,
+      "loss": 0.3909,
+      "step": 5861
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0841604804404038e-05,
+      "loss": 0.3827,
+      "step": 5862
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0839936608557846e-05,
+      "loss": 0.3822,
+      "step": 5863
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0838268412711652e-05,
+      "loss": 0.4007,
+      "step": 5864
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.083660021686546e-05,
+      "loss": 0.403,
+      "step": 5865
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.083493202101927e-05,
+      "loss": 0.4181,
+      "step": 5866
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0833263825173077e-05,
+      "loss": 0.4238,
+      "step": 5867
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0831595629326883e-05,
+      "loss": 0.4216,
+      "step": 5868
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0829927433480691e-05,
+      "loss": 0.4225,
+      "step": 5869
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.08282592376345e-05,
+      "loss": 0.3901,
+      "step": 5870
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0826591041788305e-05,
+      "loss": 0.4226,
+      "step": 5871
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0824922845942113e-05,
+      "loss": 0.415,
+      "step": 5872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0823254650095923e-05,
+      "loss": 0.3989,
+      "step": 5873
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.082158645424973e-05,
+      "loss": 0.4321,
+      "step": 5874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0819918258403537e-05,
+      "loss": 0.4007,
+      "step": 5875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0818250062557345e-05,
+      "loss": 0.3736,
+      "step": 5876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0816581866711153e-05,
+      "loss": 0.4301,
+      "step": 5877
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0814913670864959e-05,
+      "loss": 0.4143,
+      "step": 5878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0813245475018767e-05,
+      "loss": 0.3929,
+      "step": 5879
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0811577279172576e-05,
+      "loss": 0.3874,
+      "step": 5880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0809909083326384e-05,
+      "loss": 0.3833,
+      "step": 5881
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.080824088748019e-05,
+      "loss": 0.3958,
+      "step": 5882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0806572691633998e-05,
+      "loss": 0.3817,
+      "step": 5883
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0804904495787806e-05,
+      "loss": 0.4026,
+      "step": 5884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0803236299941612e-05,
+      "loss": 0.3834,
+      "step": 5885
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0801568104095422e-05,
+      "loss": 0.4071,
+      "step": 5886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079989990824923e-05,
+      "loss": 0.405,
+      "step": 5887
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0798231712403038e-05,
+      "loss": 0.3838,
+      "step": 5888
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0796563516556844e-05,
+      "loss": 0.403,
+      "step": 5889
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0794895320710652e-05,
+      "loss": 0.408,
+      "step": 5890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079322712486446e-05,
+      "loss": 0.3938,
+      "step": 5891
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.079155892901827e-05,
+      "loss": 0.4064,
+      "step": 5892
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0789890733172076e-05,
+      "loss": 0.4118,
+      "step": 5893
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0788222537325883e-05,
+      "loss": 0.3918,
+      "step": 5894
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0786554341479691e-05,
+      "loss": 0.3667,
+      "step": 5895
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0784886145633498e-05,
+      "loss": 0.4177,
+      "step": 5896
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0783217949787305e-05,
+      "loss": 0.4217,
+      "step": 5897
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0781549753941113e-05,
+      "loss": 0.374,
+      "step": 5898
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0779881558094923e-05,
+      "loss": 0.3966,
+      "step": 5899
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0778213362248729e-05,
+      "loss": 0.4033,
+      "step": 5900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0776545166402537e-05,
+      "loss": 0.3891,
+      "step": 5901
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0774876970556345e-05,
+      "loss": 0.3719,
+      "step": 5902
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0773208774710151e-05,
+      "loss": 0.3757,
+      "step": 5903
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0771540578863959e-05,
+      "loss": 0.4008,
+      "step": 5904
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0769872383017767e-05,
+      "loss": 0.3533,
+      "step": 5905
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0768204187171576e-05,
+      "loss": 0.3957,
+      "step": 5906
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0766535991325383e-05,
+      "loss": 0.4013,
+      "step": 5907
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.076486779547919e-05,
+      "loss": 0.3968,
+      "step": 5908
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0763199599632998e-05,
+      "loss": 0.393,
+      "step": 5909
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0761531403786805e-05,
+      "loss": 0.3864,
+      "step": 5910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0759863207940612e-05,
+      "loss": 0.3863,
+      "step": 5911
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0758195012094422e-05,
+      "loss": 0.3975,
+      "step": 5912
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.075652681624823e-05,
+      "loss": 0.375,
+      "step": 5913
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0754858620402036e-05,
+      "loss": 0.4229,
+      "step": 5914
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0753190424555844e-05,
+      "loss": 0.3808,
+      "step": 5915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0751522228709652e-05,
+      "loss": 0.4085,
+      "step": 5916
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0749854032863458e-05,
+      "loss": 0.3998,
+      "step": 5917
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0748185837017266e-05,
+      "loss": 0.4224,
+      "step": 5918
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0746517641171076e-05,
+      "loss": 0.3606,
+      "step": 5919
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0744849445324883e-05,
+      "loss": 0.3687,
+      "step": 5920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.074318124947869e-05,
+      "loss": 0.4111,
+      "step": 5921
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0741513053632498e-05,
+      "loss": 0.4078,
+      "step": 5922
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0739844857786305e-05,
+      "loss": 0.3507,
+      "step": 5923
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0738176661940112e-05,
+      "loss": 0.4213,
+      "step": 5924
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.073650846609392e-05,
+      "loss": 0.377,
+      "step": 5925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0734840270247729e-05,
+      "loss": 0.3534,
+      "step": 5926
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0733172074401537e-05,
+      "loss": 0.4044,
+      "step": 5927
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0731503878555343e-05,
+      "loss": 0.443,
+      "step": 5928
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0729835682709151e-05,
+      "loss": 0.4072,
+      "step": 5929
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0728167486862959e-05,
+      "loss": 0.3771,
+      "step": 5930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0726499291016765e-05,
+      "loss": 0.3637,
+      "step": 5931
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0724831095170573e-05,
+      "loss": 0.4088,
+      "step": 5932
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0723162899324383e-05,
+      "loss": 0.4034,
+      "step": 5933
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.072149470347819e-05,
+      "loss": 0.4282,
+      "step": 5934
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0719826507631997e-05,
+      "loss": 0.3787,
+      "step": 5935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0718158311785805e-05,
+      "loss": 0.3961,
+      "step": 5936
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0716490115939612e-05,
+      "loss": 0.3883,
+      "step": 5937
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0714821920093419e-05,
+      "loss": 0.4054,
+      "step": 5938
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0713153724247227e-05,
+      "loss": 0.3856,
+      "step": 5939
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0711485528401036e-05,
+      "loss": 0.379,
+      "step": 5940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0709817332554844e-05,
+      "loss": 0.3992,
+      "step": 5941
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.070814913670865e-05,
+      "loss": 0.3891,
+      "step": 5942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0706480940862458e-05,
+      "loss": 0.3838,
+      "step": 5943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0704812745016266e-05,
+      "loss": 0.3994,
+      "step": 5944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0703144549170072e-05,
+      "loss": 0.3781,
+      "step": 5945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.070147635332388e-05,
+      "loss": 0.3959,
+      "step": 5946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.069980815747769e-05,
+      "loss": 0.4005,
+      "step": 5947
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0698139961631497e-05,
+      "loss": 0.396,
+      "step": 5948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0696471765785304e-05,
+      "loss": 0.4309,
+      "step": 5949
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0694803569939112e-05,
+      "loss": 0.4001,
+      "step": 5950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.069313537409292e-05,
+      "loss": 0.4191,
+      "step": 5951
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0691467178246726e-05,
+      "loss": 0.3933,
+      "step": 5952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0689798982400534e-05,
+      "loss": 0.3948,
+      "step": 5953
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0688130786554343e-05,
+      "loss": 0.4072,
+      "step": 5954
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0686462590708151e-05,
+      "loss": 0.3867,
+      "step": 5955
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0684794394861957e-05,
+      "loss": 0.424,
+      "step": 5956
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0683126199015765e-05,
+      "loss": 0.3895,
+      "step": 5957
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0681458003169573e-05,
+      "loss": 0.4251,
+      "step": 5958
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.067978980732338e-05,
+      "loss": 0.3817,
+      "step": 5959
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0678121611477189e-05,
+      "loss": 0.4012,
+      "step": 5960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0676453415630997e-05,
+      "loss": 0.3954,
+      "step": 5961
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0674785219784805e-05,
+      "loss": 0.436,
+      "step": 5962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.067311702393861e-05,
+      "loss": 0.3949,
+      "step": 5963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0671448828092419e-05,
+      "loss": 0.4028,
+      "step": 5964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0669780632246226e-05,
+      "loss": 0.3934,
+      "step": 5965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0668112436400033e-05,
+      "loss": 0.3844,
+      "step": 5966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0666444240553842e-05,
+      "loss": 0.4211,
+      "step": 5967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.066477604470765e-05,
+      "loss": 0.399,
+      "step": 5968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0663107848861458e-05,
+      "loss": 0.4083,
+      "step": 5969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0661439653015264e-05,
+      "loss": 0.4216,
+      "step": 5970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0659771457169072e-05,
+      "loss": 0.3729,
+      "step": 5971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.065810326132288e-05,
+      "loss": 0.4565,
+      "step": 5972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0656435065476686e-05,
+      "loss": 0.3794,
+      "step": 5973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0654766869630496e-05,
+      "loss": 0.3925,
+      "step": 5974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0653098673784304e-05,
+      "loss": 0.4186,
+      "step": 5975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0651430477938112e-05,
+      "loss": 0.4089,
+      "step": 5976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0649762282091918e-05,
+      "loss": 0.4226,
+      "step": 5977
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0648094086245726e-05,
+      "loss": 0.414,
+      "step": 5978
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0646425890399534e-05,
+      "loss": 0.3991,
+      "step": 5979
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.064475769455334e-05,
+      "loss": 0.3839,
+      "step": 5980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.064308949870715e-05,
+      "loss": 0.4337,
+      "step": 5981
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0641421302860957e-05,
+      "loss": 0.4148,
+      "step": 5982
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0639753107014765e-05,
+      "loss": 0.4131,
+      "step": 5983
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0638084911168571e-05,
+      "loss": 0.4132,
+      "step": 5984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0636416715322379e-05,
+      "loss": 0.418,
+      "step": 5985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0634748519476189e-05,
+      "loss": 0.419,
+      "step": 5986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0633080323629993e-05,
+      "loss": 0.4246,
+      "step": 5987
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0631412127783803e-05,
+      "loss": 0.4097,
+      "step": 5988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.062974393193761e-05,
+      "loss": 0.399,
+      "step": 5989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0628075736091419e-05,
+      "loss": 0.4106,
+      "step": 5990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0626407540245225e-05,
+      "loss": 0.4178,
+      "step": 5991
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0624739344399033e-05,
+      "loss": 0.4086,
+      "step": 5992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0623071148552842e-05,
+      "loss": 0.4272,
+      "step": 5993
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0621402952706647e-05,
+      "loss": 0.3941,
+      "step": 5994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0619734756860456e-05,
+      "loss": 0.395,
+      "step": 5995
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0618066561014264e-05,
+      "loss": 0.4023,
+      "step": 5996
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0616398365168072e-05,
+      "loss": 0.402,
+      "step": 5997
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0614730169321878e-05,
+      "loss": 0.403,
+      "step": 5998
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0613061973475686e-05,
+      "loss": 0.4161,
+      "step": 5999
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0611393777629496e-05,
+      "loss": 0.409,
+      "step": 6000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0609725581783302e-05,
+      "loss": 0.3738,
+      "step": 6001
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.060805738593711e-05,
+      "loss": 0.4164,
+      "step": 6002
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0606389190090918e-05,
+      "loss": 0.3883,
+      "step": 6003
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0604720994244726e-05,
+      "loss": 0.4314,
+      "step": 6004
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0603052798398532e-05,
+      "loss": 0.3712,
+      "step": 6005
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.060138460255234e-05,
+      "loss": 0.381,
+      "step": 6006
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.059971640670615e-05,
+      "loss": 0.4025,
+      "step": 6007
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0598048210859955e-05,
+      "loss": 0.4035,
+      "step": 6008
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0596380015013763e-05,
+      "loss": 0.4095,
+      "step": 6009
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0594711819167571e-05,
+      "loss": 0.3982,
+      "step": 6010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0593043623321379e-05,
+      "loss": 0.4142,
+      "step": 6011
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0591375427475185e-05,
+      "loss": 0.3827,
+      "step": 6012
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0589707231628993e-05,
+      "loss": 0.3688,
+      "step": 6013
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0588039035782803e-05,
+      "loss": 0.4315,
+      "step": 6014
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.058637083993661e-05,
+      "loss": 0.4027,
+      "step": 6015
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0584702644090417e-05,
+      "loss": 0.4225,
+      "step": 6016
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0583034448244225e-05,
+      "loss": 0.4104,
+      "step": 6017
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0581366252398033e-05,
+      "loss": 0.4048,
+      "step": 6018
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0579698056551839e-05,
+      "loss": 0.3856,
+      "step": 6019
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0578029860705647e-05,
+      "loss": 0.4016,
+      "step": 6020
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0576361664859456e-05,
+      "loss": 0.4244,
+      "step": 6021
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0574693469013264e-05,
+      "loss": 0.3919,
+      "step": 6022
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.057302527316707e-05,
+      "loss": 0.3858,
+      "step": 6023
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0571357077320878e-05,
+      "loss": 0.3981,
+      "step": 6024
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0569688881474686e-05,
+      "loss": 0.3856,
+      "step": 6025
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0568020685628492e-05,
+      "loss": 0.3659,
+      "step": 6026
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0566352489782302e-05,
+      "loss": 0.4163,
+      "step": 6027
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.056468429393611e-05,
+      "loss": 0.3913,
+      "step": 6028
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0563016098089918e-05,
+      "loss": 0.3883,
+      "step": 6029
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0561347902243724e-05,
+      "loss": 0.3755,
+      "step": 6030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0559679706397532e-05,
+      "loss": 0.4059,
+      "step": 6031
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.055801151055134e-05,
+      "loss": 0.4066,
+      "step": 6032
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0556343314705146e-05,
+      "loss": 0.3888,
+      "step": 6033
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0554675118858955e-05,
+      "loss": 0.4148,
+      "step": 6034
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0553006923012763e-05,
+      "loss": 0.435,
+      "step": 6035
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0551338727166571e-05,
+      "loss": 0.3888,
+      "step": 6036
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0549670531320377e-05,
+      "loss": 0.4053,
+      "step": 6037
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0548002335474185e-05,
+      "loss": 0.4046,
+      "step": 6038
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0546334139627993e-05,
+      "loss": 0.3873,
+      "step": 6039
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.05446659437818e-05,
+      "loss": 0.4364,
+      "step": 6040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0542997747935609e-05,
+      "loss": 0.3904,
+      "step": 6041
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0541329552089417e-05,
+      "loss": 0.4153,
+      "step": 6042
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0539661356243225e-05,
+      "loss": 0.3888,
+      "step": 6043
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0537993160397031e-05,
+      "loss": 0.4048,
+      "step": 6044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0536324964550839e-05,
+      "loss": 0.4013,
+      "step": 6045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0534656768704647e-05,
+      "loss": 0.4227,
+      "step": 6046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0532988572858453e-05,
+      "loss": 0.3918,
+      "step": 6047
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0531320377012262e-05,
+      "loss": 0.363,
+      "step": 6048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.052965218116607e-05,
+      "loss": 0.3895,
+      "step": 6049
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0527983985319878e-05,
+      "loss": 0.4162,
+      "step": 6050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.4221,
+      "step": 6051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0524647593627492e-05,
+      "loss": 0.3856,
+      "step": 6052
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.05229793977813e-05,
+      "loss": 0.4102,
+      "step": 6053
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0521311201935106e-05,
+      "loss": 0.4146,
+      "step": 6054
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0519643006088916e-05,
+      "loss": 0.4264,
+      "step": 6055
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0517974810242724e-05,
+      "loss": 0.3984,
+      "step": 6056
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0516306614396532e-05,
+      "loss": 0.3995,
+      "step": 6057
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0514638418550338e-05,
+      "loss": 0.3914,
+      "step": 6058
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0512970222704146e-05,
+      "loss": 0.394,
+      "step": 6059
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0511302026857955e-05,
+      "loss": 0.4302,
+      "step": 6060
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.050963383101176e-05,
+      "loss": 0.3884,
+      "step": 6061
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.050796563516557e-05,
+      "loss": 0.3976,
+      "step": 6062
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0506297439319377e-05,
+      "loss": 0.3924,
+      "step": 6063
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0504629243473185e-05,
+      "loss": 0.4177,
+      "step": 6064
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0502961047626991e-05,
+      "loss": 0.3781,
+      "step": 6065
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.05012928517808e-05,
+      "loss": 0.3943,
+      "step": 6066
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0499624655934609e-05,
+      "loss": 0.3758,
+      "step": 6067
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0497956460088413e-05,
+      "loss": 0.4033,
+      "step": 6068
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0496288264242223e-05,
+      "loss": 0.4008,
+      "step": 6069
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0494620068396031e-05,
+      "loss": 0.386,
+      "step": 6070
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0492951872549839e-05,
+      "loss": 0.4202,
+      "step": 6071
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0491283676703645e-05,
+      "loss": 0.4007,
+      "step": 6072
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0489615480857453e-05,
+      "loss": 0.4063,
+      "step": 6073
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0487947285011262e-05,
+      "loss": 0.4029,
+      "step": 6074
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0486279089165069e-05,
+      "loss": 0.3967,
+      "step": 6075
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0484610893318877e-05,
+      "loss": 0.4162,
+      "step": 6076
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0482942697472684e-05,
+      "loss": 0.4589,
+      "step": 6077
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0481274501626492e-05,
+      "loss": 0.4029,
+      "step": 6078
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0479606305780299e-05,
+      "loss": 0.411,
+      "step": 6079
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0477938109934106e-05,
+      "loss": 0.3917,
+      "step": 6080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0476269914087916e-05,
+      "loss": 0.4045,
+      "step": 6081
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0474601718241722e-05,
+      "loss": 0.3594,
+      "step": 6082
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.047293352239553e-05,
+      "loss": 0.4066,
+      "step": 6083
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0471265326549338e-05,
+      "loss": 0.4134,
+      "step": 6084
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0469597130703146e-05,
+      "loss": 0.4019,
+      "step": 6085
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0467928934856952e-05,
+      "loss": 0.4214,
+      "step": 6086
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.046626073901076e-05,
+      "loss": 0.4335,
+      "step": 6087
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.046459254316457e-05,
+      "loss": 0.4071,
+      "step": 6088
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0462924347318376e-05,
+      "loss": 0.3922,
+      "step": 6089
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0461256151472184e-05,
+      "loss": 0.3964,
+      "step": 6090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0459587955625991e-05,
+      "loss": 0.4145,
+      "step": 6091
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.04579197597798e-05,
+      "loss": 0.4059,
+      "step": 6092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0456251563933606e-05,
+      "loss": 0.4337,
+      "step": 6093
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0454583368087413e-05,
+      "loss": 0.4037,
+      "step": 6094
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0452915172241223e-05,
+      "loss": 0.4351,
+      "step": 6095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.045124697639503e-05,
+      "loss": 0.4227,
+      "step": 6096
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0449578780548837e-05,
+      "loss": 0.4568,
+      "step": 6097
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0447910584702645e-05,
+      "loss": 0.3888,
+      "step": 6098
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0446242388856453e-05,
+      "loss": 0.3697,
+      "step": 6099
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0444574193010259e-05,
+      "loss": 0.4227,
+      "step": 6100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0442905997164069e-05,
+      "loss": 0.3886,
+      "step": 6101
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0441237801317877e-05,
+      "loss": 0.3979,
+      "step": 6102
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0439569605471683e-05,
+      "loss": 0.4088,
+      "step": 6103
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.043790140962549e-05,
+      "loss": 0.4282,
+      "step": 6104
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0436233213779298e-05,
+      "loss": 0.3762,
+      "step": 6105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0434565017933106e-05,
+      "loss": 0.3926,
+      "step": 6106
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0432896822086913e-05,
+      "loss": 0.4115,
+      "step": 6107
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0431228626240722e-05,
+      "loss": 0.3915,
+      "step": 6108
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.042956043039453e-05,
+      "loss": 0.3933,
+      "step": 6109
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0427892234548336e-05,
+      "loss": 0.3984,
+      "step": 6110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0426224038702144e-05,
+      "loss": 0.3813,
+      "step": 6111
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0424555842855952e-05,
+      "loss": 0.4147,
+      "step": 6112
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.042288764700976e-05,
+      "loss": 0.4246,
+      "step": 6113
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0421219451163566e-05,
+      "loss": 0.3863,
+      "step": 6114
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0419551255317376e-05,
+      "loss": 0.4041,
+      "step": 6115
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0417883059471184e-05,
+      "loss": 0.4057,
+      "step": 6116
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.041621486362499e-05,
+      "loss": 0.396,
+      "step": 6117
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0414546667778798e-05,
+      "loss": 0.4074,
+      "step": 6118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0412878471932606e-05,
+      "loss": 0.3948,
+      "step": 6119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0411210276086413e-05,
+      "loss": 0.3846,
+      "step": 6120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.040954208024022e-05,
+      "loss": 0.4041,
+      "step": 6121
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.040787388439403e-05,
+      "loss": 0.3819,
+      "step": 6122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0406205688547837e-05,
+      "loss": 0.3949,
+      "step": 6123
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0404537492701643e-05,
+      "loss": 0.4138,
+      "step": 6124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0402869296855451e-05,
+      "loss": 0.3711,
+      "step": 6125
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0401201101009259e-05,
+      "loss": 0.3975,
+      "step": 6126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0399532905163069e-05,
+      "loss": 0.3985,
+      "step": 6127
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0397864709316873e-05,
+      "loss": 0.4133,
+      "step": 6128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0396196513470683e-05,
+      "loss": 0.3978,
+      "step": 6129
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.039452831762449e-05,
+      "loss": 0.4092,
+      "step": 6130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0392860121778298e-05,
+      "loss": 0.4078,
+      "step": 6131
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0391191925932105e-05,
+      "loss": 0.3913,
+      "step": 6132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0389523730085913e-05,
+      "loss": 0.4195,
+      "step": 6133
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0387855534239722e-05,
+      "loss": 0.3992,
+      "step": 6134
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0386187338393527e-05,
+      "loss": 0.4117,
+      "step": 6135
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0384519142547336e-05,
+      "loss": 0.3915,
+      "step": 6136
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0382850946701144e-05,
+      "loss": 0.3878,
+      "step": 6137
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0381182750854952e-05,
+      "loss": 0.4235,
+      "step": 6138
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0379514555008758e-05,
+      "loss": 0.4033,
+      "step": 6139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0377846359162566e-05,
+      "loss": 0.4054,
+      "step": 6140
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0376178163316376e-05,
+      "loss": 0.3906,
+      "step": 6141
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0374509967470182e-05,
+      "loss": 0.4187,
+      "step": 6142
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.037284177162399e-05,
+      "loss": 0.4058,
+      "step": 6143
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0371173575777798e-05,
+      "loss": 0.4059,
+      "step": 6144
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0369505379931606e-05,
+      "loss": 0.4109,
+      "step": 6145
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0367837184085412e-05,
+      "loss": 0.4363,
+      "step": 6146
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.036616898823922e-05,
+      "loss": 0.4107,
+      "step": 6147
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.036450079239303e-05,
+      "loss": 0.3753,
+      "step": 6148
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0362832596546835e-05,
+      "loss": 0.3987,
+      "step": 6149
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0361164400700643e-05,
+      "loss": 0.4543,
+      "step": 6150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0359496204854451e-05,
+      "loss": 0.4135,
+      "step": 6151
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0357828009008259e-05,
+      "loss": 0.3717,
+      "step": 6152
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0356159813162065e-05,
+      "loss": 0.3809,
+      "step": 6153
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0354491617315873e-05,
+      "loss": 0.4239,
+      "step": 6154
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0352823421469683e-05,
+      "loss": 0.395,
+      "step": 6155
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0351155225623489e-05,
+      "loss": 0.4084,
+      "step": 6156
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0349487029777297e-05,
+      "loss": 0.3982,
+      "step": 6157
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0347818833931105e-05,
+      "loss": 0.4019,
+      "step": 6158
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0346150638084913e-05,
+      "loss": 0.3786,
+      "step": 6159
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0344482442238719e-05,
+      "loss": 0.3869,
+      "step": 6160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0342814246392527e-05,
+      "loss": 0.3988,
+      "step": 6161
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0341146050546336e-05,
+      "loss": 0.3946,
+      "step": 6162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0339477854700142e-05,
+      "loss": 0.4137,
+      "step": 6163
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.033780965885395e-05,
+      "loss": 0.4221,
+      "step": 6164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0336141463007758e-05,
+      "loss": 0.4234,
+      "step": 6165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0334473267161566e-05,
+      "loss": 0.3844,
+      "step": 6166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0332805071315372e-05,
+      "loss": 0.4021,
+      "step": 6167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0331136875469182e-05,
+      "loss": 0.3991,
+      "step": 6168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.032946867962299e-05,
+      "loss": 0.4092,
+      "step": 6169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0327800483776796e-05,
+      "loss": 0.4001,
+      "step": 6170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0326132287930604e-05,
+      "loss": 0.385,
+      "step": 6171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0324464092084412e-05,
+      "loss": 0.4008,
+      "step": 6172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.032279589623822e-05,
+      "loss": 0.4073,
+      "step": 6173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0321127700392026e-05,
+      "loss": 0.4113,
+      "step": 6174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0319459504545835e-05,
+      "loss": 0.3548,
+      "step": 6175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0317791308699643e-05,
+      "loss": 0.3735,
+      "step": 6176
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.031612311285345e-05,
+      "loss": 0.3984,
+      "step": 6177
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0314454917007257e-05,
+      "loss": 0.3818,
+      "step": 6178
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0312786721161065e-05,
+      "loss": 0.441,
+      "step": 6179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0311118525314873e-05,
+      "loss": 0.3933,
+      "step": 6180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.030945032946868e-05,
+      "loss": 0.3899,
+      "step": 6181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0307782133622489e-05,
+      "loss": 0.4157,
+      "step": 6182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0306113937776297e-05,
+      "loss": 0.3813,
+      "step": 6183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0304445741930103e-05,
+      "loss": 0.3859,
+      "step": 6184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.030277754608391e-05,
+      "loss": 0.4231,
+      "step": 6185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0301109350237719e-05,
+      "loss": 0.3874,
+      "step": 6186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0299441154391527e-05,
+      "loss": 0.4134,
+      "step": 6187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0297772958545333e-05,
+      "loss": 0.3885,
+      "step": 6188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0296104762699142e-05,
+      "loss": 0.3706,
+      "step": 6189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.029443656685295e-05,
+      "loss": 0.4291,
+      "step": 6190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0292768371006756e-05,
+      "loss": 0.3727,
+      "step": 6191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0291100175160564e-05,
+      "loss": 0.3886,
+      "step": 6192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0289431979314372e-05,
+      "loss": 0.4281,
+      "step": 6193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.028776378346818e-05,
+      "loss": 0.4039,
+      "step": 6194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0286095587621986e-05,
+      "loss": 0.3961,
+      "step": 6195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0284427391775796e-05,
+      "loss": 0.4016,
+      "step": 6196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0282759195929604e-05,
+      "loss": 0.4183,
+      "step": 6197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.028109100008341e-05,
+      "loss": 0.4275,
+      "step": 6198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0279422804237218e-05,
+      "loss": 0.4211,
+      "step": 6199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0277754608391026e-05,
+      "loss": 0.3875,
+      "step": 6200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0276086412544835e-05,
+      "loss": 0.3526,
+      "step": 6201
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.027441821669864e-05,
+      "loss": 0.3694,
+      "step": 6202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.027275002085245e-05,
+      "loss": 0.4022,
+      "step": 6203
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0271081825006257e-05,
+      "loss": 0.3735,
+      "step": 6204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0269413629160063e-05,
+      "loss": 0.4022,
+      "step": 6205
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0267745433313871e-05,
+      "loss": 0.3806,
+      "step": 6206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.026607723746768e-05,
+      "loss": 0.4208,
+      "step": 6207
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0264409041621489e-05,
+      "loss": 0.382,
+      "step": 6208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0262740845775293e-05,
+      "loss": 0.3885,
+      "step": 6209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0261072649929103e-05,
+      "loss": 0.3843,
+      "step": 6210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.025940445408291e-05,
+      "loss": 0.4013,
+      "step": 6211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0257736258236717e-05,
+      "loss": 0.3738,
+      "step": 6212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0256068062390525e-05,
+      "loss": 0.399,
+      "step": 6213
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0254399866544333e-05,
+      "loss": 0.3968,
+      "step": 6214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0252731670698142e-05,
+      "loss": 0.3953,
+      "step": 6215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0251063474851949e-05,
+      "loss": 0.4396,
+      "step": 6216
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0249395279005756e-05,
+      "loss": 0.3996,
+      "step": 6217
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0247727083159564e-05,
+      "loss": 0.4305,
+      "step": 6218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.024605888731337e-05,
+      "loss": 0.4194,
+      "step": 6219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0244390691467178e-05,
+      "loss": 0.4069,
+      "step": 6220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0242722495620986e-05,
+      "loss": 0.4102,
+      "step": 6221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0241054299774796e-05,
+      "loss": 0.4194,
+      "step": 6222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0239386103928602e-05,
+      "loss": 0.4092,
+      "step": 6223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.023771790808241e-05,
+      "loss": 0.374,
+      "step": 6224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0236049712236218e-05,
+      "loss": 0.3989,
+      "step": 6225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0234381516390024e-05,
+      "loss": 0.3772,
+      "step": 6226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0232713320543832e-05,
+      "loss": 0.3879,
+      "step": 6227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.023104512469764e-05,
+      "loss": 0.3954,
+      "step": 6228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.022937692885145e-05,
+      "loss": 0.3969,
+      "step": 6229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0227708733005256e-05,
+      "loss": 0.4031,
+      "step": 6230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0226040537159063e-05,
+      "loss": 0.3928,
+      "step": 6231
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0224372341312871e-05,
+      "loss": 0.3629,
+      "step": 6232
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0222704145466678e-05,
+      "loss": 0.3796,
+      "step": 6233
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0221035949620485e-05,
+      "loss": 0.4094,
+      "step": 6234
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0219367753774293e-05,
+      "loss": 0.4307,
+      "step": 6235
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0217699557928103e-05,
+      "loss": 0.4014,
+      "step": 6236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0216031362081909e-05,
+      "loss": 0.405,
+      "step": 6237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0214363166235717e-05,
+      "loss": 0.4042,
+      "step": 6238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0212694970389525e-05,
+      "loss": 0.3885,
+      "step": 6239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0211026774543331e-05,
+      "loss": 0.406,
+      "step": 6240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0209358578697139e-05,
+      "loss": 0.3985,
+      "step": 6241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0207690382850949e-05,
+      "loss": 0.4066,
+      "step": 6242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0206022187004756e-05,
+      "loss": 0.4101,
+      "step": 6243
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0204353991158563e-05,
+      "loss": 0.4051,
+      "step": 6244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.020268579531237e-05,
+      "loss": 0.3615,
+      "step": 6245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0201017599466178e-05,
+      "loss": 0.368,
+      "step": 6246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0199349403619985e-05,
+      "loss": 0.3849,
+      "step": 6247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0197681207773792e-05,
+      "loss": 0.3951,
+      "step": 6248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0196013011927602e-05,
+      "loss": 0.4017,
+      "step": 6249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.019434481608141e-05,
+      "loss": 0.4043,
+      "step": 6250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0192676620235216e-05,
+      "loss": 0.3918,
+      "step": 6251
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0191008424389024e-05,
+      "loss": 0.4015,
+      "step": 6252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0189340228542832e-05,
+      "loss": 0.3797,
+      "step": 6253
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.018767203269664e-05,
+      "loss": 0.4339,
+      "step": 6254
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0186003836850446e-05,
+      "loss": 0.4124,
+      "step": 6255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0184335641004256e-05,
+      "loss": 0.3718,
+      "step": 6256
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0182667445158063e-05,
+      "loss": 0.4045,
+      "step": 6257
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.018099924931187e-05,
+      "loss": 0.4104,
+      "step": 6258
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0179331053465678e-05,
+      "loss": 0.3845,
+      "step": 6259
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0177662857619485e-05,
+      "loss": 0.3916,
+      "step": 6260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0175994661773293e-05,
+      "loss": 0.4123,
+      "step": 6261
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.01743264659271e-05,
+      "loss": 0.3675,
+      "step": 6262
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0172658270080909e-05,
+      "loss": 0.3701,
+      "step": 6263
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0170990074234717e-05,
+      "loss": 0.3922,
+      "step": 6264
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0169321878388523e-05,
+      "loss": 0.3871,
+      "step": 6265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0167653682542331e-05,
+      "loss": 0.432,
+      "step": 6266
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0165985486696139e-05,
+      "loss": 0.4201,
+      "step": 6267
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0164317290849949e-05,
+      "loss": 0.3965,
+      "step": 6268
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0162649095003753e-05,
+      "loss": 0.3881,
+      "step": 6269
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0160980899157563e-05,
+      "loss": 0.4033,
+      "step": 6270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.015931270331137e-05,
+      "loss": 0.3835,
+      "step": 6271
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0157644507465177e-05,
+      "loss": 0.4175,
+      "step": 6272
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0155976311618985e-05,
+      "loss": 0.4025,
+      "step": 6273
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0154308115772792e-05,
+      "loss": 0.3657,
+      "step": 6274
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0152639919926602e-05,
+      "loss": 0.3895,
+      "step": 6275
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0150971724080407e-05,
+      "loss": 0.4088,
+      "step": 6276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0149303528234216e-05,
+      "loss": 0.3992,
+      "step": 6277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0147635332388024e-05,
+      "loss": 0.3866,
+      "step": 6278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.014596713654183e-05,
+      "loss": 0.3827,
+      "step": 6279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0144298940695638e-05,
+      "loss": 0.3681,
+      "step": 6280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0142630744849446e-05,
+      "loss": 0.3841,
+      "step": 6281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0140962549003256e-05,
+      "loss": 0.3724,
+      "step": 6282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0139294353157062e-05,
+      "loss": 0.3884,
+      "step": 6283
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013762615731087e-05,
+      "loss": 0.3972,
+      "step": 6284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135957961464678e-05,
+      "loss": 0.3707,
+      "step": 6285
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0134289765618484e-05,
+      "loss": 0.3846,
+      "step": 6286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0132621569772292e-05,
+      "loss": 0.4167,
+      "step": 6287
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.01309533739261e-05,
+      "loss": 0.3922,
+      "step": 6288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129285178079909e-05,
+      "loss": 0.3737,
+      "step": 6289
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127616982233715e-05,
+      "loss": 0.4268,
+      "step": 6290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125948786387523e-05,
+      "loss": 0.3873,
+      "step": 6291
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0124280590541331e-05,
+      "loss": 0.4072,
+      "step": 6292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122612394695137e-05,
+      "loss": 0.4105,
+      "step": 6293
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120944198848945e-05,
+      "loss": 0.4285,
+      "step": 6294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119276003002753e-05,
+      "loss": 0.391,
+      "step": 6295
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0117607807156563e-05,
+      "loss": 0.41,
+      "step": 6296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0115939611310369e-05,
+      "loss": 0.4019,
+      "step": 6297
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0114271415464177e-05,
+      "loss": 0.4051,
+      "step": 6298
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112603219617985e-05,
+      "loss": 0.3962,
+      "step": 6299
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011093502377179e-05,
+      "loss": 0.3784,
+      "step": 6300
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109266827925599e-05,
+      "loss": 0.4001,
+      "step": 6301
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0107598632079407e-05,
+      "loss": 0.4089,
+      "step": 6302
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0105930436233216e-05,
+      "loss": 0.3976,
+      "step": 6303
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0104262240387022e-05,
+      "loss": 0.3849,
+      "step": 6304
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010259404454083e-05,
+      "loss": 0.4244,
+      "step": 6305
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0100925848694638e-05,
+      "loss": 0.4283,
+      "step": 6306
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0099257652848444e-05,
+      "loss": 0.3855,
+      "step": 6307
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0097589457002252e-05,
+      "loss": 0.3837,
+      "step": 6308
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.009592126115606e-05,
+      "loss": 0.3643,
+      "step": 6309
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.009425306530987e-05,
+      "loss": 0.4025,
+      "step": 6310
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0092584869463676e-05,
+      "loss": 0.389,
+      "step": 6311
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0090916673617484e-05,
+      "loss": 0.4022,
+      "step": 6312
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0089248477771292e-05,
+      "loss": 0.4029,
+      "step": 6313
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0087580281925098e-05,
+      "loss": 0.4103,
+      "step": 6314
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0085912086078906e-05,
+      "loss": 0.3918,
+      "step": 6315
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0084243890232715e-05,
+      "loss": 0.4159,
+      "step": 6316
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0082575694386523e-05,
+      "loss": 0.4086,
+      "step": 6317
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.008090749854033e-05,
+      "loss": 0.4144,
+      "step": 6318
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0079239302694137e-05,
+      "loss": 0.3803,
+      "step": 6319
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0077571106847945e-05,
+      "loss": 0.3827,
+      "step": 6320
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0075902911001751e-05,
+      "loss": 0.4254,
+      "step": 6321
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.007423471515556e-05,
+      "loss": 0.3874,
+      "step": 6322
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0072566519309369e-05,
+      "loss": 0.4242,
+      "step": 6323
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0070898323463177e-05,
+      "loss": 0.4173,
+      "step": 6324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069230127616983e-05,
+      "loss": 0.4067,
+      "step": 6325
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006756193177079e-05,
+      "loss": 0.4075,
+      "step": 6326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0065893735924599e-05,
+      "loss": 0.3815,
+      "step": 6327
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064225540078405e-05,
+      "loss": 0.3974,
+      "step": 6328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062557344232213e-05,
+      "loss": 0.3738,
+      "step": 6329
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060889148386022e-05,
+      "loss": 0.4017,
+      "step": 6330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005922095253983e-05,
+      "loss": 0.413,
+      "step": 6331
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057552756693636e-05,
+      "loss": 0.3985,
+      "step": 6332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055884560847444e-05,
+      "loss": 0.4031,
+      "step": 6333
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054216365001252e-05,
+      "loss": 0.3844,
+      "step": 6334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0052548169155058e-05,
+      "loss": 0.3679,
+      "step": 6335
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050879973308866e-05,
+      "loss": 0.3451,
+      "step": 6336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049211777462676e-05,
+      "loss": 0.4271,
+      "step": 6337
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047543581616484e-05,
+      "loss": 0.4276,
+      "step": 6338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004587538577029e-05,
+      "loss": 0.3978,
+      "step": 6339
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044207189924098e-05,
+      "loss": 0.3904,
+      "step": 6340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0042538994077906e-05,
+      "loss": 0.3976,
+      "step": 6341
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0040870798231712e-05,
+      "loss": 0.3862,
+      "step": 6342
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003920260238552e-05,
+      "loss": 0.3824,
+      "step": 6343
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003753440653933e-05,
+      "loss": 0.4218,
+      "step": 6344
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0035866210693137e-05,
+      "loss": 0.4061,
+      "step": 6345
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034198014846943e-05,
+      "loss": 0.3832,
+      "step": 6346
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032529819000751e-05,
+      "loss": 0.3869,
+      "step": 6347
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030861623154559e-05,
+      "loss": 0.3859,
+      "step": 6348
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029193427308365e-05,
+      "loss": 0.3781,
+      "step": 6349
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027525231462173e-05,
+      "loss": 0.3997,
+      "step": 6350
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0025857035615983e-05,
+      "loss": 0.3736,
+      "step": 6351
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002418883976979e-05,
+      "loss": 0.3859,
+      "step": 6352
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022520643923597e-05,
+      "loss": 0.4442,
+      "step": 6353
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020852448077405e-05,
+      "loss": 0.4309,
+      "step": 6354
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019184252231213e-05,
+      "loss": 0.4019,
+      "step": 6355
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017516056385019e-05,
+      "loss": 0.3997,
+      "step": 6356
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015847860538828e-05,
+      "loss": 0.3825,
+      "step": 6357
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014179664692636e-05,
+      "loss": 0.4294,
+      "step": 6358
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0012511468846444e-05,
+      "loss": 0.4143,
+      "step": 6359
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001084327300025e-05,
+      "loss": 0.4013,
+      "step": 6360
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009175077154058e-05,
+      "loss": 0.4099,
+      "step": 6361
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007506881307866e-05,
+      "loss": 0.4305,
+      "step": 6362
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005838685461672e-05,
+      "loss": 0.3848,
+      "step": 6363
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0004170489615482e-05,
+      "loss": 0.379,
+      "step": 6364
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.000250229376929e-05,
+      "loss": 0.4027,
+      "step": 6365
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0000834097923098e-05,
+      "loss": 0.3922,
+      "step": 6366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.999165902076906e-06,
+      "loss": 0.3791,
+      "step": 6367
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.997497706230712e-06,
+      "loss": 0.3889,
+      "step": 6368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99582951038452e-06,
+      "loss": 0.4317,
+      "step": 6369
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994161314538328e-06,
+      "loss": 0.3421,
+      "step": 6370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992493118692135e-06,
+      "loss": 0.4246,
+      "step": 6371
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990824922845943e-06,
+      "loss": 0.3907,
+      "step": 6372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98915672699975e-06,
+      "loss": 0.3782,
+      "step": 6373
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987488531153559e-06,
+      "loss": 0.3816,
+      "step": 6374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985820335307365e-06,
+      "loss": 0.4062,
+      "step": 6375
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984152139461173e-06,
+      "loss": 0.3999,
+      "step": 6376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.982483943614981e-06,
+      "loss": 0.3902,
+      "step": 6377
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980815747768789e-06,
+      "loss": 0.4229,
+      "step": 6378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979147551922597e-06,
+      "loss": 0.4092,
+      "step": 6379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977479356076403e-06,
+      "loss": 0.4005,
+      "step": 6380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975811160230213e-06,
+      "loss": 0.4407,
+      "step": 6381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.974142964384019e-06,
+      "loss": 0.3961,
+      "step": 6382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972474768537828e-06,
+      "loss": 0.3601,
+      "step": 6383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970806572691635e-06,
+      "loss": 0.3888,
+      "step": 6384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969138376845442e-06,
+      "loss": 0.4261,
+      "step": 6385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96747018099925e-06,
+      "loss": 0.3821,
+      "step": 6386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965801985153057e-06,
+      "loss": 0.3996,
+      "step": 6387
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.964133789306866e-06,
+      "loss": 0.4184,
+      "step": 6388
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.962465593460672e-06,
+      "loss": 0.3766,
+      "step": 6389
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960797397614482e-06,
+      "loss": 0.3853,
+      "step": 6390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.959129201768288e-06,
+      "loss": 0.412,
+      "step": 6391
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.957461005922096e-06,
+      "loss": 0.4183,
+      "step": 6392
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955792810075904e-06,
+      "loss": 0.3821,
+      "step": 6393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954124614229712e-06,
+      "loss": 0.386,
+      "step": 6394
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95245641838352e-06,
+      "loss": 0.407,
+      "step": 6395
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950788222537326e-06,
+      "loss": 0.4039,
+      "step": 6396
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949120026691135e-06,
+      "loss": 0.4023,
+      "step": 6397
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947451830844942e-06,
+      "loss": 0.3827,
+      "step": 6398
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94578363499875e-06,
+      "loss": 0.4141,
+      "step": 6399
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.944115439152557e-06,
+      "loss": 0.388,
+      "step": 6400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942447243306365e-06,
+      "loss": 0.3826,
+      "step": 6401
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940779047460173e-06,
+      "loss": 0.4207,
+      "step": 6402
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93911085161398e-06,
+      "loss": 0.3704,
+      "step": 6403
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937442655767789e-06,
+      "loss": 0.382,
+      "step": 6404
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935774459921595e-06,
+      "loss": 0.4234,
+      "step": 6405
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934106264075403e-06,
+      "loss": 0.3451,
+      "step": 6406
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.932438068229211e-06,
+      "loss": 0.3878,
+      "step": 6407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.930769872383019e-06,
+      "loss": 0.4,
+      "step": 6408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.929101676536827e-06,
+      "loss": 0.4069,
+      "step": 6409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.927433480690633e-06,
+      "loss": 0.4253,
+      "step": 6410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.925765284844442e-06,
+      "loss": 0.4208,
+      "step": 6411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.924097088998249e-06,
+      "loss": 0.397,
+      "step": 6412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.922428893152057e-06,
+      "loss": 0.3974,
+      "step": 6413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.920760697305864e-06,
+      "loss": 0.4208,
+      "step": 6414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.919092501459672e-06,
+      "loss": 0.3974,
+      "step": 6415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.91742430561348e-06,
+      "loss": 0.426,
+      "step": 6416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.915756109767286e-06,
+      "loss": 0.4065,
+      "step": 6417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.914087913921096e-06,
+      "loss": 0.3979,
+      "step": 6418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.912419718074902e-06,
+      "loss": 0.382,
+      "step": 6419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.910751522228712e-06,
+      "loss": 0.378,
+      "step": 6420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.909083326382518e-06,
+      "loss": 0.4424,
+      "step": 6421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.907415130536326e-06,
+      "loss": 0.3891,
+      "step": 6422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.905746934690134e-06,
+      "loss": 0.3908,
+      "step": 6423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.904078738843942e-06,
+      "loss": 0.3904,
+      "step": 6424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.90241054299775e-06,
+      "loss": 0.3899,
+      "step": 6425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.900742347151556e-06,
+      "loss": 0.3987,
+      "step": 6426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.899074151305365e-06,
+      "loss": 0.4301,
+      "step": 6427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.897405955459171e-06,
+      "loss": 0.4336,
+      "step": 6428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.89573775961298e-06,
+      "loss": 0.4046,
+      "step": 6429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.894069563766787e-06,
+      "loss": 0.3893,
+      "step": 6430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.892401367920595e-06,
+      "loss": 0.4101,
+      "step": 6431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.890733172074403e-06,
+      "loss": 0.3723,
+      "step": 6432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88906497622821e-06,
+      "loss": 0.3821,
+      "step": 6433
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.887396780382019e-06,
+      "loss": 0.4072,
+      "step": 6434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.885728584535825e-06,
+      "loss": 0.3992,
+      "step": 6435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884060388689633e-06,
+      "loss": 0.3989,
+      "step": 6436
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88239219284344e-06,
+      "loss": 0.3975,
+      "step": 6437
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880723996997249e-06,
+      "loss": 0.4217,
+      "step": 6438
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.879055801151057e-06,
+      "loss": 0.396,
+      "step": 6439
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877387605304863e-06,
+      "loss": 0.4076,
+      "step": 6440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.875719409458672e-06,
+      "loss": 0.3916,
+      "step": 6441
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874051213612479e-06,
+      "loss": 0.3659,
+      "step": 6442
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872383017766286e-06,
+      "loss": 0.3801,
+      "step": 6443
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870714821920094e-06,
+      "loss": 0.3877,
+      "step": 6444
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.869046626073902e-06,
+      "loss": 0.3933,
+      "step": 6445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86737843022771e-06,
+      "loss": 0.3874,
+      "step": 6446
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.865710234381516e-06,
+      "loss": 0.4241,
+      "step": 6447
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864042038535326e-06,
+      "loss": 0.3741,
+      "step": 6448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.862373842689132e-06,
+      "loss": 0.38,
+      "step": 6449
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.86070564684294e-06,
+      "loss": 0.4003,
+      "step": 6450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.859037450996748e-06,
+      "loss": 0.4033,
+      "step": 6451
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.857369255150556e-06,
+      "loss": 0.3926,
+      "step": 6452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.855701059304364e-06,
+      "loss": 0.3709,
+      "step": 6453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.85403286345817e-06,
+      "loss": 0.3966,
+      "step": 6454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.85236466761198e-06,
+      "loss": 0.3946,
+      "step": 6455
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.850696471765786e-06,
+      "loss": 0.3986,
+      "step": 6456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.849028275919595e-06,
+      "loss": 0.4213,
+      "step": 6457
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.847360080073401e-06,
+      "loss": 0.3819,
+      "step": 6458
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.84569188422721e-06,
+      "loss": 0.3866,
+      "step": 6459
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.844023688381017e-06,
+      "loss": 0.4139,
+      "step": 6460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.842355492534825e-06,
+      "loss": 0.4033,
+      "step": 6461
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.840687296688633e-06,
+      "loss": 0.4026,
+      "step": 6462
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.839019100842439e-06,
+      "loss": 0.3906,
+      "step": 6463
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.837350904996249e-06,
+      "loss": 0.4136,
+      "step": 6464
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.835682709150055e-06,
+      "loss": 0.4054,
+      "step": 6465
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.834014513303863e-06,
+      "loss": 0.4017,
+      "step": 6466
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.83234631745767e-06,
+      "loss": 0.4402,
+      "step": 6467
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.830678121611478e-06,
+      "loss": 0.3841,
+      "step": 6468
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.829009925765286e-06,
+      "loss": 0.3953,
+      "step": 6469
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.827341729919093e-06,
+      "loss": 0.3991,
+      "step": 6470
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.825673534072902e-06,
+      "loss": 0.3916,
+      "step": 6471
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.824005338226708e-06,
+      "loss": 0.419,
+      "step": 6472
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.822337142380516e-06,
+      "loss": 0.4214,
+      "step": 6473
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.820668946534324e-06,
+      "loss": 0.4045,
+      "step": 6474
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.819000750688132e-06,
+      "loss": 0.4237,
+      "step": 6475
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.81733255484194e-06,
+      "loss": 0.4317,
+      "step": 6476
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.815664358995746e-06,
+      "loss": 0.392,
+      "step": 6477
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.813996163149556e-06,
+      "loss": 0.3733,
+      "step": 6478
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.812327967303362e-06,
+      "loss": 0.4256,
+      "step": 6479
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.81065977145717e-06,
+      "loss": 0.4213,
+      "step": 6480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.808991575610978e-06,
+      "loss": 0.3956,
+      "step": 6481
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.807323379764786e-06,
+      "loss": 0.3552,
+      "step": 6482
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.805655183918593e-06,
+      "loss": 0.4301,
+      "step": 6483
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.8039869880724e-06,
+      "loss": 0.4043,
+      "step": 6484
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.80231879222621e-06,
+      "loss": 0.41,
+      "step": 6485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.800650596380015e-06,
+      "loss": 0.3867,
+      "step": 6486
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.798982400533823e-06,
+      "loss": 0.4138,
+      "step": 6487
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.797314204687631e-06,
+      "loss": 0.418,
+      "step": 6488
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.795646008841439e-06,
+      "loss": 0.4048,
+      "step": 6489
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.793977812995247e-06,
+      "loss": 0.406,
+      "step": 6490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.792309617149053e-06,
+      "loss": 0.3682,
+      "step": 6491
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.790641421302863e-06,
+      "loss": 0.4133,
+      "step": 6492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.788973225456669e-06,
+      "loss": 0.3834,
+      "step": 6493
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.787305029610477e-06,
+      "loss": 0.3938,
+      "step": 6494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.785636833764285e-06,
+      "loss": 0.3866,
+      "step": 6495
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.783968637918093e-06,
+      "loss": 0.3925,
+      "step": 6496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.7823004420719e-06,
+      "loss": 0.4075,
+      "step": 6497
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.780632246225708e-06,
+      "loss": 0.4098,
+      "step": 6498
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.778964050379516e-06,
+      "loss": 0.4167,
+      "step": 6499
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.777295854533322e-06,
+      "loss": 0.3709,
+      "step": 6500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.77562765868713e-06,
+      "loss": 0.3905,
+      "step": 6501
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.773959462840938e-06,
+      "loss": 0.374,
+      "step": 6502
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.772291266994746e-06,
+      "loss": 0.3954,
+      "step": 6503
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.770623071148554e-06,
+      "loss": 0.3693,
+      "step": 6504
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.768954875302362e-06,
+      "loss": 0.4392,
+      "step": 6505
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.76728667945617e-06,
+      "loss": 0.3918,
+      "step": 6506
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.765618483609976e-06,
+      "loss": 0.4098,
+      "step": 6507
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.763950287763784e-06,
+      "loss": 0.4031,
+      "step": 6508
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.762282091917592e-06,
+      "loss": 0.3817,
+      "step": 6509
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.7606138960714e-06,
+      "loss": 0.4085,
+      "step": 6510
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.758945700225207e-06,
+      "loss": 0.4094,
+      "step": 6511
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.757277504379015e-06,
+      "loss": 0.383,
+      "step": 6512
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.755609308532823e-06,
+      "loss": 0.4155,
+      "step": 6513
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.75394111268663e-06,
+      "loss": 0.4438,
+      "step": 6514
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.752272916840439e-06,
+      "loss": 0.3957,
+      "step": 6515
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.750604720994245e-06,
+      "loss": 0.4335,
+      "step": 6516
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.748936525148053e-06,
+      "loss": 0.3799,
+      "step": 6517
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.747268329301861e-06,
+      "loss": 0.3845,
+      "step": 6518
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.745600133455669e-06,
+      "loss": 0.4067,
+      "step": 6519
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.743931937609477e-06,
+      "loss": 0.3918,
+      "step": 6520
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.742263741763283e-06,
+      "loss": 0.3813,
+      "step": 6521
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.740595545917093e-06,
+      "loss": 0.4224,
+      "step": 6522
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.738927350070899e-06,
+      "loss": 0.3917,
+      "step": 6523
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.737259154224707e-06,
+      "loss": 0.4025,
+      "step": 6524
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.735590958378514e-06,
+      "loss": 0.3809,
+      "step": 6525
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.733922762532322e-06,
+      "loss": 0.4163,
+      "step": 6526
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.73225456668613e-06,
+      "loss": 0.383,
+      "step": 6527
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.730586370839936e-06,
+      "loss": 0.3922,
+      "step": 6528
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.728918174993746e-06,
+      "loss": 0.3874,
+      "step": 6529
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.727249979147552e-06,
+      "loss": 0.415,
+      "step": 6530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.72558178330136e-06,
+      "loss": 0.4056,
+      "step": 6531
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.723913587455168e-06,
+      "loss": 0.3805,
+      "step": 6532
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.722245391608976e-06,
+      "loss": 0.3963,
+      "step": 6533
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.720577195762784e-06,
+      "loss": 0.3841,
+      "step": 6534
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.718908999916592e-06,
+      "loss": 0.4089,
+      "step": 6535
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.7172408040704e-06,
+      "loss": 0.3642,
+      "step": 6536
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.715572608224206e-06,
+      "loss": 0.4003,
+      "step": 6537
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.713904412378014e-06,
+      "loss": 0.3866,
+      "step": 6538
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.712236216531822e-06,
+      "loss": 0.4279,
+      "step": 6539
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.71056802068563e-06,
+      "loss": 0.3646,
+      "step": 6540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.708899824839437e-06,
+      "loss": 0.4152,
+      "step": 6541
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.707231628993245e-06,
+      "loss": 0.4073,
+      "step": 6542
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.705563433147053e-06,
+      "loss": 0.3585,
+      "step": 6543
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.70389523730086e-06,
+      "loss": 0.4068,
+      "step": 6544
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.702227041454667e-06,
+      "loss": 0.4055,
+      "step": 6545
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.700558845608475e-06,
+      "loss": 0.389,
+      "step": 6546
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.698890649762283e-06,
+      "loss": 0.3688,
+      "step": 6547
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.69722245391609e-06,
+      "loss": 0.4185,
+      "step": 6548
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.695554258069899e-06,
+      "loss": 0.4154,
+      "step": 6549
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.693886062223707e-06,
+      "loss": 0.379,
+      "step": 6550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.692217866377513e-06,
+      "loss": 0.4109,
+      "step": 6551
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.69054967053132e-06,
+      "loss": 0.4124,
+      "step": 6552
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.688881474685129e-06,
+      "loss": 0.379,
+      "step": 6553
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.687213278838936e-06,
+      "loss": 0.3864,
+      "step": 6554
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.685545082992744e-06,
+      "loss": 0.3912,
+      "step": 6555
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.683876887146552e-06,
+      "loss": 0.399,
+      "step": 6556
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.68220869130036e-06,
+      "loss": 0.396,
+      "step": 6557
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.680540495454166e-06,
+      "loss": 0.4205,
+      "step": 6558
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.678872299607974e-06,
+      "loss": 0.4245,
+      "step": 6559
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.677204103761782e-06,
+      "loss": 0.3987,
+      "step": 6560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.67553590791559e-06,
+      "loss": 0.4268,
+      "step": 6561
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.673867712069398e-06,
+      "loss": 0.3904,
+      "step": 6562
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.672199516223206e-06,
+      "loss": 0.3979,
+      "step": 6563
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.670531320377014e-06,
+      "loss": 0.4009,
+      "step": 6564
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.66886312453082e-06,
+      "loss": 0.3778,
+      "step": 6565
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.667194928684628e-06,
+      "loss": 0.4223,
+      "step": 6566
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.665526732838436e-06,
+      "loss": 0.3902,
+      "step": 6567
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.663858536992243e-06,
+      "loss": 0.3901,
+      "step": 6568
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.662190341146051e-06,
+      "loss": 0.4144,
+      "step": 6569
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.66052214529986e-06,
+      "loss": 0.3831,
+      "step": 6570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.658853949453667e-06,
+      "loss": 0.4084,
+      "step": 6571
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.657185753607475e-06,
+      "loss": 0.4058,
+      "step": 6572
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.655517557761281e-06,
+      "loss": 0.3887,
+      "step": 6573
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.653849361915089e-06,
+      "loss": 0.4275,
+      "step": 6574
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.652181166068897e-06,
+      "loss": 0.4236,
+      "step": 6575
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.650512970222705e-06,
+      "loss": 0.3632,
+      "step": 6576
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.648844774376513e-06,
+      "loss": 0.3866,
+      "step": 6577
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.64717657853032e-06,
+      "loss": 0.4213,
+      "step": 6578
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.645508382684129e-06,
+      "loss": 0.4138,
+      "step": 6579
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.643840186837936e-06,
+      "loss": 0.4032,
+      "step": 6580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.642171990991743e-06,
+      "loss": 0.389,
+      "step": 6581
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.64050379514555e-06,
+      "loss": 0.374,
+      "step": 6582
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.638835599299358e-06,
+      "loss": 0.4246,
+      "step": 6583
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.637167403453166e-06,
+      "loss": 0.4001,
+      "step": 6584
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.635499207606974e-06,
+      "loss": 0.4359,
+      "step": 6585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.633831011760782e-06,
+      "loss": 0.3961,
+      "step": 6586
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.63216281591459e-06,
+      "loss": 0.3811,
+      "step": 6587
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.630494620068396e-06,
+      "loss": 0.3867,
+      "step": 6588
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.628826424222204e-06,
+      "loss": 0.4019,
+      "step": 6589
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.627158228376012e-06,
+      "loss": 0.4074,
+      "step": 6590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.62549003252982e-06,
+      "loss": 0.4038,
+      "step": 6591
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.623821836683628e-06,
+      "loss": 0.3856,
+      "step": 6592
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.622153640837436e-06,
+      "loss": 0.3946,
+      "step": 6593
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.620485444991243e-06,
+      "loss": 0.3956,
+      "step": 6594
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.61881724914505e-06,
+      "loss": 0.3993,
+      "step": 6595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.617149053298858e-06,
+      "loss": 0.4093,
+      "step": 6596
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.615480857452665e-06,
+      "loss": 0.3946,
+      "step": 6597
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.613812661606473e-06,
+      "loss": 0.3935,
+      "step": 6598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.612144465760281e-06,
+      "loss": 0.3863,
+      "step": 6599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.610476269914089e-06,
+      "loss": 0.4178,
+      "step": 6600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.608808074067897e-06,
+      "loss": 0.3847,
+      "step": 6601
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.607139878221705e-06,
+      "loss": 0.4072,
+      "step": 6602
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.605471682375511e-06,
+      "loss": 0.4029,
+      "step": 6603
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.603803486529319e-06,
+      "loss": 0.4073,
+      "step": 6604
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.602135290683127e-06,
+      "loss": 0.4072,
+      "step": 6605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.600467094836935e-06,
+      "loss": 0.4342,
+      "step": 6606
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.598798898990743e-06,
+      "loss": 0.4108,
+      "step": 6607
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.59713070314455e-06,
+      "loss": 0.3893,
+      "step": 6608
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.595462507298358e-06,
+      "loss": 0.4132,
+      "step": 6609
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.593794311452165e-06,
+      "loss": 0.4144,
+      "step": 6610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.592126115605972e-06,
+      "loss": 0.4034,
+      "step": 6611
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.59045791975978e-06,
+      "loss": 0.3806,
+      "step": 6612
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.588789723913588e-06,
+      "loss": 0.4019,
+      "step": 6613
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.587121528067396e-06,
+      "loss": 0.3988,
+      "step": 6614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.585453332221204e-06,
+      "loss": 0.3737,
+      "step": 6615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.583785136375012e-06,
+      "loss": 0.3842,
+      "step": 6616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.582116940528818e-06,
+      "loss": 0.3936,
+      "step": 6617
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.580448744682626e-06,
+      "loss": 0.3785,
+      "step": 6618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.578780548836434e-06,
+      "loss": 0.4202,
+      "step": 6619
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.577112352990242e-06,
+      "loss": 0.4211,
+      "step": 6620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.57544415714405e-06,
+      "loss": 0.3975,
+      "step": 6621
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.573775961297858e-06,
+      "loss": 0.433,
+      "step": 6622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.572107765451665e-06,
+      "loss": 0.3753,
+      "step": 6623
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.570439569605472e-06,
+      "loss": 0.3621,
+      "step": 6624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.56877137375928e-06,
+      "loss": 0.3853,
+      "step": 6625
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.567103177913087e-06,
+      "loss": 0.4088,
+      "step": 6626
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.565434982066895e-06,
+      "loss": 0.3646,
+      "step": 6627
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.563766786220703e-06,
+      "loss": 0.3895,
+      "step": 6628
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.562098590374511e-06,
+      "loss": 0.3869,
+      "step": 6629
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.560430394528319e-06,
+      "loss": 0.4038,
+      "step": 6630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.558762198682125e-06,
+      "loss": 0.4176,
+      "step": 6631
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.557094002835933e-06,
+      "loss": 0.3946,
+      "step": 6632
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.555425806989741e-06,
+      "loss": 0.4187,
+      "step": 6633
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.553757611143549e-06,
+      "loss": 0.3787,
+      "step": 6634
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.552089415297357e-06,
+      "loss": 0.3734,
+      "step": 6635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.550421219451165e-06,
+      "loss": 0.3932,
+      "step": 6636
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.548753023604972e-06,
+      "loss": 0.3904,
+      "step": 6637
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.54708482775878e-06,
+      "loss": 0.3882,
+      "step": 6638
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.545416631912588e-06,
+      "loss": 0.4091,
+      "step": 6639
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.543748436066394e-06,
+      "loss": 0.3918,
+      "step": 6640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.542080240220202e-06,
+      "loss": 0.4134,
+      "step": 6641
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.54041204437401e-06,
+      "loss": 0.3913,
+      "step": 6642
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.538743848527818e-06,
+      "loss": 0.3893,
+      "step": 6643
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.537075652681626e-06,
+      "loss": 0.4039,
+      "step": 6644
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.535407456835434e-06,
+      "loss": 0.4053,
+      "step": 6645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.533739260989242e-06,
+      "loss": 0.3985,
+      "step": 6646
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.532071065143048e-06,
+      "loss": 0.4155,
+      "step": 6647
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.530402869296856e-06,
+      "loss": 0.4003,
+      "step": 6648
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.528734673450664e-06,
+      "loss": 0.3726,
+      "step": 6649
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.527066477604472e-06,
+      "loss": 0.4054,
+      "step": 6650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.52539828175828e-06,
+      "loss": 0.4072,
+      "step": 6651
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.523730085912087e-06,
+      "loss": 0.417,
+      "step": 6652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.522061890065895e-06,
+      "loss": 0.4229,
+      "step": 6653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.520393694219701e-06,
+      "loss": 0.4185,
+      "step": 6654
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.51872549837351e-06,
+      "loss": 0.4369,
+      "step": 6655
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.517057302527317e-06,
+      "loss": 0.4155,
+      "step": 6656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.515389106681125e-06,
+      "loss": 0.4068,
+      "step": 6657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.513720910834933e-06,
+      "loss": 0.3867,
+      "step": 6658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.512052714988741e-06,
+      "loss": 0.4191,
+      "step": 6659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.510384519142549e-06,
+      "loss": 0.4139,
+      "step": 6660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.508716323296355e-06,
+      "loss": 0.4023,
+      "step": 6661
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.507048127450163e-06,
+      "loss": 0.3569,
+      "step": 6662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.50537993160397e-06,
+      "loss": 0.3982,
+      "step": 6663
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.503711735757779e-06,
+      "loss": 0.4299,
+      "step": 6664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.502043539911586e-06,
+      "loss": 0.4073,
+      "step": 6665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.500375344065394e-06,
+      "loss": 0.3789,
+      "step": 6666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.498707148219202e-06,
+      "loss": 0.4141,
+      "step": 6667
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.497038952373008e-06,
+      "loss": 0.4215,
+      "step": 6668
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.495370756526816e-06,
+      "loss": 0.3772,
+      "step": 6669
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.493702560680624e-06,
+      "loss": 0.4054,
+      "step": 6670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.492034364834432e-06,
+      "loss": 0.4109,
+      "step": 6671
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.49036616898824e-06,
+      "loss": 0.4094,
+      "step": 6672
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.488697973142048e-06,
+      "loss": 0.4265,
+      "step": 6673
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.487029777295856e-06,
+      "loss": 0.4124,
+      "step": 6674
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.485361581449662e-06,
+      "loss": 0.3823,
+      "step": 6675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.483693385603472e-06,
+      "loss": 0.3924,
+      "step": 6676
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.482025189757278e-06,
+      "loss": 0.4179,
+      "step": 6677
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.480356993911086e-06,
+      "loss": 0.4164,
+      "step": 6678
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.478688798064894e-06,
+      "loss": 0.3893,
+      "step": 6679
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.477020602218701e-06,
+      "loss": 0.4184,
+      "step": 6680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.47535240637251e-06,
+      "loss": 0.4005,
+      "step": 6681
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 0.4539,
+      "step": 6682
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.472016014680125e-06,
+      "loss": 0.3999,
+      "step": 6683
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.470347818833931e-06,
+      "loss": 0.4242,
+      "step": 6684
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.46867962298774e-06,
+      "loss": 0.3946,
+      "step": 6685
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.467011427141547e-06,
+      "loss": 0.4125,
+      "step": 6686
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.465343231295355e-06,
+      "loss": 0.3842,
+      "step": 6687
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.463675035449163e-06,
+      "loss": 0.3852,
+      "step": 6688
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.462006839602969e-06,
+      "loss": 0.403,
+      "step": 6689
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.460338643756779e-06,
+      "loss": 0.3783,
+      "step": 6690
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.458670447910585e-06,
+      "loss": 0.41,
+      "step": 6691
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.457002252064393e-06,
+      "loss": 0.3752,
+      "step": 6692
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.4553340562182e-06,
+      "loss": 0.4045,
+      "step": 6693
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.453665860372008e-06,
+      "loss": 0.3637,
+      "step": 6694
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.451997664525816e-06,
+      "loss": 0.4195,
+      "step": 6695
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.450329468679624e-06,
+      "loss": 0.4011,
+      "step": 6696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.448661272833432e-06,
+      "loss": 0.3946,
+      "step": 6697
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.446993076987238e-06,
+      "loss": 0.3924,
+      "step": 6698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.445324881141046e-06,
+      "loss": 0.4137,
+      "step": 6699
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.443656685294854e-06,
+      "loss": 0.4047,
+      "step": 6700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.441988489448662e-06,
+      "loss": 0.4022,
+      "step": 6701
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.44032029360247e-06,
+      "loss": 0.4151,
+      "step": 6702
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.438652097756278e-06,
+      "loss": 0.4075,
+      "step": 6703
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.436983901910086e-06,
+      "loss": 0.3838,
+      "step": 6704
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.435315706063892e-06,
+      "loss": 0.3844,
+      "step": 6705
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.4336475102177e-06,
+      "loss": 0.4045,
+      "step": 6706
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.431979314371508e-06,
+      "loss": 0.409,
+      "step": 6707
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.430311118525315e-06,
+      "loss": 0.4073,
+      "step": 6708
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.428642922679123e-06,
+      "loss": 0.4151,
+      "step": 6709
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.426974726832931e-06,
+      "loss": 0.3681,
+      "step": 6710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.425306530986739e-06,
+      "loss": 0.3842,
+      "step": 6711
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.423638335140545e-06,
+      "loss": 0.4179,
+      "step": 6712
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.421970139294355e-06,
+      "loss": 0.4071,
+      "step": 6713
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.420301943448161e-06,
+      "loss": 0.4326,
+      "step": 6714
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.418633747601969e-06,
+      "loss": 0.4167,
+      "step": 6715
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.416965551755777e-06,
+      "loss": 0.404,
+      "step": 6716
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.415297355909585e-06,
+      "loss": 0.3771,
+      "step": 6717
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.413629160063393e-06,
+      "loss": 0.4219,
+      "step": 6718
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.411960964217199e-06,
+      "loss": 0.3976,
+      "step": 6719
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.410292768371008e-06,
+      "loss": 0.4044,
+      "step": 6720
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.408624572524815e-06,
+      "loss": 0.4018,
+      "step": 6721
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.406956376678623e-06,
+      "loss": 0.4254,
+      "step": 6722
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.40528818083243e-06,
+      "loss": 0.408,
+      "step": 6723
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.403619984986238e-06,
+      "loss": 0.4111,
+      "step": 6724
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.401951789140046e-06,
+      "loss": 0.4018,
+      "step": 6725
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.400283593293852e-06,
+      "loss": 0.4197,
+      "step": 6726
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.398615397447662e-06,
+      "loss": 0.3837,
+      "step": 6727
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.396947201601468e-06,
+      "loss": 0.3982,
+      "step": 6728
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.395279005755276e-06,
+      "loss": 0.3839,
+      "step": 6729
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.393610809909084e-06,
+      "loss": 0.355,
+      "step": 6730
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.391942614062892e-06,
+      "loss": 0.3988,
+      "step": 6731
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.3902744182167e-06,
+      "loss": 0.4359,
+      "step": 6732
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.388606222370506e-06,
+      "loss": 0.4212,
+      "step": 6733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.386938026524315e-06,
+      "loss": 0.4159,
+      "step": 6734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.385269830678122e-06,
+      "loss": 0.3965,
+      "step": 6735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.38360163483193e-06,
+      "loss": 0.3883,
+      "step": 6736
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.381933438985737e-06,
+      "loss": 0.4307,
+      "step": 6737
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.380265243139545e-06,
+      "loss": 0.3843,
+      "step": 6738
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.378597047293353e-06,
+      "loss": 0.4321,
+      "step": 6739
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.37692885144716e-06,
+      "loss": 0.4038,
+      "step": 6740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.375260655600969e-06,
+      "loss": 0.3899,
+      "step": 6741
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.373592459754775e-06,
+      "loss": 0.3488,
+      "step": 6742
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.371924263908583e-06,
+      "loss": 0.4174,
+      "step": 6743
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.370256068062391e-06,
+      "loss": 0.4089,
+      "step": 6744
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.368587872216199e-06,
+      "loss": 0.4097,
+      "step": 6745
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.366919676370007e-06,
+      "loss": 0.3964,
+      "step": 6746
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.365251480523813e-06,
+      "loss": 0.4203,
+      "step": 6747
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.363583284677622e-06,
+      "loss": 0.4205,
+      "step": 6748
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.361915088831429e-06,
+      "loss": 0.3944,
+      "step": 6749
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.360246892985238e-06,
+      "loss": 0.4101,
+      "step": 6750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.358578697139044e-06,
+      "loss": 0.4052,
+      "step": 6751
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.356910501292852e-06,
+      "loss": 0.3962,
+      "step": 6752
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.35524230544666e-06,
+      "loss": 0.39,
+      "step": 6753
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.353574109600468e-06,
+      "loss": 0.3792,
+      "step": 6754
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.351905913754276e-06,
+      "loss": 0.3884,
+      "step": 6755
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.350237717908082e-06,
+      "loss": 0.38,
+      "step": 6756
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.348569522061892e-06,
+      "loss": 0.3965,
+      "step": 6757
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.346901326215698e-06,
+      "loss": 0.4047,
+      "step": 6758
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.345233130369506e-06,
+      "loss": 0.4035,
+      "step": 6759
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.343564934523314e-06,
+      "loss": 0.3993,
+      "step": 6760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.341896738677122e-06,
+      "loss": 0.386,
+      "step": 6761
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.34022854283093e-06,
+      "loss": 0.3795,
+      "step": 6762
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.338560346984736e-06,
+      "loss": 0.3974,
+      "step": 6763
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.336892151138545e-06,
+      "loss": 0.38,
+      "step": 6764
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.335223955292351e-06,
+      "loss": 0.4083,
+      "step": 6765
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.33355575944616e-06,
+      "loss": 0.4081,
+      "step": 6766
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.331887563599967e-06,
+      "loss": 0.3999,
+      "step": 6767
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.330219367753775e-06,
+      "loss": 0.4076,
+      "step": 6768
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.328551171907583e-06,
+      "loss": 0.3534,
+      "step": 6769
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.32688297606139e-06,
+      "loss": 0.4111,
+      "step": 6770
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.325214780215199e-06,
+      "loss": 0.3779,
+      "step": 6771
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.323546584369005e-06,
+      "loss": 0.3471,
+      "step": 6772
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.321878388522813e-06,
+      "loss": 0.3799,
+      "step": 6773
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.32021019267662e-06,
+      "loss": 0.3976,
+      "step": 6774
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.318541996830429e-06,
+      "loss": 0.409,
+      "step": 6775
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.316873800984237e-06,
+      "loss": 0.379,
+      "step": 6776
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.315205605138043e-06,
+      "loss": 0.3903,
+      "step": 6777
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.313537409291852e-06,
+      "loss": 0.3763,
+      "step": 6778
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.311869213445659e-06,
+      "loss": 0.3999,
+      "step": 6779
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.310201017599468e-06,
+      "loss": 0.3844,
+      "step": 6780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.308532821753274e-06,
+      "loss": 0.3986,
+      "step": 6781
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.306864625907082e-06,
+      "loss": 0.4136,
+      "step": 6782
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.30519643006089e-06,
+      "loss": 0.4241,
+      "step": 6783
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.303528234214696e-06,
+      "loss": 0.4372,
+      "step": 6784
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.301860038368506e-06,
+      "loss": 0.3919,
+      "step": 6785
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.300191842522312e-06,
+      "loss": 0.4128,
+      "step": 6786
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.298523646676122e-06,
+      "loss": 0.3867,
+      "step": 6787
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.296855450829928e-06,
+      "loss": 0.4188,
+      "step": 6788
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.295187254983736e-06,
+      "loss": 0.4218,
+      "step": 6789
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.293519059137544e-06,
+      "loss": 0.3799,
+      "step": 6790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.291850863291351e-06,
+      "loss": 0.4054,
+      "step": 6791
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.29018266744516e-06,
+      "loss": 0.3856,
+      "step": 6792
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.288514471598966e-06,
+      "loss": 0.4079,
+      "step": 6793
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.286846275752775e-06,
+      "loss": 0.384,
+      "step": 6794
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.285178079906581e-06,
+      "loss": 0.4046,
+      "step": 6795
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.28350988406039e-06,
+      "loss": 0.3867,
+      "step": 6796
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.281841688214197e-06,
+      "loss": 0.4202,
+      "step": 6797
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.280173492368005e-06,
+      "loss": 0.3847,
+      "step": 6798
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.278505296521813e-06,
+      "loss": 0.4135,
+      "step": 6799
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.276837100675619e-06,
+      "loss": 0.4123,
+      "step": 6800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.275168904829429e-06,
+      "loss": 0.4174,
+      "step": 6801
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.273500708983235e-06,
+      "loss": 0.4106,
+      "step": 6802
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.271832513137043e-06,
+      "loss": 0.3689,
+      "step": 6803
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.27016431729085e-06,
+      "loss": 0.4119,
+      "step": 6804
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.268496121444658e-06,
+      "loss": 0.4022,
+      "step": 6805
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.266827925598466e-06,
+      "loss": 0.4091,
+      "step": 6806
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.265159729752273e-06,
+      "loss": 0.4031,
+      "step": 6807
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.263491533906082e-06,
+      "loss": 0.4028,
+      "step": 6808
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.261823338059888e-06,
+      "loss": 0.3882,
+      "step": 6809
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.260155142213696e-06,
+      "loss": 0.3825,
+      "step": 6810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.258486946367504e-06,
+      "loss": 0.3914,
+      "step": 6811
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.256818750521312e-06,
+      "loss": 0.4025,
+      "step": 6812
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.25515055467512e-06,
+      "loss": 0.3726,
+      "step": 6813
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.253482358828926e-06,
+      "loss": 0.3874,
+      "step": 6814
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.251814162982736e-06,
+      "loss": 0.3968,
+      "step": 6815
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.250145967136542e-06,
+      "loss": 0.4001,
+      "step": 6816
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.248477771290351e-06,
+      "loss": 0.4066,
+      "step": 6817
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.246809575444158e-06,
+      "loss": 0.3871,
+      "step": 6818
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.245141379597966e-06,
+      "loss": 0.3923,
+      "step": 6819
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.243473183751773e-06,
+      "loss": 0.3974,
+      "step": 6820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.24180498790558e-06,
+      "loss": 0.3731,
+      "step": 6821
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.24013679205939e-06,
+      "loss": 0.4142,
+      "step": 6822
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.238468596213195e-06,
+      "loss": 0.3744,
+      "step": 6823
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.236800400367005e-06,
+      "loss": 0.3938,
+      "step": 6824
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.235132204520811e-06,
+      "loss": 0.4142,
+      "step": 6825
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.233464008674619e-06,
+      "loss": 0.3824,
+      "step": 6826
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.231795812828427e-06,
+      "loss": 0.3855,
+      "step": 6827
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.230127616982235e-06,
+      "loss": 0.4128,
+      "step": 6828
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.228459421136043e-06,
+      "loss": 0.3713,
+      "step": 6829
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.226791225289849e-06,
+      "loss": 0.401,
+      "step": 6830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.225123029443658e-06,
+      "loss": 0.3918,
+      "step": 6831
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.223454833597465e-06,
+      "loss": 0.3833,
+      "step": 6832
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.221786637751273e-06,
+      "loss": 0.407,
+      "step": 6833
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.22011844190508e-06,
+      "loss": 0.4113,
+      "step": 6834
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.218450246058888e-06,
+      "loss": 0.3833,
+      "step": 6835
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.216782050212696e-06,
+      "loss": 0.3934,
+      "step": 6836
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.215113854366502e-06,
+      "loss": 0.4101,
+      "step": 6837
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.213445658520312e-06,
+      "loss": 0.403,
+      "step": 6838
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.211777462674118e-06,
+      "loss": 0.3901,
+      "step": 6839
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.210109266827926e-06,
+      "loss": 0.3636,
+      "step": 6840
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.208441070981734e-06,
+      "loss": 0.4343,
+      "step": 6841
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.206772875135542e-06,
+      "loss": 0.3885,
+      "step": 6842
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.20510467928935e-06,
+      "loss": 0.3906,
+      "step": 6843
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.203436483443156e-06,
+      "loss": 0.4262,
+      "step": 6844
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.201768287596966e-06,
+      "loss": 0.3884,
+      "step": 6845
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.200100091750772e-06,
+      "loss": 0.4027,
+      "step": 6846
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.19843189590458e-06,
+      "loss": 0.4144,
+      "step": 6847
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.196763700058387e-06,
+      "loss": 0.3923,
+      "step": 6848
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.195095504212195e-06,
+      "loss": 0.3936,
+      "step": 6849
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.193427308366003e-06,
+      "loss": 0.4278,
+      "step": 6850
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.19175911251981e-06,
+      "loss": 0.4053,
+      "step": 6851
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.190090916673619e-06,
+      "loss": 0.4303,
+      "step": 6852
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.188422720827425e-06,
+      "loss": 0.4146,
+      "step": 6853
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.186754524981235e-06,
+      "loss": 0.3993,
+      "step": 6854
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.185086329135041e-06,
+      "loss": 0.4232,
+      "step": 6855
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.183418133288849e-06,
+      "loss": 0.3766,
+      "step": 6856
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.181749937442657e-06,
+      "loss": 0.3936,
+      "step": 6857
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.180081741596465e-06,
+      "loss": 0.3924,
+      "step": 6858
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.178413545750273e-06,
+      "loss": 0.4271,
+      "step": 6859
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.176745349904079e-06,
+      "loss": 0.424,
+      "step": 6860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.175077154057888e-06,
+      "loss": 0.3749,
+      "step": 6861
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.173408958211695e-06,
+      "loss": 0.4008,
+      "step": 6862
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.171740762365502e-06,
+      "loss": 0.4298,
+      "step": 6863
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.17007256651931e-06,
+      "loss": 0.3854,
+      "step": 6864
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.168404370673118e-06,
+      "loss": 0.3854,
+      "step": 6865
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.166736174826926e-06,
+      "loss": 0.389,
+      "step": 6866
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.165067978980732e-06,
+      "loss": 0.3774,
+      "step": 6867
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.163399783134542e-06,
+      "loss": 0.3989,
+      "step": 6868
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.161731587288348e-06,
+      "loss": 0.413,
+      "step": 6869
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.160063391442156e-06,
+      "loss": 0.4056,
+      "step": 6870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.158395195595964e-06,
+      "loss": 0.4046,
+      "step": 6871
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.156726999749772e-06,
+      "loss": 0.4154,
+      "step": 6872
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.15505880390358e-06,
+      "loss": 0.402,
+      "step": 6873
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.153390608057386e-06,
+      "loss": 0.411,
+      "step": 6874
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.151722412211195e-06,
+      "loss": 0.384,
+      "step": 6875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.150054216365002e-06,
+      "loss": 0.3968,
+      "step": 6876
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.14838602051881e-06,
+      "loss": 0.3823,
+      "step": 6877
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.146717824672617e-06,
+      "loss": 0.3811,
+      "step": 6878
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.145049628826425e-06,
+      "loss": 0.3924,
+      "step": 6879
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.143381432980233e-06,
+      "loss": 0.3919,
+      "step": 6880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.14171323713404e-06,
+      "loss": 0.4148,
+      "step": 6881
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.140045041287849e-06,
+      "loss": 0.3593,
+      "step": 6882
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.138376845441655e-06,
+      "loss": 0.4109,
+      "step": 6883
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.136708649595463e-06,
+      "loss": 0.3996,
+      "step": 6884
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.13504045374927e-06,
+      "loss": 0.369,
+      "step": 6885
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.133372257903079e-06,
+      "loss": 0.4189,
+      "step": 6886
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.131704062056887e-06,
+      "loss": 0.4038,
+      "step": 6887
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.130035866210693e-06,
+      "loss": 0.3988,
+      "step": 6888
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.128367670364502e-06,
+      "loss": 0.3612,
+      "step": 6889
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.126699474518309e-06,
+      "loss": 0.4036,
+      "step": 6890
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.125031278672118e-06,
+      "loss": 0.3967,
+      "step": 6891
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.123363082825924e-06,
+      "loss": 0.4298,
+      "step": 6892
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.121694886979732e-06,
+      "loss": 0.396,
+      "step": 6893
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.12002669113354e-06,
+      "loss": 0.3812,
+      "step": 6894
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.118358495287348e-06,
+      "loss": 0.3975,
+      "step": 6895
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.116690299441156e-06,
+      "loss": 0.4109,
+      "step": 6896
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.115022103594962e-06,
+      "loss": 0.4098,
+      "step": 6897
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.113353907748772e-06,
+      "loss": 0.3755,
+      "step": 6898
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.111685711902578e-06,
+      "loss": 0.3759,
+      "step": 6899
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.110017516056386e-06,
+      "loss": 0.4027,
+      "step": 6900
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.108349320210194e-06,
+      "loss": 0.3893,
+      "step": 6901
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.106681124364002e-06,
+      "loss": 0.3903,
+      "step": 6902
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.10501292851781e-06,
+      "loss": 0.4133,
+      "step": 6903
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.103344732671616e-06,
+      "loss": 0.4044,
+      "step": 6904
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.101676536825425e-06,
+      "loss": 0.4114,
+      "step": 6905
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.100008340979231e-06,
+      "loss": 0.3772,
+      "step": 6906
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.09834014513304e-06,
+      "loss": 0.3811,
+      "step": 6907
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.096671949286847e-06,
+      "loss": 0.4107,
+      "step": 6908
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.095003753440655e-06,
+      "loss": 0.4113,
+      "step": 6909
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.093335557594463e-06,
+      "loss": 0.3831,
+      "step": 6910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.091667361748269e-06,
+      "loss": 0.424,
+      "step": 6911
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.089999165902079e-06,
+      "loss": 0.4016,
+      "step": 6912
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.088330970055885e-06,
+      "loss": 0.4343,
+      "step": 6913
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.086662774209693e-06,
+      "loss": 0.3676,
+      "step": 6914
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.0849945783635e-06,
+      "loss": 0.4061,
+      "step": 6915
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.083326382517309e-06,
+      "loss": 0.4102,
+      "step": 6916
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.081658186671116e-06,
+      "loss": 0.3931,
+      "step": 6917
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.079989990824923e-06,
+      "loss": 0.4018,
+      "step": 6918
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.078321794978732e-06,
+      "loss": 0.4133,
+      "step": 6919
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.076653599132538e-06,
+      "loss": 0.4377,
+      "step": 6920
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.074985403286346e-06,
+      "loss": 0.3616,
+      "step": 6921
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.073317207440154e-06,
+      "loss": 0.3967,
+      "step": 6922
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.071649011593962e-06,
+      "loss": 0.3762,
+      "step": 6923
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.06998081574777e-06,
+      "loss": 0.3983,
+      "step": 6924
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.068312619901576e-06,
+      "loss": 0.3856,
+      "step": 6925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.066644424055386e-06,
+      "loss": 0.4129,
+      "step": 6926
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.064976228209192e-06,
+      "loss": 0.3774,
+      "step": 6927
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.063308032363e-06,
+      "loss": 0.4091,
+      "step": 6928
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.061639836516808e-06,
+      "loss": 0.3865,
+      "step": 6929
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.059971640670616e-06,
+      "loss": 0.4259,
+      "step": 6930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.058303444824423e-06,
+      "loss": 0.402,
+      "step": 6931
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.056635248978231e-06,
+      "loss": 0.3943,
+      "step": 6932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.05496705313204e-06,
+      "loss": 0.3957,
+      "step": 6933
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.053298857285845e-06,
+      "loss": 0.3972,
+      "step": 6934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.051630661439655e-06,
+      "loss": 0.3706,
+      "step": 6935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.049962465593461e-06,
+      "loss": 0.4202,
+      "step": 6936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.048294269747269e-06,
+      "loss": 0.4238,
+      "step": 6937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.046626073901077e-06,
+      "loss": 0.4168,
+      "step": 6938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.044957878054885e-06,
+      "loss": 0.3789,
+      "step": 6939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.043289682208693e-06,
+      "loss": 0.367,
+      "step": 6940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.041621486362499e-06,
+      "loss": 0.3943,
+      "step": 6941
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.039953290516309e-06,
+      "loss": 0.3988,
+      "step": 6942
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.038285094670115e-06,
+      "loss": 0.4189,
+      "step": 6943
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.036616898823923e-06,
+      "loss": 0.418,
+      "step": 6944
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.03494870297773e-06,
+      "loss": 0.3523,
+      "step": 6945
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.033280507131538e-06,
+      "loss": 0.4012,
+      "step": 6946
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.031612311285346e-06,
+      "loss": 0.4123,
+      "step": 6947
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.029944115439152e-06,
+      "loss": 0.3767,
+      "step": 6948
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.028275919592962e-06,
+      "loss": 0.3935,
+      "step": 6949
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.026607723746768e-06,
+      "loss": 0.4168,
+      "step": 6950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.024939527900576e-06,
+      "loss": 0.4253,
+      "step": 6951
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.023271332054384e-06,
+      "loss": 0.4067,
+      "step": 6952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.021603136208192e-06,
+      "loss": 0.3803,
+      "step": 6953
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.019934940362e-06,
+      "loss": 0.3834,
+      "step": 6954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.018266744515806e-06,
+      "loss": 0.3994,
+      "step": 6955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.016598548669616e-06,
+      "loss": 0.3899,
+      "step": 6956
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.014930352823422e-06,
+      "loss": 0.4063,
+      "step": 6957
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.01326215697723e-06,
+      "loss": 0.4087,
+      "step": 6958
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.011593961131038e-06,
+      "loss": 0.4111,
+      "step": 6959
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.009925765284845e-06,
+      "loss": 0.4178,
+      "step": 6960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.008257569438653e-06,
+      "loss": 0.3726,
+      "step": 6961
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.00658937359246e-06,
+      "loss": 0.3779,
+      "step": 6962
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.004921177746269e-06,
+      "loss": 0.4074,
+      "step": 6963
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.003252981900075e-06,
+      "loss": 0.412,
+      "step": 6964
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.001584786053883e-06,
+      "loss": 0.3975,
+      "step": 6965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.999916590207691e-06,
+      "loss": 0.3789,
+      "step": 6966
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.998248394361499e-06,
+      "loss": 0.4007,
+      "step": 6967
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.996580198515307e-06,
+      "loss": 0.3994,
+      "step": 6968
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.994912002669115e-06,
+      "loss": 0.4,
+      "step": 6969
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.993243806822923e-06,
+      "loss": 0.4032,
+      "step": 6970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.991575610976729e-06,
+      "loss": 0.401,
+      "step": 6971
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.989907415130537e-06,
+      "loss": 0.3889,
+      "step": 6972
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.988239219284345e-06,
+      "loss": 0.4168,
+      "step": 6973
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.986571023438152e-06,
+      "loss": 0.3788,
+      "step": 6974
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.98490282759196e-06,
+      "loss": 0.4037,
+      "step": 6975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.983234631745768e-06,
+      "loss": 0.4213,
+      "step": 6976
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.981566435899576e-06,
+      "loss": 0.4292,
+      "step": 6977
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.979898240053382e-06,
+      "loss": 0.3897,
+      "step": 6978
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.97823004420719e-06,
+      "loss": 0.4142,
+      "step": 6979
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.976561848360998e-06,
+      "loss": 0.3872,
+      "step": 6980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.974893652514806e-06,
+      "loss": 0.38,
+      "step": 6981
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.973225456668614e-06,
+      "loss": 0.3924,
+      "step": 6982
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.971557260822422e-06,
+      "loss": 0.4141,
+      "step": 6983
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.96988906497623e-06,
+      "loss": 0.4,
+      "step": 6984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.968220869130036e-06,
+      "loss": 0.4199,
+      "step": 6985
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.966552673283844e-06,
+      "loss": 0.3769,
+      "step": 6986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.964884477437652e-06,
+      "loss": 0.3893,
+      "step": 6987
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.96321628159146e-06,
+      "loss": 0.3846,
+      "step": 6988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.961548085745267e-06,
+      "loss": 0.3901,
+      "step": 6989
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.959879889899075e-06,
+      "loss": 0.4005,
+      "step": 6990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.958211694052883e-06,
+      "loss": 0.3982,
+      "step": 6991
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.95654349820669e-06,
+      "loss": 0.4134,
+      "step": 6992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.954875302360499e-06,
+      "loss": 0.363,
+      "step": 6993
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.953207106514305e-06,
+      "loss": 0.3873,
+      "step": 6994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.951538910668113e-06,
+      "loss": 0.412,
+      "step": 6995
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.949870714821921e-06,
+      "loss": 0.406,
+      "step": 6996
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.948202518975729e-06,
+      "loss": 0.3854,
+      "step": 6997
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.946534323129537e-06,
+      "loss": 0.3796,
+      "step": 6998
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.944866127283343e-06,
+      "loss": 0.3912,
+      "step": 6999
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.943197931437152e-06,
+      "loss": 0.3943,
+      "step": 7000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.941529735590959e-06,
+      "loss": 0.4107,
+      "step": 7001
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.939861539744767e-06,
+      "loss": 0.3758,
+      "step": 7002
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.938193343898574e-06,
+      "loss": 0.4064,
+      "step": 7003
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.936525148052382e-06,
+      "loss": 0.4019,
+      "step": 7004
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.93485695220619e-06,
+      "loss": 0.3927,
+      "step": 7005
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.933188756359998e-06,
+      "loss": 0.389,
+      "step": 7006
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.931520560513806e-06,
+      "loss": 0.4321,
+      "step": 7007
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.929852364667612e-06,
+      "loss": 0.4412,
+      "step": 7008
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.92818416882142e-06,
+      "loss": 0.3869,
+      "step": 7009
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.926515972975228e-06,
+      "loss": 0.385,
+      "step": 7010
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.924847777129036e-06,
+      "loss": 0.3724,
+      "step": 7011
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.923179581282844e-06,
+      "loss": 0.3912,
+      "step": 7012
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.921511385436652e-06,
+      "loss": 0.3798,
+      "step": 7013
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.91984318959046e-06,
+      "loss": 0.3881,
+      "step": 7014
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.918174993744266e-06,
+      "loss": 0.3833,
+      "step": 7015
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.916506797898074e-06,
+      "loss": 0.4097,
+      "step": 7016
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.914838602051881e-06,
+      "loss": 0.4297,
+      "step": 7017
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.91317040620569e-06,
+      "loss": 0.4207,
+      "step": 7018
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.911502210359497e-06,
+      "loss": 0.3911,
+      "step": 7019
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.909834014513305e-06,
+      "loss": 0.4154,
+      "step": 7020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.908165818667113e-06,
+      "loss": 0.4073,
+      "step": 7021
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.90649762282092e-06,
+      "loss": 0.3941,
+      "step": 7022
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.904829426974727e-06,
+      "loss": 0.4019,
+      "step": 7023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.903161231128535e-06,
+      "loss": 0.3929,
+      "step": 7024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.901493035282343e-06,
+      "loss": 0.3669,
+      "step": 7025
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.89982483943615e-06,
+      "loss": 0.416,
+      "step": 7026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.898156643589959e-06,
+      "loss": 0.3849,
+      "step": 7027
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.896488447743766e-06,
+      "loss": 0.3767,
+      "step": 7028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.894820251897573e-06,
+      "loss": 0.4032,
+      "step": 7029
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.89315205605138e-06,
+      "loss": 0.39,
+      "step": 7030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.891483860205188e-06,
+      "loss": 0.4135,
+      "step": 7031
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.889815664358996e-06,
+      "loss": 0.386,
+      "step": 7032
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.888147468512804e-06,
+      "loss": 0.3869,
+      "step": 7033
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.886479272666612e-06,
+      "loss": 0.4199,
+      "step": 7034
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.88481107682042e-06,
+      "loss": 0.3833,
+      "step": 7035
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.883142880974228e-06,
+      "loss": 0.4014,
+      "step": 7036
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.881474685128034e-06,
+      "loss": 0.3952,
+      "step": 7037
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.879806489281842e-06,
+      "loss": 0.4061,
+      "step": 7038
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.87813829343565e-06,
+      "loss": 0.4366,
+      "step": 7039
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.876470097589458e-06,
+      "loss": 0.4173,
+      "step": 7040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.874801901743266e-06,
+      "loss": 0.4061,
+      "step": 7041
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.873133705897074e-06,
+      "loss": 0.3922,
+      "step": 7042
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.871465510050881e-06,
+      "loss": 0.4325,
+      "step": 7043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.869797314204688e-06,
+      "loss": 0.4106,
+      "step": 7044
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.868129118358495e-06,
+      "loss": 0.3664,
+      "step": 7045
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.866460922512303e-06,
+      "loss": 0.4083,
+      "step": 7046
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.864792726666111e-06,
+      "loss": 0.3883,
+      "step": 7047
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.863124530819919e-06,
+      "loss": 0.3696,
+      "step": 7048
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.861456334973727e-06,
+      "loss": 0.4026,
+      "step": 7049
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.859788139127535e-06,
+      "loss": 0.3701,
+      "step": 7050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.858119943281343e-06,
+      "loss": 0.381,
+      "step": 7051
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.856451747435149e-06,
+      "loss": 0.4237,
+      "step": 7052
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.854783551588957e-06,
+      "loss": 0.4156,
+      "step": 7053
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.853115355742765e-06,
+      "loss": 0.4343,
+      "step": 7054
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.851447159896573e-06,
+      "loss": 0.4117,
+      "step": 7055
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.84977896405038e-06,
+      "loss": 0.399,
+      "step": 7056
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.848110768204188e-06,
+      "loss": 0.3807,
+      "step": 7057
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.846442572357996e-06,
+      "loss": 0.3922,
+      "step": 7058
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.844774376511803e-06,
+      "loss": 0.3782,
+      "step": 7059
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.84310618066561e-06,
+      "loss": 0.3933,
+      "step": 7060
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.841437984819418e-06,
+      "loss": 0.3963,
+      "step": 7061
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.839769788973226e-06,
+      "loss": 0.3946,
+      "step": 7062
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.838101593127034e-06,
+      "loss": 0.3783,
+      "step": 7063
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.836433397280842e-06,
+      "loss": 0.3941,
+      "step": 7064
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.83476520143465e-06,
+      "loss": 0.3602,
+      "step": 7065
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.833097005588456e-06,
+      "loss": 0.402,
+      "step": 7066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.831428809742264e-06,
+      "loss": 0.42,
+      "step": 7067
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.829760613896072e-06,
+      "loss": 0.4218,
+      "step": 7068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.82809241804988e-06,
+      "loss": 0.3793,
+      "step": 7069
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.826424222203688e-06,
+      "loss": 0.3998,
+      "step": 7070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.824756026357495e-06,
+      "loss": 0.4092,
+      "step": 7071
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.823087830511303e-06,
+      "loss": 0.4178,
+      "step": 7072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.821419634665111e-06,
+      "loss": 0.397,
+      "step": 7073
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.819751438818917e-06,
+      "loss": 0.3895,
+      "step": 7074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.818083242972725e-06,
+      "loss": 0.4082,
+      "step": 7075
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.816415047126533e-06,
+      "loss": 0.4285,
+      "step": 7076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.814746851280341e-06,
+      "loss": 0.4001,
+      "step": 7077
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.813078655434149e-06,
+      "loss": 0.3971,
+      "step": 7078
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.811410459587957e-06,
+      "loss": 0.4062,
+      "step": 7079
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.809742263741765e-06,
+      "loss": 0.4039,
+      "step": 7080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.808074067895571e-06,
+      "loss": 0.4147,
+      "step": 7081
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.806405872049379e-06,
+      "loss": 0.3911,
+      "step": 7082
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.804737676203187e-06,
+      "loss": 0.3751,
+      "step": 7083
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.803069480356995e-06,
+      "loss": 0.3777,
+      "step": 7084
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.801401284510803e-06,
+      "loss": 0.4399,
+      "step": 7085
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.79973308866461e-06,
+      "loss": 0.4246,
+      "step": 7086
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.798064892818418e-06,
+      "loss": 0.4291,
+      "step": 7087
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.796396696972224e-06,
+      "loss": 0.4118,
+      "step": 7088
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.794728501126032e-06,
+      "loss": 0.3694,
+      "step": 7089
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.79306030527984e-06,
+      "loss": 0.413,
+      "step": 7090
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.791392109433648e-06,
+      "loss": 0.3717,
+      "step": 7091
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.789723913587456e-06,
+      "loss": 0.3838,
+      "step": 7092
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.788055717741264e-06,
+      "loss": 0.4074,
+      "step": 7093
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.786387521895072e-06,
+      "loss": 0.3929,
+      "step": 7094
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.784719326048878e-06,
+      "loss": 0.3727,
+      "step": 7095
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.783051130202686e-06,
+      "loss": 0.3838,
+      "step": 7096
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.781382934356494e-06,
+      "loss": 0.3616,
+      "step": 7097
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.779714738510302e-06,
+      "loss": 0.3895,
+      "step": 7098
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.77804654266411e-06,
+      "loss": 0.3677,
+      "step": 7099
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.776378346817917e-06,
+      "loss": 0.3696,
+      "step": 7100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.774710150971725e-06,
+      "loss": 0.402,
+      "step": 7101
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.773041955125531e-06,
+      "loss": 0.3979,
+      "step": 7102
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.77137375927934e-06,
+      "loss": 0.3728,
+      "step": 7103
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.769705563433147e-06,
+      "loss": 0.3655,
+      "step": 7104
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.768037367586955e-06,
+      "loss": 0.3884,
+      "step": 7105
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.766369171740763e-06,
+      "loss": 0.352,
+      "step": 7106
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.764700975894571e-06,
+      "loss": 0.3594,
+      "step": 7107
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.763032780048379e-06,
+      "loss": 0.4149,
+      "step": 7108
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.761364584202187e-06,
+      "loss": 0.4104,
+      "step": 7109
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.759696388355995e-06,
+      "loss": 0.3833,
+      "step": 7110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.7580281925098e-06,
+      "loss": 0.4128,
+      "step": 7111
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.756359996663609e-06,
+      "loss": 0.4045,
+      "step": 7112
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.754691800817417e-06,
+      "loss": 0.3886,
+      "step": 7113
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.753023604971224e-06,
+      "loss": 0.387,
+      "step": 7114
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.751355409125032e-06,
+      "loss": 0.4084,
+      "step": 7115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.74968721327884e-06,
+      "loss": 0.3927,
+      "step": 7116
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.748019017432648e-06,
+      "loss": 0.3766,
+      "step": 7117
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.746350821586454e-06,
+      "loss": 0.3986,
+      "step": 7118
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.744682625740262e-06,
+      "loss": 0.4077,
+      "step": 7119
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.74301442989407e-06,
+      "loss": 0.4089,
+      "step": 7120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.741346234047878e-06,
+      "loss": 0.4117,
+      "step": 7121
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.739678038201686e-06,
+      "loss": 0.4021,
+      "step": 7122
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.738009842355494e-06,
+      "loss": 0.4205,
+      "step": 7123
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.736341646509302e-06,
+      "loss": 0.3895,
+      "step": 7124
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.734673450663108e-06,
+      "loss": 0.3862,
+      "step": 7125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.733005254816916e-06,
+      "loss": 0.3854,
+      "step": 7126
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.731337058970724e-06,
+      "loss": 0.4133,
+      "step": 7127
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.729668863124531e-06,
+      "loss": 0.389,
+      "step": 7128
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.72800066727834e-06,
+      "loss": 0.4005,
+      "step": 7129
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.726332471432147e-06,
+      "loss": 0.3972,
+      "step": 7130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.724664275585955e-06,
+      "loss": 0.3915,
+      "step": 7131
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.722996079739761e-06,
+      "loss": 0.4296,
+      "step": 7132
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.72132788389357e-06,
+      "loss": 0.4044,
+      "step": 7133
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.719659688047377e-06,
+      "loss": 0.3974,
+      "step": 7134
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.717991492201185e-06,
+      "loss": 0.3977,
+      "step": 7135
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.716323296354993e-06,
+      "loss": 0.3963,
+      "step": 7136
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.7146551005088e-06,
+      "loss": 0.406,
+      "step": 7137
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.712986904662609e-06,
+      "loss": 0.3774,
+      "step": 7138
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.711318708816415e-06,
+      "loss": 0.3697,
+      "step": 7139
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.709650512970223e-06,
+      "loss": 0.401,
+      "step": 7140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.70798231712403e-06,
+      "loss": 0.4331,
+      "step": 7141
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.706314121277839e-06,
+      "loss": 0.4007,
+      "step": 7142
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.704645925431646e-06,
+      "loss": 0.3727,
+      "step": 7143
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.702977729585454e-06,
+      "loss": 0.3963,
+      "step": 7144
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.701309533739262e-06,
+      "loss": 0.3817,
+      "step": 7145
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.699641337893068e-06,
+      "loss": 0.3697,
+      "step": 7146
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.697973142046878e-06,
+      "loss": 0.3994,
+      "step": 7147
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.696304946200684e-06,
+      "loss": 0.4476,
+      "step": 7148
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.694636750354492e-06,
+      "loss": 0.4077,
+      "step": 7149
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.6929685545083e-06,
+      "loss": 0.3889,
+      "step": 7150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.691300358662108e-06,
+      "loss": 0.3899,
+      "step": 7151
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.689632162815916e-06,
+      "loss": 0.3933,
+      "step": 7152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.687963966969722e-06,
+      "loss": 0.3955,
+      "step": 7153
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.686295771123531e-06,
+      "loss": 0.4411,
+      "step": 7154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.684627575277338e-06,
+      "loss": 0.39,
+      "step": 7155
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.682959379431146e-06,
+      "loss": 0.3982,
+      "step": 7156
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.681291183584953e-06,
+      "loss": 0.3737,
+      "step": 7157
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.679622987738761e-06,
+      "loss": 0.4043,
+      "step": 7158
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.67795479189257e-06,
+      "loss": 0.4085,
+      "step": 7159
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.676286596046375e-06,
+      "loss": 0.415,
+      "step": 7160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.674618400200185e-06,
+      "loss": 0.4142,
+      "step": 7161
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.672950204353991e-06,
+      "loss": 0.3778,
+      "step": 7162
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.671282008507799e-06,
+      "loss": 0.3841,
+      "step": 7163
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.669613812661607e-06,
+      "loss": 0.3956,
+      "step": 7164
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.667945616815415e-06,
+      "loss": 0.3962,
+      "step": 7165
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.666277420969223e-06,
+      "loss": 0.3933,
+      "step": 7166
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.664609225123029e-06,
+      "loss": 0.4243,
+      "step": 7167
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.662941029276838e-06,
+      "loss": 0.3791,
+      "step": 7168
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.661272833430645e-06,
+      "loss": 0.4049,
+      "step": 7169
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.659604637584453e-06,
+      "loss": 0.385,
+      "step": 7170
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.65793644173826e-06,
+      "loss": 0.3675,
+      "step": 7171
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.656268245892068e-06,
+      "loss": 0.3886,
+      "step": 7172
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.654600050045876e-06,
+      "loss": 0.3719,
+      "step": 7173
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.652931854199684e-06,
+      "loss": 0.385,
+      "step": 7174
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.651263658353492e-06,
+      "loss": 0.4126,
+      "step": 7175
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.649595462507298e-06,
+      "loss": 0.4231,
+      "step": 7176
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.647927266661106e-06,
+      "loss": 0.4084,
+      "step": 7177
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.646259070814914e-06,
+      "loss": 0.4286,
+      "step": 7178
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.644590874968722e-06,
+      "loss": 0.4197,
+      "step": 7179
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.64292267912253e-06,
+      "loss": 0.3918,
+      "step": 7180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.641254483276338e-06,
+      "loss": 0.3812,
+      "step": 7181
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.639586287430146e-06,
+      "loss": 0.4125,
+      "step": 7182
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.637918091583952e-06,
+      "loss": 0.4102,
+      "step": 7183
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.636249895737761e-06,
+      "loss": 0.4359,
+      "step": 7184
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.634581699891567e-06,
+      "loss": 0.3915,
+      "step": 7185
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.632913504045375e-06,
+      "loss": 0.4006,
+      "step": 7186
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.631245308199183e-06,
+      "loss": 0.3927,
+      "step": 7187
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.629577112352991e-06,
+      "loss": 0.3935,
+      "step": 7188
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.627908916506799e-06,
+      "loss": 0.377,
+      "step": 7189
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.626240720660605e-06,
+      "loss": 0.3969,
+      "step": 7190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.624572524814415e-06,
+      "loss": 0.3916,
+      "step": 7191
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.622904328968221e-06,
+      "loss": 0.3799,
+      "step": 7192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.621236133122029e-06,
+      "loss": 0.4108,
+      "step": 7193
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.619567937275837e-06,
+      "loss": 0.3905,
+      "step": 7194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.617899741429645e-06,
+      "loss": 0.4263,
+      "step": 7195
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.616231545583453e-06,
+      "loss": 0.3867,
+      "step": 7196
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.614563349737259e-06,
+      "loss": 0.3986,
+      "step": 7197
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.612895153891068e-06,
+      "loss": 0.411,
+      "step": 7198
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.611226958044875e-06,
+      "loss": 0.3843,
+      "step": 7199
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.609558762198682e-06,
+      "loss": 0.3865,
+      "step": 7200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.60789056635249e-06,
+      "loss": 0.4089,
+      "step": 7201
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.606222370506298e-06,
+      "loss": 0.3776,
+      "step": 7202
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.604554174660106e-06,
+      "loss": 0.3871,
+      "step": 7203
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.602885978813912e-06,
+      "loss": 0.4016,
+      "step": 7204
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.601217782967722e-06,
+      "loss": 0.3988,
+      "step": 7205
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.599549587121528e-06,
+      "loss": 0.3804,
+      "step": 7206
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.597881391275336e-06,
+      "loss": 0.3691,
+      "step": 7207
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.596213195429144e-06,
+      "loss": 0.3888,
+      "step": 7208
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.594544999582952e-06,
+      "loss": 0.4028,
+      "step": 7209
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.59287680373676e-06,
+      "loss": 0.3646,
+      "step": 7210
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.591208607890566e-06,
+      "loss": 0.3915,
+      "step": 7211
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.589540412044375e-06,
+      "loss": 0.3771,
+      "step": 7212
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.587872216198182e-06,
+      "loss": 0.3591,
+      "step": 7213
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.586204020351991e-06,
+      "loss": 0.3636,
+      "step": 7214
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.584535824505797e-06,
+      "loss": 0.391,
+      "step": 7215
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.582867628659605e-06,
+      "loss": 0.33,
+      "step": 7216
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.581199432813413e-06,
+      "loss": 0.3987,
+      "step": 7217
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.57953123696722e-06,
+      "loss": 0.3692,
+      "step": 7218
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.577863041121029e-06,
+      "loss": 0.3893,
+      "step": 7219
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.576194845274835e-06,
+      "loss": 0.4097,
+      "step": 7220
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.574526649428645e-06,
+      "loss": 0.4257,
+      "step": 7221
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.57285845358245e-06,
+      "loss": 0.3698,
+      "step": 7222
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.571190257736259e-06,
+      "loss": 0.3981,
+      "step": 7223
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.569522061890067e-06,
+      "loss": 0.3693,
+      "step": 7224
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.567853866043875e-06,
+      "loss": 0.4124,
+      "step": 7225
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.566185670197682e-06,
+      "loss": 0.3852,
+      "step": 7226
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.564517474351489e-06,
+      "loss": 0.4274,
+      "step": 7227
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.562849278505298e-06,
+      "loss": 0.3956,
+      "step": 7228
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.561181082659104e-06,
+      "loss": 0.4059,
+      "step": 7229
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.559512886812912e-06,
+      "loss": 0.393,
+      "step": 7230
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.55784469096672e-06,
+      "loss": 0.4091,
+      "step": 7231
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.556176495120528e-06,
+      "loss": 0.375,
+      "step": 7232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.554508299274336e-06,
+      "loss": 0.4094,
+      "step": 7233
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.552840103428142e-06,
+      "loss": 0.3821,
+      "step": 7234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.551171907581952e-06,
+      "loss": 0.3994,
+      "step": 7235
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.549503711735758e-06,
+      "loss": 0.4087,
+      "step": 7236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.547835515889566e-06,
+      "loss": 0.3897,
+      "step": 7237
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.546167320043374e-06,
+      "loss": 0.4063,
+      "step": 7238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.544499124197182e-06,
+      "loss": 0.4088,
+      "step": 7239
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.54283092835099e-06,
+      "loss": 0.3972,
+      "step": 7240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.541162732504796e-06,
+      "loss": 0.3896,
+      "step": 7241
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.539494536658605e-06,
+      "loss": 0.4164,
+      "step": 7242
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.537826340812411e-06,
+      "loss": 0.395,
+      "step": 7243
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.53615814496622e-06,
+      "loss": 0.4102,
+      "step": 7244
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.534489949120027e-06,
+      "loss": 0.4076,
+      "step": 7245
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.532821753273835e-06,
+      "loss": 0.4276,
+      "step": 7246
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.531153557427643e-06,
+      "loss": 0.3896,
+      "step": 7247
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.529485361581449e-06,
+      "loss": 0.3874,
+      "step": 7248
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.527817165735259e-06,
+      "loss": 0.3852,
+      "step": 7249
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.526148969889065e-06,
+      "loss": 0.4166,
+      "step": 7250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.524480774042874e-06,
+      "loss": 0.3951,
+      "step": 7251
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.52281257819668e-06,
+      "loss": 0.3852,
+      "step": 7252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.521144382350489e-06,
+      "loss": 0.4126,
+      "step": 7253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.519476186504296e-06,
+      "loss": 0.3673,
+      "step": 7254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.517807990658103e-06,
+      "loss": 0.3795,
+      "step": 7255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.516139794811912e-06,
+      "loss": 0.4062,
+      "step": 7256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.514471598965718e-06,
+      "loss": 0.4212,
+      "step": 7257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.512803403119528e-06,
+      "loss": 0.4029,
+      "step": 7258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.511135207273334e-06,
+      "loss": 0.4166,
+      "step": 7259
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.509467011427142e-06,
+      "loss": 0.3654,
+      "step": 7260
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.50779881558095e-06,
+      "loss": 0.3978,
+      "step": 7261
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.506130619734758e-06,
+      "loss": 0.361,
+      "step": 7262
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.504462423888566e-06,
+      "loss": 0.4146,
+      "step": 7263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.502794228042372e-06,
+      "loss": 0.4138,
+      "step": 7264
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.501126032196182e-06,
+      "loss": 0.426,
+      "step": 7265
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.499457836349988e-06,
+      "loss": 0.3837,
+      "step": 7266
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.497789640503796e-06,
+      "loss": 0.4115,
+      "step": 7267
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.496121444657603e-06,
+      "loss": 0.4195,
+      "step": 7268
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.494453248811411e-06,
+      "loss": 0.3852,
+      "step": 7269
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.49278505296522e-06,
+      "loss": 0.4005,
+      "step": 7270
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.491116857119025e-06,
+      "loss": 0.3837,
+      "step": 7271
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.489448661272835e-06,
+      "loss": 0.3932,
+      "step": 7272
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.487780465426641e-06,
+      "loss": 0.4121,
+      "step": 7273
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.486112269580449e-06,
+      "loss": 0.3685,
+      "step": 7274
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.484444073734257e-06,
+      "loss": 0.3846,
+      "step": 7275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.482775877888065e-06,
+      "loss": 0.4118,
+      "step": 7276
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.481107682041873e-06,
+      "loss": 0.3978,
+      "step": 7277
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.479439486195679e-06,
+      "loss": 0.3734,
+      "step": 7278
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.477771290349489e-06,
+      "loss": 0.4041,
+      "step": 7279
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.476103094503295e-06,
+      "loss": 0.3984,
+      "step": 7280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.474434898657103e-06,
+      "loss": 0.4173,
+      "step": 7281
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.47276670281091e-06,
+      "loss": 0.4013,
+      "step": 7282
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.471098506964718e-06,
+      "loss": 0.4199,
+      "step": 7283
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.469430311118526e-06,
+      "loss": 0.3948,
+      "step": 7284
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.467762115272332e-06,
+      "loss": 0.3784,
+      "step": 7285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.466093919426142e-06,
+      "loss": 0.3945,
+      "step": 7286
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.464425723579948e-06,
+      "loss": 0.3849,
+      "step": 7287
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.462757527733758e-06,
+      "loss": 0.391,
+      "step": 7288
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.461089331887564e-06,
+      "loss": 0.3945,
+      "step": 7289
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.459421136041372e-06,
+      "loss": 0.4094,
+      "step": 7290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.45775294019518e-06,
+      "loss": 0.4191,
+      "step": 7291
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.456084744348988e-06,
+      "loss": 0.4355,
+      "step": 7292
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.454416548502796e-06,
+      "loss": 0.4272,
+      "step": 7293
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.452748352656602e-06,
+      "loss": 0.3652,
+      "step": 7294
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.451080156810411e-06,
+      "loss": 0.3759,
+      "step": 7295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.449411960964218e-06,
+      "loss": 0.4178,
+      "step": 7296
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.447743765118025e-06,
+      "loss": 0.4011,
+      "step": 7297
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.446075569271833e-06,
+      "loss": 0.4025,
+      "step": 7298
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.444407373425641e-06,
+      "loss": 0.3988,
+      "step": 7299
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.442739177579449e-06,
+      "loss": 0.4124,
+      "step": 7300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.441070981733255e-06,
+      "loss": 0.4339,
+      "step": 7301
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.439402785887065e-06,
+      "loss": 0.4165,
+      "step": 7302
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.437734590040871e-06,
+      "loss": 0.3972,
+      "step": 7303
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.436066394194679e-06,
+      "loss": 0.3856,
+      "step": 7304
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.434398198348487e-06,
+      "loss": 0.3977,
+      "step": 7305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.432730002502295e-06,
+      "loss": 0.3937,
+      "step": 7306
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.431061806656103e-06,
+      "loss": 0.4001,
+      "step": 7307
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.429393610809909e-06,
+      "loss": 0.3839,
+      "step": 7308
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.427725414963718e-06,
+      "loss": 0.3855,
+      "step": 7309
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.426057219117525e-06,
+      "loss": 0.4111,
+      "step": 7310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.424389023271332e-06,
+      "loss": 0.4114,
+      "step": 7311
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.42272082742514e-06,
+      "loss": 0.4232,
+      "step": 7312
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 0.3944,
+      "step": 7313
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.419384435732756e-06,
+      "loss": 0.4241,
+      "step": 7314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.417716239886562e-06,
+      "loss": 0.4432,
+      "step": 7315
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.416048044040372e-06,
+      "loss": 0.3816,
+      "step": 7316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.414379848194178e-06,
+      "loss": 0.3891,
+      "step": 7317
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.412711652347986e-06,
+      "loss": 0.3918,
+      "step": 7318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.411043456501794e-06,
+      "loss": 0.3708,
+      "step": 7319
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.409375260655602e-06,
+      "loss": 0.3844,
+      "step": 7320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.40770706480941e-06,
+      "loss": 0.406,
+      "step": 7321
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.406038868963216e-06,
+      "loss": 0.3814,
+      "step": 7322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.404370673117025e-06,
+      "loss": 0.388,
+      "step": 7323
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.402702477270832e-06,
+      "loss": 0.3773,
+      "step": 7324
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.401034281424641e-06,
+      "loss": 0.4241,
+      "step": 7325
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.399366085578447e-06,
+      "loss": 0.3729,
+      "step": 7326
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.397697889732255e-06,
+      "loss": 0.4251,
+      "step": 7327
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.396029693886063e-06,
+      "loss": 0.3792,
+      "step": 7328
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.394361498039871e-06,
+      "loss": 0.4038,
+      "step": 7329
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.392693302193679e-06,
+      "loss": 0.3955,
+      "step": 7330
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.391025106347485e-06,
+      "loss": 0.4269,
+      "step": 7331
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.389356910501295e-06,
+      "loss": 0.4289,
+      "step": 7332
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.387688714655101e-06,
+      "loss": 0.3825,
+      "step": 7333
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.386020518808909e-06,
+      "loss": 0.4126,
+      "step": 7334
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.384352322962717e-06,
+      "loss": 0.3928,
+      "step": 7335
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.382684127116525e-06,
+      "loss": 0.3874,
+      "step": 7336
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.381015931270332e-06,
+      "loss": 0.3478,
+      "step": 7337
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.379347735424139e-06,
+      "loss": 0.3584,
+      "step": 7338
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.377679539577948e-06,
+      "loss": 0.4006,
+      "step": 7339
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.376011343731754e-06,
+      "loss": 0.4163,
+      "step": 7340
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.374343147885562e-06,
+      "loss": 0.4015,
+      "step": 7341
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.37267495203937e-06,
+      "loss": 0.4121,
+      "step": 7342
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.371006756193178e-06,
+      "loss": 0.4236,
+      "step": 7343
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.369338560346986e-06,
+      "loss": 0.3833,
+      "step": 7344
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.367670364500792e-06,
+      "loss": 0.3929,
+      "step": 7345
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.366002168654602e-06,
+      "loss": 0.3883,
+      "step": 7346
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.364333972808408e-06,
+      "loss": 0.3782,
+      "step": 7347
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.362665776962216e-06,
+      "loss": 0.4304,
+      "step": 7348
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.360997581116024e-06,
+      "loss": 0.3919,
+      "step": 7349
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.359329385269832e-06,
+      "loss": 0.4022,
+      "step": 7350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.35766118942364e-06,
+      "loss": 0.4352,
+      "step": 7351
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.355992993577446e-06,
+      "loss": 0.387,
+      "step": 7352
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.354324797731255e-06,
+      "loss": 0.3708,
+      "step": 7353
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.352656601885061e-06,
+      "loss": 0.3893,
+      "step": 7354
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.350988406038871e-06,
+      "loss": 0.387,
+      "step": 7355
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.349320210192677e-06,
+      "loss": 0.3827,
+      "step": 7356
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.347652014346485e-06,
+      "loss": 0.3647,
+      "step": 7357
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.345983818500293e-06,
+      "loss": 0.4245,
+      "step": 7358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.3443156226541e-06,
+      "loss": 0.3883,
+      "step": 7359
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.342647426807909e-06,
+      "loss": 0.3852,
+      "step": 7360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.340979230961715e-06,
+      "loss": 0.4244,
+      "step": 7361
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.339311035115525e-06,
+      "loss": 0.4302,
+      "step": 7362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.33764283926933e-06,
+      "loss": 0.3619,
+      "step": 7363
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.335974643423139e-06,
+      "loss": 0.4136,
+      "step": 7364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.334306447576947e-06,
+      "loss": 0.3961,
+      "step": 7365
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.332638251730754e-06,
+      "loss": 0.4178,
+      "step": 7366
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.330970055884562e-06,
+      "loss": 0.3875,
+      "step": 7367
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.329301860038368e-06,
+      "loss": 0.3979,
+      "step": 7368
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.327633664192178e-06,
+      "loss": 0.3921,
+      "step": 7369
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.325965468345984e-06,
+      "loss": 0.3716,
+      "step": 7370
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.324297272499792e-06,
+      "loss": 0.3686,
+      "step": 7371
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.3226290766536e-06,
+      "loss": 0.3903,
+      "step": 7372
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.320960880807408e-06,
+      "loss": 0.3865,
+      "step": 7373
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.319292684961216e-06,
+      "loss": 0.3743,
+      "step": 7374
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.317624489115022e-06,
+      "loss": 0.396,
+      "step": 7375
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.315956293268832e-06,
+      "loss": 0.4319,
+      "step": 7376
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.314288097422638e-06,
+      "loss": 0.4015,
+      "step": 7377
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.312619901576446e-06,
+      "loss": 0.403,
+      "step": 7378
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.310951705730254e-06,
+      "loss": 0.3816,
+      "step": 7379
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.309283509884061e-06,
+      "loss": 0.3891,
+      "step": 7380
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.30761531403787e-06,
+      "loss": 0.3497,
+      "step": 7381
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.305947118191676e-06,
+      "loss": 0.3906,
+      "step": 7382
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.304278922345485e-06,
+      "loss": 0.4215,
+      "step": 7383
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.302610726499291e-06,
+      "loss": 0.3826,
+      "step": 7384
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.3009425306531e-06,
+      "loss": 0.3528,
+      "step": 7385
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.299274334806907e-06,
+      "loss": 0.4181,
+      "step": 7386
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.297606138960715e-06,
+      "loss": 0.3875,
+      "step": 7387
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.295937943114523e-06,
+      "loss": 0.4447,
+      "step": 7388
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.294269747268329e-06,
+      "loss": 0.4351,
+      "step": 7389
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.292601551422139e-06,
+      "loss": 0.4413,
+      "step": 7390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.290933355575945e-06,
+      "loss": 0.3633,
+      "step": 7391
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.289265159729753e-06,
+      "loss": 0.3888,
+      "step": 7392
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.28759696388356e-06,
+      "loss": 0.3757,
+      "step": 7393
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.285928768037368e-06,
+      "loss": 0.3736,
+      "step": 7394
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.284260572191176e-06,
+      "loss": 0.3913,
+      "step": 7395
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.282592376344983e-06,
+      "loss": 0.4051,
+      "step": 7396
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.280924180498792e-06,
+      "loss": 0.38,
+      "step": 7397
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.279255984652598e-06,
+      "loss": 0.4041,
+      "step": 7398
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.277587788806406e-06,
+      "loss": 0.4037,
+      "step": 7399
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.275919592960214e-06,
+      "loss": 0.4215,
+      "step": 7400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.274251397114022e-06,
+      "loss": 0.4244,
+      "step": 7401
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.27258320126783e-06,
+      "loss": 0.3826,
+      "step": 7402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.270915005421638e-06,
+      "loss": 0.3879,
+      "step": 7403
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.269246809575446e-06,
+      "loss": 0.3834,
+      "step": 7404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.267578613729252e-06,
+      "loss": 0.3976,
+      "step": 7405
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.265910417883061e-06,
+      "loss": 0.4091,
+      "step": 7406
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.264242222036868e-06,
+      "loss": 0.4124,
+      "step": 7407
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.262574026190675e-06,
+      "loss": 0.3822,
+      "step": 7408
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.260905830344483e-06,
+      "loss": 0.3956,
+      "step": 7409
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.259237634498291e-06,
+      "loss": 0.4385,
+      "step": 7410
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.257569438652099e-06,
+      "loss": 0.3922,
+      "step": 7411
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.255901242805905e-06,
+      "loss": 0.4014,
+      "step": 7412
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.254233046959715e-06,
+      "loss": 0.4217,
+      "step": 7413
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.252564851113521e-06,
+      "loss": 0.4033,
+      "step": 7414
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.250896655267329e-06,
+      "loss": 0.4078,
+      "step": 7415
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.249228459421137e-06,
+      "loss": 0.3542,
+      "step": 7416
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.247560263574945e-06,
+      "loss": 0.3899,
+      "step": 7417
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.245892067728753e-06,
+      "loss": 0.3859,
+      "step": 7418
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.244223871882559e-06,
+      "loss": 0.3791,
+      "step": 7419
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.242555676036368e-06,
+      "loss": 0.3603,
+      "step": 7420
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.240887480190175e-06,
+      "loss": 0.4018,
+      "step": 7421
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.239219284343983e-06,
+      "loss": 0.4332,
+      "step": 7422
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.23755108849779e-06,
+      "loss": 0.3662,
+      "step": 7423
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.235882892651598e-06,
+      "loss": 0.4074,
+      "step": 7424
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.234214696805406e-06,
+      "loss": 0.3792,
+      "step": 7425
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.232546500959212e-06,
+      "loss": 0.4027,
+      "step": 7426
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.230878305113022e-06,
+      "loss": 0.4394,
+      "step": 7427
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.229210109266828e-06,
+      "loss": 0.3933,
+      "step": 7428
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.227541913420636e-06,
+      "loss": 0.3705,
+      "step": 7429
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.225873717574444e-06,
+      "loss": 0.3915,
+      "step": 7430
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.224205521728252e-06,
+      "loss": 0.4039,
+      "step": 7431
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.22253732588206e-06,
+      "loss": 0.4027,
+      "step": 7432
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.220869130035866e-06,
+      "loss": 0.413,
+      "step": 7433
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.219200934189675e-06,
+      "loss": 0.3949,
+      "step": 7434
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.217532738343482e-06,
+      "loss": 0.3953,
+      "step": 7435
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.21586454249729e-06,
+      "loss": 0.3817,
+      "step": 7436
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.214196346651097e-06,
+      "loss": 0.3965,
+      "step": 7437
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.212528150804905e-06,
+      "loss": 0.353,
+      "step": 7438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.210859954958713e-06,
+      "loss": 0.3967,
+      "step": 7439
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.209191759112521e-06,
+      "loss": 0.3826,
+      "step": 7440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.207523563266329e-06,
+      "loss": 0.4039,
+      "step": 7441
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.205855367420135e-06,
+      "loss": 0.3818,
+      "step": 7442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.204187171573943e-06,
+      "loss": 0.4335,
+      "step": 7443
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.202518975727751e-06,
+      "loss": 0.3831,
+      "step": 7444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.200850779881559e-06,
+      "loss": 0.3931,
+      "step": 7445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.199182584035367e-06,
+      "loss": 0.3965,
+      "step": 7446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.197514388189175e-06,
+      "loss": 0.3771,
+      "step": 7447
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.195846192342982e-06,
+      "loss": 0.3863,
+      "step": 7448
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.194177996496789e-06,
+      "loss": 0.3803,
+      "step": 7449
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.192509800650597e-06,
+      "loss": 0.3741,
+      "step": 7450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.190841604804404e-06,
+      "loss": 0.3685,
+      "step": 7451
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.189173408958212e-06,
+      "loss": 0.4152,
+      "step": 7452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.18750521311202e-06,
+      "loss": 0.4028,
+      "step": 7453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.185837017265828e-06,
+      "loss": 0.4058,
+      "step": 7454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.184168821419636e-06,
+      "loss": 0.3755,
+      "step": 7455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.182500625573442e-06,
+      "loss": 0.4263,
+      "step": 7456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.18083242972725e-06,
+      "loss": 0.4096,
+      "step": 7457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.179164233881058e-06,
+      "loss": 0.3879,
+      "step": 7458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.177496038034866e-06,
+      "loss": 0.4058,
+      "step": 7459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.175827842188674e-06,
+      "loss": 0.3695,
+      "step": 7460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.174159646342482e-06,
+      "loss": 0.375,
+      "step": 7461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.17249145049629e-06,
+      "loss": 0.372,
+      "step": 7462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.170823254650096e-06,
+      "loss": 0.4149,
+      "step": 7463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.169155058803904e-06,
+      "loss": 0.3871,
+      "step": 7464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.167486862957711e-06,
+      "loss": 0.3818,
+      "step": 7465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.16581866711152e-06,
+      "loss": 0.3921,
+      "step": 7466
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.164150471265327e-06,
+      "loss": 0.4429,
+      "step": 7467
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.162482275419135e-06,
+      "loss": 0.3625,
+      "step": 7468
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.160814079572943e-06,
+      "loss": 0.3982,
+      "step": 7469
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.159145883726751e-06,
+      "loss": 0.3986,
+      "step": 7470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.157477687880559e-06,
+      "loss": 0.4166,
+      "step": 7471
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.155809492034365e-06,
+      "loss": 0.3835,
+      "step": 7472
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.154141296188173e-06,
+      "loss": 0.3949,
+      "step": 7473
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.15247310034198e-06,
+      "loss": 0.3799,
+      "step": 7474
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.150804904495789e-06,
+      "loss": 0.4123,
+      "step": 7475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.149136708649597e-06,
+      "loss": 0.4214,
+      "step": 7476
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.147468512803404e-06,
+      "loss": 0.4089,
+      "step": 7477
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.145800316957212e-06,
+      "loss": 0.381,
+      "step": 7478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.144132121111019e-06,
+      "loss": 0.4161,
+      "step": 7479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.142463925264826e-06,
+      "loss": 0.3816,
+      "step": 7480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.140795729418634e-06,
+      "loss": 0.391,
+      "step": 7481
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.139127533572442e-06,
+      "loss": 0.405,
+      "step": 7482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.13745933772625e-06,
+      "loss": 0.4148,
+      "step": 7483
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.135791141880058e-06,
+      "loss": 0.3995,
+      "step": 7484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.134122946033866e-06,
+      "loss": 0.4191,
+      "step": 7485
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.132454750187672e-06,
+      "loss": 0.4091,
+      "step": 7486
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.13078655434148e-06,
+      "loss": 0.4349,
+      "step": 7487
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.129118358495288e-06,
+      "loss": 0.4162,
+      "step": 7488
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.127450162649096e-06,
+      "loss": 0.3849,
+      "step": 7489
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.125781966802904e-06,
+      "loss": 0.3918,
+      "step": 7490
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.124113770956711e-06,
+      "loss": 0.3697,
+      "step": 7491
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.12244557511052e-06,
+      "loss": 0.4002,
+      "step": 7492
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.120777379264326e-06,
+      "loss": 0.392,
+      "step": 7493
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.119109183418133e-06,
+      "loss": 0.3844,
+      "step": 7494
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.117440987571941e-06,
+      "loss": 0.3975,
+      "step": 7495
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.11577279172575e-06,
+      "loss": 0.3993,
+      "step": 7496
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.114104595879557e-06,
+      "loss": 0.4108,
+      "step": 7497
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.112436400033365e-06,
+      "loss": 0.4428,
+      "step": 7498
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.110768204187173e-06,
+      "loss": 0.4038,
+      "step": 7499
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.109100008340979e-06,
+      "loss": 0.4052,
+      "step": 7500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.107431812494787e-06,
+      "loss": 0.4155,
+      "step": 7501
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.105763616648595e-06,
+      "loss": 0.3843,
+      "step": 7502
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.104095420802403e-06,
+      "loss": 0.3898,
+      "step": 7503
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.10242722495621e-06,
+      "loss": 0.3911,
+      "step": 7504
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.100759029110019e-06,
+      "loss": 0.3897,
+      "step": 7505
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.099090833263826e-06,
+      "loss": 0.4045,
+      "step": 7506
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.097422637417634e-06,
+      "loss": 0.3998,
+      "step": 7507
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.09575444157144e-06,
+      "loss": 0.3851,
+      "step": 7508
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.094086245725248e-06,
+      "loss": 0.3874,
+      "step": 7509
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.092418049879056e-06,
+      "loss": 0.4115,
+      "step": 7510
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.090749854032864e-06,
+      "loss": 0.3935,
+      "step": 7511
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.089081658186672e-06,
+      "loss": 0.3984,
+      "step": 7512
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.08741346234048e-06,
+      "loss": 0.3921,
+      "step": 7513
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.085745266494288e-06,
+      "loss": 0.3797,
+      "step": 7514
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.084077070648094e-06,
+      "loss": 0.3758,
+      "step": 7515
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.082408874801902e-06,
+      "loss": 0.3657,
+      "step": 7516
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.08074067895571e-06,
+      "loss": 0.4047,
+      "step": 7517
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.079072483109518e-06,
+      "loss": 0.4291,
+      "step": 7518
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.077404287263326e-06,
+      "loss": 0.4186,
+      "step": 7519
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.075736091417133e-06,
+      "loss": 0.4062,
+      "step": 7520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.074067895570941e-06,
+      "loss": 0.3737,
+      "step": 7521
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.072399699724748e-06,
+      "loss": 0.4413,
+      "step": 7522
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.070731503878555e-06,
+      "loss": 0.3815,
+      "step": 7523
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.069063308032363e-06,
+      "loss": 0.4076,
+      "step": 7524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.067395112186171e-06,
+      "loss": 0.4015,
+      "step": 7525
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.065726916339979e-06,
+      "loss": 0.3849,
+      "step": 7526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.064058720493787e-06,
+      "loss": 0.365,
+      "step": 7527
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.062390524647595e-06,
+      "loss": 0.4036,
+      "step": 7528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.060722328801403e-06,
+      "loss": 0.42,
+      "step": 7529
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.059054132955209e-06,
+      "loss": 0.4292,
+      "step": 7530
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.057385937109017e-06,
+      "loss": 0.3821,
+      "step": 7531
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.055717741262825e-06,
+      "loss": 0.3966,
+      "step": 7532
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.054049545416633e-06,
+      "loss": 0.3944,
+      "step": 7533
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.05238134957044e-06,
+      "loss": 0.4033,
+      "step": 7534
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.050713153724248e-06,
+      "loss": 0.3915,
+      "step": 7535
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.049044957878056e-06,
+      "loss": 0.3893,
+      "step": 7536
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.047376762031862e-06,
+      "loss": 0.4079,
+      "step": 7537
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.04570856618567e-06,
+      "loss": 0.4329,
+      "step": 7538
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.044040370339478e-06,
+      "loss": 0.3957,
+      "step": 7539
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.042372174493286e-06,
+      "loss": 0.3893,
+      "step": 7540
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.040703978647094e-06,
+      "loss": 0.4099,
+      "step": 7541
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.039035782800902e-06,
+      "loss": 0.3957,
+      "step": 7542
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.03736758695471e-06,
+      "loss": 0.3826,
+      "step": 7543
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.035699391108518e-06,
+      "loss": 0.4256,
+      "step": 7544
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.034031195262324e-06,
+      "loss": 0.427,
+      "step": 7545
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.032362999416132e-06,
+      "loss": 0.3776,
+      "step": 7546
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.03069480356994e-06,
+      "loss": 0.3806,
+      "step": 7547
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.029026607723747e-06,
+      "loss": 0.4255,
+      "step": 7548
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.027358411877555e-06,
+      "loss": 0.4083,
+      "step": 7549
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.025690216031363e-06,
+      "loss": 0.3953,
+      "step": 7550
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.024022020185171e-06,
+      "loss": 0.3931,
+      "step": 7551
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.022353824338977e-06,
+      "loss": 0.4103,
+      "step": 7552
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.020685628492785e-06,
+      "loss": 0.4185,
+      "step": 7553
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.019017432646593e-06,
+      "loss": 0.3795,
+      "step": 7554
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.017349236800401e-06,
+      "loss": 0.4078,
+      "step": 7555
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.015681040954209e-06,
+      "loss": 0.4235,
+      "step": 7556
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.014012845108017e-06,
+      "loss": 0.3995,
+      "step": 7557
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.012344649261825e-06,
+      "loss": 0.44,
+      "step": 7558
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.010676453415631e-06,
+      "loss": 0.3879,
+      "step": 7559
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.009008257569439e-06,
+      "loss": 0.4112,
+      "step": 7560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 8.007340061723247e-06,
+      "loss": 0.3904,
+      "step": 7561
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.005671865877055e-06,
+      "loss": 0.3924,
+      "step": 7562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.004003670030862e-06,
+      "loss": 0.4021,
+      "step": 7563
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.00233547418467e-06,
+      "loss": 0.3564,
+      "step": 7564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 8.000667278338478e-06,
+      "loss": 0.4272,
+      "step": 7565
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.998999082492284e-06,
+      "loss": 0.4121,
+      "step": 7566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.997330886646092e-06,
+      "loss": 0.3806,
+      "step": 7567
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.9956626907999e-06,
+      "loss": 0.3863,
+      "step": 7568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.993994494953708e-06,
+      "loss": 0.4144,
+      "step": 7569
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.992326299107516e-06,
+      "loss": 0.3671,
+      "step": 7570
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.990658103261324e-06,
+      "loss": 0.3859,
+      "step": 7571
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.988989907415132e-06,
+      "loss": 0.3945,
+      "step": 7572
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.987321711568938e-06,
+      "loss": 0.4054,
+      "step": 7573
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.985653515722746e-06,
+      "loss": 0.3845,
+      "step": 7574
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.983985319876554e-06,
+      "loss": 0.4162,
+      "step": 7575
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.982317124030362e-06,
+      "loss": 0.4168,
+      "step": 7576
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.98064892818417e-06,
+      "loss": 0.3848,
+      "step": 7577
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.978980732337977e-06,
+      "loss": 0.3722,
+      "step": 7578
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.977312536491785e-06,
+      "loss": 0.3876,
+      "step": 7579
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.975644340645591e-06,
+      "loss": 0.3895,
+      "step": 7580
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.973976144799401e-06,
+      "loss": 0.3722,
+      "step": 7581
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.972307948953207e-06,
+      "loss": 0.4209,
+      "step": 7582
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.970639753107015e-06,
+      "loss": 0.3801,
+      "step": 7583
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.968971557260823e-06,
+      "loss": 0.3833,
+      "step": 7584
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.96730336141463e-06,
+      "loss": 0.3969,
+      "step": 7585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.965635165568439e-06,
+      "loss": 0.3869,
+      "step": 7586
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.963966969722247e-06,
+      "loss": 0.405,
+      "step": 7587
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.962298773876054e-06,
+      "loss": 0.4156,
+      "step": 7588
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.96063057802986e-06,
+      "loss": 0.4138,
+      "step": 7589
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.958962382183669e-06,
+      "loss": 0.387,
+      "step": 7590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.957294186337476e-06,
+      "loss": 0.3952,
+      "step": 7591
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.955625990491284e-06,
+      "loss": 0.4239,
+      "step": 7592
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.953957794645092e-06,
+      "loss": 0.4,
+      "step": 7593
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.9522895987989e-06,
+      "loss": 0.4083,
+      "step": 7594
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.950621402952708e-06,
+      "loss": 0.3838,
+      "step": 7595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.948953207106514e-06,
+      "loss": 0.3979,
+      "step": 7596
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.947285011260322e-06,
+      "loss": 0.375,
+      "step": 7597
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.94561681541413e-06,
+      "loss": 0.3847,
+      "step": 7598
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.943948619567938e-06,
+      "loss": 0.3839,
+      "step": 7599
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.942280423721746e-06,
+      "loss": 0.3894,
+      "step": 7600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.940612227875554e-06,
+      "loss": 0.3614,
+      "step": 7601
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.938944032029362e-06,
+      "loss": 0.4017,
+      "step": 7602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.937275836183168e-06,
+      "loss": 0.3807,
+      "step": 7603
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.935607640336976e-06,
+      "loss": 0.3652,
+      "step": 7604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.933939444490783e-06,
+      "loss": 0.3997,
+      "step": 7605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.932271248644591e-06,
+      "loss": 0.3779,
+      "step": 7606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.9306030527984e-06,
+      "loss": 0.4196,
+      "step": 7607
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.928934856952207e-06,
+      "loss": 0.3883,
+      "step": 7608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.927266661106015e-06,
+      "loss": 0.4483,
+      "step": 7609
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.925598465259821e-06,
+      "loss": 0.3837,
+      "step": 7610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.92393026941363e-06,
+      "loss": 0.3992,
+      "step": 7611
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.922262073567437e-06,
+      "loss": 0.3957,
+      "step": 7612
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.920593877721245e-06,
+      "loss": 0.409,
+      "step": 7613
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.918925681875053e-06,
+      "loss": 0.4232,
+      "step": 7614
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.91725748602886e-06,
+      "loss": 0.3852,
+      "step": 7615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.915589290182669e-06,
+      "loss": 0.3988,
+      "step": 7616
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.913921094336475e-06,
+      "loss": 0.3655,
+      "step": 7617
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.912252898490284e-06,
+      "loss": 0.3846,
+      "step": 7618
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.91058470264409e-06,
+      "loss": 0.4195,
+      "step": 7619
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.908916506797898e-06,
+      "loss": 0.426,
+      "step": 7620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.907248310951706e-06,
+      "loss": 0.3966,
+      "step": 7621
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.905580115105514e-06,
+      "loss": 0.4325,
+      "step": 7622
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.903911919259322e-06,
+      "loss": 0.4198,
+      "step": 7623
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.902243723413128e-06,
+      "loss": 0.3982,
+      "step": 7624
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.900575527566938e-06,
+      "loss": 0.3902,
+      "step": 7625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.898907331720744e-06,
+      "loss": 0.3926,
+      "step": 7626
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.897239135874552e-06,
+      "loss": 0.4026,
+      "step": 7627
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.89557094002836e-06,
+      "loss": 0.3909,
+      "step": 7628
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.893902744182168e-06,
+      "loss": 0.4216,
+      "step": 7629
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.892234548335976e-06,
+      "loss": 0.3859,
+      "step": 7630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.890566352489782e-06,
+      "loss": 0.3922,
+      "step": 7631
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.888898156643591e-06,
+      "loss": 0.3862,
+      "step": 7632
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.887229960797398e-06,
+      "loss": 0.3891,
+      "step": 7633
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.885561764951205e-06,
+      "loss": 0.3988,
+      "step": 7634
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.883893569105013e-06,
+      "loss": 0.3973,
+      "step": 7635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.882225373258821e-06,
+      "loss": 0.4024,
+      "step": 7636
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.880557177412629e-06,
+      "loss": 0.3822,
+      "step": 7637
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.878888981566435e-06,
+      "loss": 0.4313,
+      "step": 7638
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.877220785720245e-06,
+      "loss": 0.4087,
+      "step": 7639
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.875552589874051e-06,
+      "loss": 0.3973,
+      "step": 7640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.873884394027859e-06,
+      "loss": 0.4253,
+      "step": 7641
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.872216198181667e-06,
+      "loss": 0.4106,
+      "step": 7642
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.870548002335475e-06,
+      "loss": 0.4018,
+      "step": 7643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.868879806489283e-06,
+      "loss": 0.4044,
+      "step": 7644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.86721161064309e-06,
+      "loss": 0.4213,
+      "step": 7645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.865543414796898e-06,
+      "loss": 0.3811,
+      "step": 7646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.863875218950705e-06,
+      "loss": 0.4324,
+      "step": 7647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.862207023104514e-06,
+      "loss": 0.4102,
+      "step": 7648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.86053882725832e-06,
+      "loss": 0.3805,
+      "step": 7649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.858870631412128e-06,
+      "loss": 0.4072,
+      "step": 7650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.857202435565936e-06,
+      "loss": 0.3937,
+      "step": 7651
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.855534239719744e-06,
+      "loss": 0.3732,
+      "step": 7652
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.853866043873552e-06,
+      "loss": 0.3883,
+      "step": 7653
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.852197848027358e-06,
+      "loss": 0.379,
+      "step": 7654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.850529652181168e-06,
+      "loss": 0.4194,
+      "step": 7655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.848861456334974e-06,
+      "loss": 0.3959,
+      "step": 7656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.847193260488782e-06,
+      "loss": 0.401,
+      "step": 7657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.84552506464259e-06,
+      "loss": 0.3866,
+      "step": 7658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.843856868796398e-06,
+      "loss": 0.3913,
+      "step": 7659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.842188672950205e-06,
+      "loss": 0.3941,
+      "step": 7660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.840520477104012e-06,
+      "loss": 0.3853,
+      "step": 7661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.838852281257821e-06,
+      "loss": 0.4004,
+      "step": 7662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.837184085411627e-06,
+      "loss": 0.3945,
+      "step": 7663
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.835515889565435e-06,
+      "loss": 0.4043,
+      "step": 7664
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.833847693719243e-06,
+      "loss": 0.4182,
+      "step": 7665
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.832179497873051e-06,
+      "loss": 0.3945,
+      "step": 7666
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.830511302026859e-06,
+      "loss": 0.406,
+      "step": 7667
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.828843106180665e-06,
+      "loss": 0.4185,
+      "step": 7668
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.827174910334475e-06,
+      "loss": 0.402,
+      "step": 7669
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.825506714488281e-06,
+      "loss": 0.3969,
+      "step": 7670
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.823838518642089e-06,
+      "loss": 0.3714,
+      "step": 7671
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.822170322795897e-06,
+      "loss": 0.3971,
+      "step": 7672
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.820502126949705e-06,
+      "loss": 0.4167,
+      "step": 7673
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.818833931103512e-06,
+      "loss": 0.4063,
+      "step": 7674
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.817165735257319e-06,
+      "loss": 0.4133,
+      "step": 7675
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.815497539411128e-06,
+      "loss": 0.4043,
+      "step": 7676
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.813829343564934e-06,
+      "loss": 0.3899,
+      "step": 7677
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.812161147718742e-06,
+      "loss": 0.3923,
+      "step": 7678
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.81049295187255e-06,
+      "loss": 0.4001,
+      "step": 7679
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.808824756026358e-06,
+      "loss": 0.3912,
+      "step": 7680
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.807156560180166e-06,
+      "loss": 0.3764,
+      "step": 7681
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.805488364333972e-06,
+      "loss": 0.3861,
+      "step": 7682
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.803820168487782e-06,
+      "loss": 0.3997,
+      "step": 7683
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.802151972641588e-06,
+      "loss": 0.4193,
+      "step": 7684
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.800483776795398e-06,
+      "loss": 0.3772,
+      "step": 7685
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.798815580949204e-06,
+      "loss": 0.4294,
+      "step": 7686
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.797147385103012e-06,
+      "loss": 0.4018,
+      "step": 7687
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.79547918925682e-06,
+      "loss": 0.4174,
+      "step": 7688
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.793810993410626e-06,
+      "loss": 0.4,
+      "step": 7689
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.792142797564435e-06,
+      "loss": 0.3879,
+      "step": 7690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.790474601718241e-06,
+      "loss": 0.4022,
+      "step": 7691
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.788806405872051e-06,
+      "loss": 0.3975,
+      "step": 7692
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.787138210025857e-06,
+      "loss": 0.4031,
+      "step": 7693
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.785470014179665e-06,
+      "loss": 0.3707,
+      "step": 7694
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.783801818333473e-06,
+      "loss": 0.3954,
+      "step": 7695
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.782133622487281e-06,
+      "loss": 0.4019,
+      "step": 7696
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.780465426641089e-06,
+      "loss": 0.3875,
+      "step": 7697
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.778797230794895e-06,
+      "loss": 0.3779,
+      "step": 7698
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.777129034948705e-06,
+      "loss": 0.411,
+      "step": 7699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.77546083910251e-06,
+      "loss": 0.3999,
+      "step": 7700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.773792643256319e-06,
+      "loss": 0.4057,
+      "step": 7701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.772124447410127e-06,
+      "loss": 0.4087,
+      "step": 7702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.770456251563934e-06,
+      "loss": 0.406,
+      "step": 7703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.768788055717742e-06,
+      "loss": 0.3984,
+      "step": 7704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.767119859871548e-06,
+      "loss": 0.3848,
+      "step": 7705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.765451664025358e-06,
+      "loss": 0.3854,
+      "step": 7706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.763783468179164e-06,
+      "loss": 0.4144,
+      "step": 7707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.762115272332972e-06,
+      "loss": 0.415,
+      "step": 7708
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.76044707648678e-06,
+      "loss": 0.365,
+      "step": 7709
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.758778880640588e-06,
+      "loss": 0.3879,
+      "step": 7710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.757110684794396e-06,
+      "loss": 0.379,
+      "step": 7711
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.755442488948202e-06,
+      "loss": 0.3925,
+      "step": 7712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.753774293102012e-06,
+      "loss": 0.3954,
+      "step": 7713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.752106097255818e-06,
+      "loss": 0.3741,
+      "step": 7714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.750437901409626e-06,
+      "loss": 0.3876,
+      "step": 7715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.748769705563434e-06,
+      "loss": 0.4019,
+      "step": 7716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.747101509717241e-06,
+      "loss": 0.3944,
+      "step": 7717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.74543331387105e-06,
+      "loss": 0.4425,
+      "step": 7718
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.743765118024856e-06,
+      "loss": 0.3954,
+      "step": 7719
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.742096922178665e-06,
+      "loss": 0.417,
+      "step": 7720
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.740428726332471e-06,
+      "loss": 0.4453,
+      "step": 7721
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.738760530486281e-06,
+      "loss": 0.397,
+      "step": 7722
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.737092334640087e-06,
+      "loss": 0.3809,
+      "step": 7723
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.735424138793895e-06,
+      "loss": 0.4045,
+      "step": 7724
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.733755942947703e-06,
+      "loss": 0.3894,
+      "step": 7725
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.73208774710151e-06,
+      "loss": 0.3962,
+      "step": 7726
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.730419551255319e-06,
+      "loss": 0.4061,
+      "step": 7727
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.728751355409125e-06,
+      "loss": 0.4011,
+      "step": 7728
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.727083159562934e-06,
+      "loss": 0.3842,
+      "step": 7729
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.72541496371674e-06,
+      "loss": 0.4093,
+      "step": 7730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.723746767870548e-06,
+      "loss": 0.3691,
+      "step": 7731
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.722078572024356e-06,
+      "loss": 0.3864,
+      "step": 7732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.720410376178164e-06,
+      "loss": 0.3658,
+      "step": 7733
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.718742180331972e-06,
+      "loss": 0.4014,
+      "step": 7734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.717073984485778e-06,
+      "loss": 0.3976,
+      "step": 7735
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.715405788639588e-06,
+      "loss": 0.3816,
+      "step": 7736
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.713737592793394e-06,
+      "loss": 0.4102,
+      "step": 7737
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.712069396947202e-06,
+      "loss": 0.3739,
+      "step": 7738
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.71040120110101e-06,
+      "loss": 0.3736,
+      "step": 7739
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.708733005254818e-06,
+      "loss": 0.425,
+      "step": 7740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.707064809408626e-06,
+      "loss": 0.425,
+      "step": 7741
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.705396613562432e-06,
+      "loss": 0.3942,
+      "step": 7742
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.703728417716241e-06,
+      "loss": 0.4215,
+      "step": 7743
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.702060221870048e-06,
+      "loss": 0.3843,
+      "step": 7744
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.700392026023855e-06,
+      "loss": 0.3967,
+      "step": 7745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.698723830177663e-06,
+      "loss": 0.4059,
+      "step": 7746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.697055634331471e-06,
+      "loss": 0.4006,
+      "step": 7747
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.69538743848528e-06,
+      "loss": 0.3818,
+      "step": 7748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.693719242639085e-06,
+      "loss": 0.3949,
+      "step": 7749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.692051046792895e-06,
+      "loss": 0.4125,
+      "step": 7750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.690382850946701e-06,
+      "loss": 0.3925,
+      "step": 7751
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.688714655100509e-06,
+      "loss": 0.4142,
+      "step": 7752
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.687046459254317e-06,
+      "loss": 0.3665,
+      "step": 7753
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.685378263408125e-06,
+      "loss": 0.3973,
+      "step": 7754
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.683710067561933e-06,
+      "loss": 0.3896,
+      "step": 7755
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.682041871715739e-06,
+      "loss": 0.3938,
+      "step": 7756
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.680373675869548e-06,
+      "loss": 0.3892,
+      "step": 7757
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.678705480023355e-06,
+      "loss": 0.387,
+      "step": 7758
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.677037284177164e-06,
+      "loss": 0.4036,
+      "step": 7759
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.67536908833097e-06,
+      "loss": 0.4032,
+      "step": 7760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.673700892484778e-06,
+      "loss": 0.3867,
+      "step": 7761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.672032696638586e-06,
+      "loss": 0.3626,
+      "step": 7762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.670364500792394e-06,
+      "loss": 0.4112,
+      "step": 7763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.668696304946202e-06,
+      "loss": 0.3862,
+      "step": 7764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.667028109100008e-06,
+      "loss": 0.3917,
+      "step": 7765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.665359913253818e-06,
+      "loss": 0.3727,
+      "step": 7766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.663691717407624e-06,
+      "loss": 0.3803,
+      "step": 7767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.662023521561432e-06,
+      "loss": 0.3812,
+      "step": 7768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.66035532571524e-06,
+      "loss": 0.3778,
+      "step": 7769
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.658687129869048e-06,
+      "loss": 0.433,
+      "step": 7770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.657018934022855e-06,
+      "loss": 0.3804,
+      "step": 7771
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.655350738176662e-06,
+      "loss": 0.4026,
+      "step": 7772
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.653682542330471e-06,
+      "loss": 0.3957,
+      "step": 7773
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.652014346484277e-06,
+      "loss": 0.4134,
+      "step": 7774
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.650346150638085e-06,
+      "loss": 0.3934,
+      "step": 7775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.648677954791893e-06,
+      "loss": 0.4286,
+      "step": 7776
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.647009758945701e-06,
+      "loss": 0.4206,
+      "step": 7777
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.645341563099509e-06,
+      "loss": 0.3563,
+      "step": 7778
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.643673367253315e-06,
+      "loss": 0.4008,
+      "step": 7779
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.642005171407125e-06,
+      "loss": 0.3811,
+      "step": 7780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.640336975560931e-06,
+      "loss": 0.3918,
+      "step": 7781
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.638668779714739e-06,
+      "loss": 0.3877,
+      "step": 7782
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.637000583868547e-06,
+      "loss": 0.3647,
+      "step": 7783
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.635332388022355e-06,
+      "loss": 0.4153,
+      "step": 7784
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.633664192176163e-06,
+      "loss": 0.3749,
+      "step": 7785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.631995996329969e-06,
+      "loss": 0.3836,
+      "step": 7786
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.630327800483778e-06,
+      "loss": 0.3854,
+      "step": 7787
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.628659604637585e-06,
+      "loss": 0.425,
+      "step": 7788
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.626991408791393e-06,
+      "loss": 0.4007,
+      "step": 7789
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.6253232129452e-06,
+      "loss": 0.4087,
+      "step": 7790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.623655017099007e-06,
+      "loss": 0.3884,
+      "step": 7791
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.621986821252816e-06,
+      "loss": 0.3812,
+      "step": 7792
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.620318625406623e-06,
+      "loss": 0.4002,
+      "step": 7793
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.618650429560431e-06,
+      "loss": 0.4378,
+      "step": 7794
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.616982233714239e-06,
+      "loss": 0.3948,
+      "step": 7795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.615314037868047e-06,
+      "loss": 0.3897,
+      "step": 7796
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.613645842021854e-06,
+      "loss": 0.4286,
+      "step": 7797
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.611977646175661e-06,
+      "loss": 0.3782,
+      "step": 7798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.6103094503294695e-06,
+      "loss": 0.3997,
+      "step": 7799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.608641254483277e-06,
+      "loss": 0.4052,
+      "step": 7800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.606973058637085e-06,
+      "loss": 0.3766,
+      "step": 7801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.605304862790892e-06,
+      "loss": 0.3793,
+      "step": 7802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.6036366669447e-06,
+      "loss": 0.3863,
+      "step": 7803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.601968471098507e-06,
+      "loss": 0.3651,
+      "step": 7804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.600300275252314e-06,
+      "loss": 0.3888,
+      "step": 7805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.598632079406123e-06,
+      "loss": 0.3838,
+      "step": 7806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.59696388355993e-06,
+      "loss": 0.4075,
+      "step": 7807
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.595295687713739e-06,
+      "loss": 0.4012,
+      "step": 7808
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.593627491867546e-06,
+      "loss": 0.3606,
+      "step": 7809
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.591959296021354e-06,
+      "loss": 0.3983,
+      "step": 7810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.590291100175161e-06,
+      "loss": 0.4158,
+      "step": 7811
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.588622904328969e-06,
+      "loss": 0.4066,
+      "step": 7812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5869547084827766e-06,
+      "loss": 0.3845,
+      "step": 7813
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.585286512636584e-06,
+      "loss": 0.3906,
+      "step": 7814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.583618316790392e-06,
+      "loss": 0.4,
+      "step": 7815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.581950120944199e-06,
+      "loss": 0.3842,
+      "step": 7816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.580281925098007e-06,
+      "loss": 0.3927,
+      "step": 7817
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.578613729251814e-06,
+      "loss": 0.4315,
+      "step": 7818
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.576945533405622e-06,
+      "loss": 0.4007,
+      "step": 7819
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.57527733755943e-06,
+      "loss": 0.3908,
+      "step": 7820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.573609141713237e-06,
+      "loss": 0.3838,
+      "step": 7821
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.571940945867046e-06,
+      "loss": 0.3993,
+      "step": 7822
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.570272750020853e-06,
+      "loss": 0.4039,
+      "step": 7823
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.568604554174661e-06,
+      "loss": 0.3906,
+      "step": 7824
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.566936358328469e-06,
+      "loss": 0.409,
+      "step": 7825
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5652681624822766e-06,
+      "loss": 0.3954,
+      "step": 7826
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.563599966636084e-06,
+      "loss": 0.3927,
+      "step": 7827
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.561931770789891e-06,
+      "loss": 0.4188,
+      "step": 7828
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.560263574943699e-06,
+      "loss": 0.3816,
+      "step": 7829
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.558595379097506e-06,
+      "loss": 0.3897,
+      "step": 7830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.556927183251314e-06,
+      "loss": 0.4102,
+      "step": 7831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.555258987405122e-06,
+      "loss": 0.4216,
+      "step": 7832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.55359079155893e-06,
+      "loss": 0.3979,
+      "step": 7833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.551922595712737e-06,
+      "loss": 0.4095,
+      "step": 7834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.550254399866544e-06,
+      "loss": 0.383,
+      "step": 7835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.548586204020353e-06,
+      "loss": 0.3985,
+      "step": 7836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.54691800817416e-06,
+      "loss": 0.4059,
+      "step": 7837
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.545249812327969e-06,
+      "loss": 0.4062,
+      "step": 7838
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.543581616481776e-06,
+      "loss": 0.3871,
+      "step": 7839
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.541913420635584e-06,
+      "loss": 0.4245,
+      "step": 7840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.540245224789391e-06,
+      "loss": 0.3854,
+      "step": 7841
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.538577028943198e-06,
+      "loss": 0.381,
+      "step": 7842
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.536908833097006e-06,
+      "loss": 0.4393,
+      "step": 7843
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.5352406372508134e-06,
+      "loss": 0.3785,
+      "step": 7844
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.533572441404622e-06,
+      "loss": 0.4321,
+      "step": 7845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.531904245558429e-06,
+      "loss": 0.3719,
+      "step": 7846
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.530236049712237e-06,
+      "loss": 0.4107,
+      "step": 7847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.528567853866044e-06,
+      "loss": 0.3933,
+      "step": 7848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.526899658019852e-06,
+      "loss": 0.4443,
+      "step": 7849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.52523146217366e-06,
+      "loss": 0.3801,
+      "step": 7850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.523563266327467e-06,
+      "loss": 0.3781,
+      "step": 7851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.521895070481276e-06,
+      "loss": 0.3995,
+      "step": 7852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.520226874635083e-06,
+      "loss": 0.3835,
+      "step": 7853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.518558678788891e-06,
+      "loss": 0.3972,
+      "step": 7854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.516890482942698e-06,
+      "loss": 0.3559,
+      "step": 7855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.5152222870965055e-06,
+      "loss": 0.3941,
+      "step": 7856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.5135540912503134e-06,
+      "loss": 0.4217,
+      "step": 7857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.5118858954041205e-06,
+      "loss": 0.4111,
+      "step": 7858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.510217699557929e-06,
+      "loss": 0.3861,
+      "step": 7859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.508549503711736e-06,
+      "loss": 0.4016,
+      "step": 7860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.506881307865544e-06,
+      "loss": 0.3613,
+      "step": 7861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.505213112019352e-06,
+      "loss": 0.4013,
+      "step": 7862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.503544916173159e-06,
+      "loss": 0.3819,
+      "step": 7863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.501876720326967e-06,
+      "loss": 0.3755,
+      "step": 7864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.500208524480774e-06,
+      "loss": 0.3821,
+      "step": 7865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.498540328634583e-06,
+      "loss": 0.3766,
+      "step": 7866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.49687213278839e-06,
+      "loss": 0.3938,
+      "step": 7867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.495203936942198e-06,
+      "loss": 0.3826,
+      "step": 7868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4935357410960055e-06,
+      "loss": 0.3815,
+      "step": 7869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.491867545249813e-06,
+      "loss": 0.4045,
+      "step": 7870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4901993494036205e-06,
+      "loss": 0.4146,
+      "step": 7871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4885311535574275e-06,
+      "loss": 0.3846,
+      "step": 7872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.486862957711236e-06,
+      "loss": 0.3813,
+      "step": 7873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.485194761865043e-06,
+      "loss": 0.4039,
+      "step": 7874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.483526566018852e-06,
+      "loss": 0.4032,
+      "step": 7875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.481858370172659e-06,
+      "loss": 0.3875,
+      "step": 7876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.480190174326466e-06,
+      "loss": 0.3894,
+      "step": 7877
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.478521978480274e-06,
+      "loss": 0.3856,
+      "step": 7878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.476853782634082e-06,
+      "loss": 0.3905,
+      "step": 7879
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.47518558678789e-06,
+      "loss": 0.3911,
+      "step": 7880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.473517390941697e-06,
+      "loss": 0.3864,
+      "step": 7881
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4718491950955055e-06,
+      "loss": 0.3953,
+      "step": 7882
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4701809992493126e-06,
+      "loss": 0.4021,
+      "step": 7883
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4685128034031205e-06,
+      "loss": 0.3934,
+      "step": 7884
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.4668446075569275e-06,
+      "loss": 0.406,
+      "step": 7885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.465176411710735e-06,
+      "loss": 0.3929,
+      "step": 7886
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.463508215864543e-06,
+      "loss": 0.41,
+      "step": 7887
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.46184002001835e-06,
+      "loss": 0.4422,
+      "step": 7888
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.460171824172159e-06,
+      "loss": 0.385,
+      "step": 7889
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.458503628325966e-06,
+      "loss": 0.3685,
+      "step": 7890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.456835432479774e-06,
+      "loss": 0.4195,
+      "step": 7891
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.455167236633582e-06,
+      "loss": 0.3767,
+      "step": 7892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.453499040787389e-06,
+      "loss": 0.4082,
+      "step": 7893
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.451830844941197e-06,
+      "loss": 0.3891,
+      "step": 7894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.450162649095004e-06,
+      "loss": 0.4163,
+      "step": 7895
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4484944532488126e-06,
+      "loss": 0.4125,
+      "step": 7896
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.44682625740262e-06,
+      "loss": 0.424,
+      "step": 7897
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4451580615564275e-06,
+      "loss": 0.3965,
+      "step": 7898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.443489865710235e-06,
+      "loss": 0.387,
+      "step": 7899
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.441821669864042e-06,
+      "loss": 0.3731,
+      "step": 7900
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.44015347401785e-06,
+      "loss": 0.4042,
+      "step": 7901
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.438485278171657e-06,
+      "loss": 0.408,
+      "step": 7902
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.436817082325466e-06,
+      "loss": 0.39,
+      "step": 7903
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.435148886479273e-06,
+      "loss": 0.3809,
+      "step": 7904
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.433480690633081e-06,
+      "loss": 0.3831,
+      "step": 7905
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.431812494786889e-06,
+      "loss": 0.3814,
+      "step": 7906
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.430144298940696e-06,
+      "loss": 0.4062,
+      "step": 7907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.428476103094504e-06,
+      "loss": 0.3974,
+      "step": 7908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.426807907248311e-06,
+      "loss": 0.3919,
+      "step": 7909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.42513971140212e-06,
+      "loss": 0.4166,
+      "step": 7910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.423471515555927e-06,
+      "loss": 0.3879,
+      "step": 7911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.421803319709735e-06,
+      "loss": 0.4236,
+      "step": 7912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.420135123863542e-06,
+      "loss": 0.4144,
+      "step": 7913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.4184669280173494e-06,
+      "loss": 0.3937,
+      "step": 7914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.416798732171157e-06,
+      "loss": 0.3692,
+      "step": 7915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.415130536324965e-06,
+      "loss": 0.3747,
+      "step": 7916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.413462340478773e-06,
+      "loss": 0.3954,
+      "step": 7917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.41179414463258e-06,
+      "loss": 0.395,
+      "step": 7918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.410125948786389e-06,
+      "loss": 0.4214,
+      "step": 7919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.408457752940196e-06,
+      "loss": 0.4006,
+      "step": 7920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.406789557094003e-06,
+      "loss": 0.4116,
+      "step": 7921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.405121361247811e-06,
+      "loss": 0.3805,
+      "step": 7922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.403453165401619e-06,
+      "loss": 0.4194,
+      "step": 7923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.401784969555427e-06,
+      "loss": 0.406,
+      "step": 7924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.400116773709234e-06,
+      "loss": 0.4214,
+      "step": 7925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.398448577863042e-06,
+      "loss": 0.3696,
+      "step": 7926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.3967803820168494e-06,
+      "loss": 0.4191,
+      "step": 7927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.3951121861706565e-06,
+      "loss": 0.418,
+      "step": 7928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.393443990324465e-06,
+      "loss": 0.3941,
+      "step": 7929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.391775794478272e-06,
+      "loss": 0.4362,
+      "step": 7930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.39010759863208e-06,
+      "loss": 0.387,
+      "step": 7931
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.388439402785887e-06,
+      "loss": 0.427,
+      "step": 7932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.386771206939696e-06,
+      "loss": 0.4072,
+      "step": 7933
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.385103011093503e-06,
+      "loss": 0.3562,
+      "step": 7934
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.38343481524731e-06,
+      "loss": 0.3809,
+      "step": 7935
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.381766619401119e-06,
+      "loss": 0.3926,
+      "step": 7936
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.380098423554926e-06,
+      "loss": 0.3919,
+      "step": 7937
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.378430227708734e-06,
+      "loss": 0.4011,
+      "step": 7938
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.376762031862541e-06,
+      "loss": 0.371,
+      "step": 7939
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3750938360163494e-06,
+      "loss": 0.3769,
+      "step": 7940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3734256401701565e-06,
+      "loss": 0.4168,
+      "step": 7941
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.371757444323965e-06,
+      "loss": 0.4158,
+      "step": 7942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.370089248477772e-06,
+      "loss": 0.3939,
+      "step": 7943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 0.4115,
+      "step": 7944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.366752856785387e-06,
+      "loss": 0.3796,
+      "step": 7945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.365084660939194e-06,
+      "loss": 0.3872,
+      "step": 7946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.363416465093003e-06,
+      "loss": 0.4041,
+      "step": 7947
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.36174826924681e-06,
+      "loss": 0.3916,
+      "step": 7948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.360080073400619e-06,
+      "loss": 0.3795,
+      "step": 7949
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.358411877554426e-06,
+      "loss": 0.3903,
+      "step": 7950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.356743681708233e-06,
+      "loss": 0.3797,
+      "step": 7951
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.355075485862041e-06,
+      "loss": 0.4188,
+      "step": 7952
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3534072900158486e-06,
+      "loss": 0.3783,
+      "step": 7953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3517390941696565e-06,
+      "loss": 0.3824,
+      "step": 7954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3500708983234635e-06,
+      "loss": 0.3989,
+      "step": 7955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.348402702477272e-06,
+      "loss": 0.4048,
+      "step": 7956
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.346734506631079e-06,
+      "loss": 0.3852,
+      "step": 7957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.345066310784886e-06,
+      "loss": 0.3971,
+      "step": 7958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.343398114938694e-06,
+      "loss": 0.411,
+      "step": 7959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.341729919092502e-06,
+      "loss": 0.3736,
+      "step": 7960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.34006172324631e-06,
+      "loss": 0.419,
+      "step": 7961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.338393527400117e-06,
+      "loss": 0.3982,
+      "step": 7962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.336725331553926e-06,
+      "loss": 0.3993,
+      "step": 7963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.335057135707733e-06,
+      "loss": 0.3775,
+      "step": 7964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.33338893986154e-06,
+      "loss": 0.3865,
+      "step": 7965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3317207440153486e-06,
+      "loss": 0.3983,
+      "step": 7966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.330052548169156e-06,
+      "loss": 0.3678,
+      "step": 7967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3283843523229635e-06,
+      "loss": 0.366,
+      "step": 7968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.3267161564767705e-06,
+      "loss": 0.4015,
+      "step": 7969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.325047960630579e-06,
+      "loss": 0.3746,
+      "step": 7970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.323379764784386e-06,
+      "loss": 0.3889,
+      "step": 7971
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.321711568938193e-06,
+      "loss": 0.4052,
+      "step": 7972
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.320043373092002e-06,
+      "loss": 0.3775,
+      "step": 7973
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.318375177245809e-06,
+      "loss": 0.3983,
+      "step": 7974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.316706981399617e-06,
+      "loss": 0.3664,
+      "step": 7975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.315038785553424e-06,
+      "loss": 0.416,
+      "step": 7976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.313370589707233e-06,
+      "loss": 0.3779,
+      "step": 7977
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.31170239386104e-06,
+      "loss": 0.4055,
+      "step": 7978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.310034198014847e-06,
+      "loss": 0.4141,
+      "step": 7979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.308366002168656e-06,
+      "loss": 0.4161,
+      "step": 7980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.306697806322463e-06,
+      "loss": 0.4013,
+      "step": 7981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.3050296104762705e-06,
+      "loss": 0.4106,
+      "step": 7982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.3033614146300776e-06,
+      "loss": 0.4245,
+      "step": 7983
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.301693218783886e-06,
+      "loss": 0.351,
+      "step": 7984
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.300025022937693e-06,
+      "loss": 0.3951,
+      "step": 7985
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2983568270915e-06,
+      "loss": 0.4084,
+      "step": 7986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.296688631245309e-06,
+      "loss": 0.3975,
+      "step": 7987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.295020435399116e-06,
+      "loss": 0.3979,
+      "step": 7988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.293352239552924e-06,
+      "loss": 0.3833,
+      "step": 7989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.291684043706732e-06,
+      "loss": 0.409,
+      "step": 7990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.29001584786054e-06,
+      "loss": 0.4309,
+      "step": 7991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.288347652014347e-06,
+      "loss": 0.3792,
+      "step": 7992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.286679456168154e-06,
+      "loss": 0.3939,
+      "step": 7993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.285011260321963e-06,
+      "loss": 0.4016,
+      "step": 7994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.28334306447577e-06,
+      "loss": 0.3732,
+      "step": 7995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2816748686295776e-06,
+      "loss": 0.403,
+      "step": 7996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2800066727833854e-06,
+      "loss": 0.4242,
+      "step": 7997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.278338476937193e-06,
+      "loss": 0.4062,
+      "step": 7998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.276670281091e-06,
+      "loss": 0.4185,
+      "step": 7999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.275002085244809e-06,
+      "loss": 0.4206,
+      "step": 8000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.273333889398616e-06,
+      "loss": 0.4007,
+      "step": 8001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.271665693552423e-06,
+      "loss": 0.3944,
+      "step": 8002
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.269997497706232e-06,
+      "loss": 0.4169,
+      "step": 8003
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.268329301860039e-06,
+      "loss": 0.4171,
+      "step": 8004
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.266661106013847e-06,
+      "loss": 0.3935,
+      "step": 8005
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.264992910167654e-06,
+      "loss": 0.3612,
+      "step": 8006
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.263324714321463e-06,
+      "loss": 0.3833,
+      "step": 8007
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.26165651847527e-06,
+      "loss": 0.3705,
+      "step": 8008
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.259988322629077e-06,
+      "loss": 0.4131,
+      "step": 8009
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2583201267828854e-06,
+      "loss": 0.4037,
+      "step": 8010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2566519309366925e-06,
+      "loss": 0.4075,
+      "step": 8011
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.2549837350905e-06,
+      "loss": 0.4223,
+      "step": 8012
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.253315539244307e-06,
+      "loss": 0.3843,
+      "step": 8013
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.251647343398116e-06,
+      "loss": 0.387,
+      "step": 8014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.249979147551923e-06,
+      "loss": 0.3898,
+      "step": 8015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.24831095170573e-06,
+      "loss": 0.3927,
+      "step": 8016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.246642755859539e-06,
+      "loss": 0.383,
+      "step": 8017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.244974560013346e-06,
+      "loss": 0.3673,
+      "step": 8018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.243306364167154e-06,
+      "loss": 0.4157,
+      "step": 8019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.241638168320962e-06,
+      "loss": 0.4156,
+      "step": 8020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.23996997247477e-06,
+      "loss": 0.3726,
+      "step": 8021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.238301776628577e-06,
+      "loss": 0.3924,
+      "step": 8022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.236633580782384e-06,
+      "loss": 0.3971,
+      "step": 8023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2349653849361925e-06,
+      "loss": 0.378,
+      "step": 8024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2332971890899995e-06,
+      "loss": 0.4309,
+      "step": 8025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.231628993243807e-06,
+      "loss": 0.3909,
+      "step": 8026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.229960797397615e-06,
+      "loss": 0.4121,
+      "step": 8027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.228292601551423e-06,
+      "loss": 0.4158,
+      "step": 8028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.22662440570523e-06,
+      "loss": 0.4034,
+      "step": 8029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.224956209859037e-06,
+      "loss": 0.4002,
+      "step": 8030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.223288014012846e-06,
+      "loss": 0.3987,
+      "step": 8031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.221619818166653e-06,
+      "loss": 0.4012,
+      "step": 8032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.219951622320461e-06,
+      "loss": 0.4262,
+      "step": 8033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.218283426474269e-06,
+      "loss": 0.3998,
+      "step": 8034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.216615230628077e-06,
+      "loss": 0.4054,
+      "step": 8035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.214947034781884e-06,
+      "loss": 0.4122,
+      "step": 8036
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.213278838935691e-06,
+      "loss": 0.3816,
+      "step": 8037
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2116106430894995e-06,
+      "loss": 0.3776,
+      "step": 8038
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.2099424472433065e-06,
+      "loss": 0.393,
+      "step": 8039
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.208274251397115e-06,
+      "loss": 0.3992,
+      "step": 8040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.206606055550922e-06,
+      "loss": 0.3892,
+      "step": 8041
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.20493785970473e-06,
+      "loss": 0.3893,
+      "step": 8042
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.203269663858537e-06,
+      "loss": 0.3956,
+      "step": 8043
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.201601468012345e-06,
+      "loss": 0.3831,
+      "step": 8044
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.199933272166153e-06,
+      "loss": 0.4097,
+      "step": 8045
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.19826507631996e-06,
+      "loss": 0.3527,
+      "step": 8046
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.196596880473769e-06,
+      "loss": 0.4285,
+      "step": 8047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.194928684627576e-06,
+      "loss": 0.4216,
+      "step": 8048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.193260488781384e-06,
+      "loss": 0.4133,
+      "step": 8049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.191592292935191e-06,
+      "loss": 0.3928,
+      "step": 8050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.189924097088999e-06,
+      "loss": 0.4047,
+      "step": 8051
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.1882559012428065e-06,
+      "loss": 0.3974,
+      "step": 8052
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.1865877053966136e-06,
+      "loss": 0.3925,
+      "step": 8053
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.184919509550422e-06,
+      "loss": 0.415,
+      "step": 8054
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.183251313704229e-06,
+      "loss": 0.3801,
+      "step": 8055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.181583117858037e-06,
+      "loss": 0.3842,
+      "step": 8056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.179914922011845e-06,
+      "loss": 0.4347,
+      "step": 8057
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.178246726165652e-06,
+      "loss": 0.3955,
+      "step": 8058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.17657853031946e-06,
+      "loss": 0.379,
+      "step": 8059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.174910334473267e-06,
+      "loss": 0.3933,
+      "step": 8060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.173242138627076e-06,
+      "loss": 0.402,
+      "step": 8061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.171573942780883e-06,
+      "loss": 0.4191,
+      "step": 8062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.169905746934691e-06,
+      "loss": 0.3675,
+      "step": 8063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.168237551088499e-06,
+      "loss": 0.3735,
+      "step": 8064
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1665693552423065e-06,
+      "loss": 0.3893,
+      "step": 8065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1649011593961136e-06,
+      "loss": 0.4139,
+      "step": 8066
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.163232963549921e-06,
+      "loss": 0.4086,
+      "step": 8067
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.161564767703729e-06,
+      "loss": 0.4057,
+      "step": 8068
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.159896571857536e-06,
+      "loss": 0.363,
+      "step": 8069
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.158228376011345e-06,
+      "loss": 0.3848,
+      "step": 8070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.156560180165152e-06,
+      "loss": 0.3995,
+      "step": 8071
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.15489198431896e-06,
+      "loss": 0.4009,
+      "step": 8072
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.153223788472767e-06,
+      "loss": 0.4044,
+      "step": 8073
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.151555592626574e-06,
+      "loss": 0.3809,
+      "step": 8074
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.149887396780383e-06,
+      "loss": 0.3931,
+      "step": 8075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.14821920093419e-06,
+      "loss": 0.4133,
+      "step": 8076
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.146551005087999e-06,
+      "loss": 0.3811,
+      "step": 8077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.144882809241806e-06,
+      "loss": 0.4177,
+      "step": 8078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1432146133956135e-06,
+      "loss": 0.387,
+      "step": 8079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.141546417549421e-06,
+      "loss": 0.3978,
+      "step": 8080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1398782217032285e-06,
+      "loss": 0.3848,
+      "step": 8081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.138210025857036e-06,
+      "loss": 0.3999,
+      "step": 8082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.136541830010843e-06,
+      "loss": 0.3548,
+      "step": 8083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.134873634164652e-06,
+      "loss": 0.4026,
+      "step": 8084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.133205438318459e-06,
+      "loss": 0.4163,
+      "step": 8085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.131537242472267e-06,
+      "loss": 0.3992,
+      "step": 8086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.129869046626074e-06,
+      "loss": 0.3436,
+      "step": 8087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.128200850779882e-06,
+      "loss": 0.412,
+      "step": 8088
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.12653265493369e-06,
+      "loss": 0.44,
+      "step": 8089
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.124864459087497e-06,
+      "loss": 0.3764,
+      "step": 8090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.123196263241306e-06,
+      "loss": 0.4062,
+      "step": 8091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.121528067395113e-06,
+      "loss": 0.392,
+      "step": 8092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1198598715489206e-06,
+      "loss": 0.3856,
+      "step": 8093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1181916757027285e-06,
+      "loss": 0.4042,
+      "step": 8094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.1165234798565355e-06,
+      "loss": 0.3571,
+      "step": 8095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.114855284010343e-06,
+      "loss": 0.3906,
+      "step": 8096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.1131870881641504e-06,
+      "loss": 0.3893,
+      "step": 8097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.111518892317959e-06,
+      "loss": 0.3842,
+      "step": 8098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.109850696471766e-06,
+      "loss": 0.3843,
+      "step": 8099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.108182500625574e-06,
+      "loss": 0.3973,
+      "step": 8100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.106514304779382e-06,
+      "loss": 0.3919,
+      "step": 8101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.104846108933189e-06,
+      "loss": 0.3993,
+      "step": 8102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.103177913086997e-06,
+      "loss": 0.3496,
+      "step": 8103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.101509717240804e-06,
+      "loss": 0.3949,
+      "step": 8104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.099841521394613e-06,
+      "loss": 0.3966,
+      "step": 8105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.09817332554842e-06,
+      "loss": 0.3876,
+      "step": 8106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0965051297022285e-06,
+      "loss": 0.3925,
+      "step": 8107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0948369338560355e-06,
+      "loss": 0.4036,
+      "step": 8108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0931687380098425e-06,
+      "loss": 0.3753,
+      "step": 8109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.09150054216365e-06,
+      "loss": 0.4099,
+      "step": 8110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0898323463174575e-06,
+      "loss": 0.3866,
+      "step": 8111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.088164150471266e-06,
+      "loss": 0.38,
+      "step": 8112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.086495954625073e-06,
+      "loss": 0.4193,
+      "step": 8113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.084827758778882e-06,
+      "loss": 0.3898,
+      "step": 8114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.083159562932689e-06,
+      "loss": 0.386,
+      "step": 8115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.081491367086496e-06,
+      "loss": 0.3956,
+      "step": 8116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.079823171240304e-06,
+      "loss": 0.4158,
+      "step": 8117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.078154975394112e-06,
+      "loss": 0.3627,
+      "step": 8118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.07648677954792e-06,
+      "loss": 0.3696,
+      "step": 8119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.074818583701727e-06,
+      "loss": 0.4311,
+      "step": 8120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0731503878555355e-06,
+      "loss": 0.3975,
+      "step": 8121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0714821920093425e-06,
+      "loss": 0.4094,
+      "step": 8122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.06981399616315e-06,
+      "loss": 0.3749,
+      "step": 8123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0681458003169574e-06,
+      "loss": 0.3843,
+      "step": 8124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.066477604470765e-06,
+      "loss": 0.4031,
+      "step": 8125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.064809408624573e-06,
+      "loss": 0.3823,
+      "step": 8126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.06314121277838e-06,
+      "loss": 0.4129,
+      "step": 8127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.061473016932189e-06,
+      "loss": 0.3657,
+      "step": 8128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.059804821085996e-06,
+      "loss": 0.4155,
+      "step": 8129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.058136625239804e-06,
+      "loss": 0.4014,
+      "step": 8130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.056468429393612e-06,
+      "loss": 0.3945,
+      "step": 8131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.054800233547419e-06,
+      "loss": 0.3905,
+      "step": 8132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.053132037701227e-06,
+      "loss": 0.4036,
+      "step": 8133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.051463841855034e-06,
+      "loss": 0.388,
+      "step": 8134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0497956460088425e-06,
+      "loss": 0.3918,
+      "step": 8135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0481274501626496e-06,
+      "loss": 0.3849,
+      "step": 8136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.0464592543164574e-06,
+      "loss": 0.3912,
+      "step": 8137
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.044791058470265e-06,
+      "loss": 0.3754,
+      "step": 8138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.043122862624072e-06,
+      "loss": 0.4027,
+      "step": 8139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.04145466677788e-06,
+      "loss": 0.3929,
+      "step": 8140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.039786470931687e-06,
+      "loss": 0.3993,
+      "step": 8141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.038118275085496e-06,
+      "loss": 0.3921,
+      "step": 8142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.036450079239303e-06,
+      "loss": 0.3722,
+      "step": 8143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.034781883393112e-06,
+      "loss": 0.3975,
+      "step": 8144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.033113687546919e-06,
+      "loss": 0.4173,
+      "step": 8145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.031445491700726e-06,
+      "loss": 0.4098,
+      "step": 8146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.029777295854534e-06,
+      "loss": 0.3926,
+      "step": 8147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.028109100008342e-06,
+      "loss": 0.3767,
+      "step": 8148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0264409041621495e-06,
+      "loss": 0.3995,
+      "step": 8149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.024772708315957e-06,
+      "loss": 0.3768,
+      "step": 8150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.023104512469765e-06,
+      "loss": 0.397,
+      "step": 8151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.021436316623572e-06,
+      "loss": 0.3939,
+      "step": 8152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.019768120777379e-06,
+      "loss": 0.4032,
+      "step": 8153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.018099924931187e-06,
+      "loss": 0.3863,
+      "step": 8154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.016431729084995e-06,
+      "loss": 0.3895,
+      "step": 8155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.014763533238803e-06,
+      "loss": 0.3916,
+      "step": 8156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.01309533739261e-06,
+      "loss": 0.3913,
+      "step": 8157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.011427141546419e-06,
+      "loss": 0.3938,
+      "step": 8158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.009758945700226e-06,
+      "loss": 0.3948,
+      "step": 8159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.008090749854033e-06,
+      "loss": 0.3999,
+      "step": 8160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.006422554007841e-06,
+      "loss": 0.4045,
+      "step": 8161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.004754358161649e-06,
+      "loss": 0.4036,
+      "step": 8162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.0030861623154566e-06,
+      "loss": 0.4039,
+      "step": 8163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.001417966469264e-06,
+      "loss": 0.3821,
+      "step": 8164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.999749770623072e-06,
+      "loss": 0.3759,
+      "step": 8165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.998081574776879e-06,
+      "loss": 0.4026,
+      "step": 8166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.9964133789306864e-06,
+      "loss": 0.4075,
+      "step": 8167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.994745183084495e-06,
+      "loss": 0.4216,
+      "step": 8168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.993076987238302e-06,
+      "loss": 0.4156,
+      "step": 8169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.99140879139211e-06,
+      "loss": 0.3868,
+      "step": 8170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.989740595545917e-06,
+      "loss": 0.3711,
+      "step": 8171
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.988072399699726e-06,
+      "loss": 0.4109,
+      "step": 8172
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.986404203853533e-06,
+      "loss": 0.3701,
+      "step": 8173
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.98473600800734e-06,
+      "loss": 0.3941,
+      "step": 8174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.983067812161149e-06,
+      "loss": 0.3743,
+      "step": 8175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.981399616314956e-06,
+      "loss": 0.3914,
+      "step": 8176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.979731420468764e-06,
+      "loss": 0.3842,
+      "step": 8177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.978063224622571e-06,
+      "loss": 0.3699,
+      "step": 8178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.976395028776379e-06,
+      "loss": 0.3679,
+      "step": 8179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.974726832930186e-06,
+      "loss": 0.4286,
+      "step": 8180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.973058637083995e-06,
+      "loss": 0.3626,
+      "step": 8181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.971390441237802e-06,
+      "loss": 0.4051,
+      "step": 8182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.969722245391609e-06,
+      "loss": 0.389,
+      "step": 8183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.968054049545417e-06,
+      "loss": 0.3564,
+      "step": 8184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.966385853699225e-06,
+      "loss": 0.4269,
+      "step": 8185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.964717657853033e-06,
+      "loss": 0.4102,
+      "step": 8186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.96304946200684e-06,
+      "loss": 0.3776,
+      "step": 8187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.961381266160649e-06,
+      "loss": 0.3993,
+      "step": 8188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.959713070314456e-06,
+      "loss": 0.3782,
+      "step": 8189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.958044874468263e-06,
+      "loss": 0.389,
+      "step": 8190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.956376678622071e-06,
+      "loss": 0.3682,
+      "step": 8191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9547084827758785e-06,
+      "loss": 0.4267,
+      "step": 8192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.953040286929686e-06,
+      "loss": 0.3468,
+      "step": 8193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9513720910834935e-06,
+      "loss": 0.3902,
+      "step": 8194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.949703895237302e-06,
+      "loss": 0.3915,
+      "step": 8195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.948035699391109e-06,
+      "loss": 0.4191,
+      "step": 8196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.946367503544916e-06,
+      "loss": 0.4061,
+      "step": 8197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.944699307698725e-06,
+      "loss": 0.367,
+      "step": 8198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.943031111852532e-06,
+      "loss": 0.3897,
+      "step": 8199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.94136291600634e-06,
+      "loss": 0.3961,
+      "step": 8200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.939694720160147e-06,
+      "loss": 0.3753,
+      "step": 8201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.938026524313956e-06,
+      "loss": 0.3858,
+      "step": 8202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.936358328467763e-06,
+      "loss": 0.4011,
+      "step": 8203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.93469013262157e-06,
+      "loss": 0.3838,
+      "step": 8204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9330219367753785e-06,
+      "loss": 0.3819,
+      "step": 8205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9313537409291856e-06,
+      "loss": 0.4074,
+      "step": 8206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9296855450829934e-06,
+      "loss": 0.3857,
+      "step": 8207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9280173492368005e-06,
+      "loss": 0.3867,
+      "step": 8208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.926349153390609e-06,
+      "loss": 0.3978,
+      "step": 8209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.924680957544416e-06,
+      "loss": 0.3965,
+      "step": 8210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.923012761698223e-06,
+      "loss": 0.386,
+      "step": 8211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.921344565852032e-06,
+      "loss": 0.4066,
+      "step": 8212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.919676370005839e-06,
+      "loss": 0.4072,
+      "step": 8213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.918008174159647e-06,
+      "loss": 0.396,
+      "step": 8214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.916339978313454e-06,
+      "loss": 0.4297,
+      "step": 8215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.914671782467263e-06,
+      "loss": 0.3784,
+      "step": 8216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.91300358662107e-06,
+      "loss": 0.3788,
+      "step": 8217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.911335390774877e-06,
+      "loss": 0.4048,
+      "step": 8218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.9096671949286855e-06,
+      "loss": 0.4187,
+      "step": 8219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.907998999082493e-06,
+      "loss": 0.4296,
+      "step": 8220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.9063308032363005e-06,
+      "loss": 0.3925,
+      "step": 8221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.904662607390108e-06,
+      "loss": 0.4123,
+      "step": 8222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.902994411543916e-06,
+      "loss": 0.405,
+      "step": 8223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.901326215697723e-06,
+      "loss": 0.3658,
+      "step": 8224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.89965801985153e-06,
+      "loss": 0.3905,
+      "step": 8225
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.897989824005339e-06,
+      "loss": 0.4212,
+      "step": 8226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.896321628159146e-06,
+      "loss": 0.3941,
+      "step": 8227
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.894653432312954e-06,
+      "loss": 0.394,
+      "step": 8228
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.892985236466762e-06,
+      "loss": 0.4252,
+      "step": 8229
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.89131704062057e-06,
+      "loss": 0.3812,
+      "step": 8230
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.889648844774377e-06,
+      "loss": 0.3681,
+      "step": 8231
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.887980648928184e-06,
+      "loss": 0.4012,
+      "step": 8232
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8863124530819926e-06,
+      "loss": 0.4025,
+      "step": 8233
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8846442572358e-06,
+      "loss": 0.4227,
+      "step": 8234
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.882976061389608e-06,
+      "loss": 0.4245,
+      "step": 8235
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.881307865543415e-06,
+      "loss": 0.3897,
+      "step": 8236
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.879639669697223e-06,
+      "loss": 0.4008,
+      "step": 8237
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.87797147385103e-06,
+      "loss": 0.3882,
+      "step": 8238
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.876303278004839e-06,
+      "loss": 0.3948,
+      "step": 8239
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.874635082158646e-06,
+      "loss": 0.3874,
+      "step": 8240
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.872966886312453e-06,
+      "loss": 0.3639,
+      "step": 8241
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.871298690466262e-06,
+      "loss": 0.2955,
+      "step": 8242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.869630494620069e-06,
+      "loss": 0.2983,
+      "step": 8243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.867962298773877e-06,
+      "loss": 0.2967,
+      "step": 8244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.866294102927684e-06,
+      "loss": 0.2878,
+      "step": 8245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8646259070814926e-06,
+      "loss": 0.283,
+      "step": 8246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8629577112353e-06,
+      "loss": 0.2916,
+      "step": 8247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.861289515389107e-06,
+      "loss": 0.3065,
+      "step": 8248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.859621319542915e-06,
+      "loss": 0.2627,
+      "step": 8249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.857953123696722e-06,
+      "loss": 0.286,
+      "step": 8250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.85628492785053e-06,
+      "loss": 0.2917,
+      "step": 8251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.854616732004337e-06,
+      "loss": 0.2865,
+      "step": 8252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.852948536158146e-06,
+      "loss": 0.2931,
+      "step": 8253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.851280340311953e-06,
+      "loss": 0.2771,
+      "step": 8254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.84961214446576e-06,
+      "loss": 0.3048,
+      "step": 8255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.847943948619569e-06,
+      "loss": 0.282,
+      "step": 8256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.846275752773376e-06,
+      "loss": 0.2836,
+      "step": 8257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.844607556927184e-06,
+      "loss": 0.2599,
+      "step": 8258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.842939361080992e-06,
+      "loss": 0.2618,
+      "step": 8259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8412711652348e-06,
+      "loss": 0.2876,
+      "step": 8260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.839602969388607e-06,
+      "loss": 0.2885,
+      "step": 8261
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.837934773542414e-06,
+      "loss": 0.2738,
+      "step": 8262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.836266577696222e-06,
+      "loss": 0.2877,
+      "step": 8263
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8345983818500295e-06,
+      "loss": 0.2931,
+      "step": 8264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.832930186003837e-06,
+      "loss": 0.2643,
+      "step": 8265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.831261990157645e-06,
+      "loss": 0.2708,
+      "step": 8266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.829593794311453e-06,
+      "loss": 0.2621,
+      "step": 8267
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.82792559846526e-06,
+      "loss": 0.288,
+      "step": 8268
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.826257402619067e-06,
+      "loss": 0.2825,
+      "step": 8269
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.824589206772876e-06,
+      "loss": 0.2902,
+      "step": 8270
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.822921010926683e-06,
+      "loss": 0.2639,
+      "step": 8271
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.821252815080492e-06,
+      "loss": 0.266,
+      "step": 8272
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.819584619234299e-06,
+      "loss": 0.3024,
+      "step": 8273
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.817916423388107e-06,
+      "loss": 0.2982,
+      "step": 8274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.816248227541914e-06,
+      "loss": 0.2713,
+      "step": 8275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8145800316957216e-06,
+      "loss": 0.2801,
+      "step": 8276
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8129118358495294e-06,
+      "loss": 0.2934,
+      "step": 8277
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.8112436400033365e-06,
+      "loss": 0.2772,
+      "step": 8278
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.809575444157145e-06,
+      "loss": 0.2799,
+      "step": 8279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.807907248310952e-06,
+      "loss": 0.2794,
+      "step": 8280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.80623905246476e-06,
+      "loss": 0.2933,
+      "step": 8281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.804570856618567e-06,
+      "loss": 0.2788,
+      "step": 8282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.802902660772375e-06,
+      "loss": 0.2721,
+      "step": 8283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.801234464926183e-06,
+      "loss": 0.2742,
+      "step": 8284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.79956626907999e-06,
+      "loss": 0.2709,
+      "step": 8285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.797898073233799e-06,
+      "loss": 0.28,
+      "step": 8286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.796229877387606e-06,
+      "loss": 0.2725,
+      "step": 8287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.794561681541414e-06,
+      "loss": 0.2702,
+      "step": 8288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.792893485695221e-06,
+      "loss": 0.2929,
+      "step": 8289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.791225289849029e-06,
+      "loss": 0.2814,
+      "step": 8290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.7895570940028365e-06,
+      "loss": 0.2682,
+      "step": 8291
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.7878888981566435e-06,
+      "loss": 0.2758,
+      "step": 8292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.786220702310452e-06,
+      "loss": 0.28,
+      "step": 8293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.784552506464259e-06,
+      "loss": 0.268,
+      "step": 8294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.782884310618067e-06,
+      "loss": 0.2813,
+      "step": 8295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.781216114771875e-06,
+      "loss": 0.2784,
+      "step": 8296
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.779547918925683e-06,
+      "loss": 0.2761,
+      "step": 8297
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.77787972307949e-06,
+      "loss": 0.2822,
+      "step": 8298
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.776211527233297e-06,
+      "loss": 0.2855,
+      "step": 8299
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.774543331387106e-06,
+      "loss": 0.2792,
+      "step": 8300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.772875135540913e-06,
+      "loss": 0.2592,
+      "step": 8301
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.771206939694721e-06,
+      "loss": 0.2903,
+      "step": 8302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7695387438485286e-06,
+      "loss": 0.2696,
+      "step": 8303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7678705480023365e-06,
+      "loss": 0.267,
+      "step": 8304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7662023521561435e-06,
+      "loss": 0.2721,
+      "step": 8305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7645341563099505e-06,
+      "loss": 0.2633,
+      "step": 8306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.762865960463759e-06,
+      "loss": 0.2664,
+      "step": 8307
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.761197764617566e-06,
+      "loss": 0.2622,
+      "step": 8308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.759529568771375e-06,
+      "loss": 0.279,
+      "step": 8309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.757861372925182e-06,
+      "loss": 0.3001,
+      "step": 8310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.75619317707899e-06,
+      "loss": 0.2835,
+      "step": 8311
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.754524981232797e-06,
+      "loss": 0.2812,
+      "step": 8312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.752856785386605e-06,
+      "loss": 0.2865,
+      "step": 8313
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.751188589540413e-06,
+      "loss": 0.2956,
+      "step": 8314
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.74952039369422e-06,
+      "loss": 0.2801,
+      "step": 8315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7478521978480286e-06,
+      "loss": 0.2582,
+      "step": 8316
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.746184002001836e-06,
+      "loss": 0.2904,
+      "step": 8317
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7445158061556435e-06,
+      "loss": 0.288,
+      "step": 8318
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7428476103094505e-06,
+      "loss": 0.2572,
+      "step": 8319
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.741179414463258e-06,
+      "loss": 0.2776,
+      "step": 8320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.739511218617066e-06,
+      "loss": 0.2474,
+      "step": 8321
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.737843022770873e-06,
+      "loss": 0.2919,
+      "step": 8322
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.736174826924682e-06,
+      "loss": 0.2687,
+      "step": 8323
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.734506631078489e-06,
+      "loss": 0.2742,
+      "step": 8324
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.732838435232297e-06,
+      "loss": 0.2633,
+      "step": 8325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.731170239386105e-06,
+      "loss": 0.2855,
+      "step": 8326
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.729502043539912e-06,
+      "loss": 0.2767,
+      "step": 8327
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.72783384769372e-06,
+      "loss": 0.272,
+      "step": 8328
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.726165651847527e-06,
+      "loss": 0.275,
+      "step": 8329
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.724497456001336e-06,
+      "loss": 0.2739,
+      "step": 8330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.722829260155143e-06,
+      "loss": 0.2614,
+      "step": 8331
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7211610643089505e-06,
+      "loss": 0.281,
+      "step": 8332
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.719492868462758e-06,
+      "loss": 0.2548,
+      "step": 8333
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7178246726165655e-06,
+      "loss": 0.2832,
+      "step": 8334
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.716156476770373e-06,
+      "loss": 0.2772,
+      "step": 8335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.71448828092418e-06,
+      "loss": 0.2853,
+      "step": 8336
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.712820085077989e-06,
+      "loss": 0.279,
+      "step": 8337
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.711151889231796e-06,
+      "loss": 0.2673,
+      "step": 8338
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.709483693385604e-06,
+      "loss": 0.2788,
+      "step": 8339
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.707815497539412e-06,
+      "loss": 0.2908,
+      "step": 8340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.706147301693219e-06,
+      "loss": 0.2536,
+      "step": 8341
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.704479105847027e-06,
+      "loss": 0.2768,
+      "step": 8342
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.702810910000834e-06,
+      "loss": 0.2693,
+      "step": 8343
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.701142714154643e-06,
+      "loss": 0.2819,
+      "step": 8344
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.69947451830845e-06,
+      "loss": 0.2629,
+      "step": 8345
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.697806322462258e-06,
+      "loss": 0.2929,
+      "step": 8346
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6961381266160654e-06,
+      "loss": 0.2755,
+      "step": 8347
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6944699307698725e-06,
+      "loss": 0.2852,
+      "step": 8348
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.69280173492368e-06,
+      "loss": 0.2897,
+      "step": 8349
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.691133539077488e-06,
+      "loss": 0.2643,
+      "step": 8350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.689465343231296e-06,
+      "loss": 0.2637,
+      "step": 8351
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.687797147385103e-06,
+      "loss": 0.2778,
+      "step": 8352
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.686128951538912e-06,
+      "loss": 0.2829,
+      "step": 8353
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.684460755692719e-06,
+      "loss": 0.2735,
+      "step": 8354
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.682792559846527e-06,
+      "loss": 0.2668,
+      "step": 8355
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.681124364000334e-06,
+      "loss": 0.2652,
+      "step": 8356
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.679456168154142e-06,
+      "loss": 0.2736,
+      "step": 8357
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.67778797230795e-06,
+      "loss": 0.2577,
+      "step": 8358
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.676119776461757e-06,
+      "loss": 0.2853,
+      "step": 8359
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6744515806155654e-06,
+      "loss": 0.2575,
+      "step": 8360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6727833847693725e-06,
+      "loss": 0.2765,
+      "step": 8361
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.67111518892318e-06,
+      "loss": 0.2649,
+      "step": 8362
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.669446993076988e-06,
+      "loss": 0.2491,
+      "step": 8363
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.667778797230795e-06,
+      "loss": 0.2761,
+      "step": 8364
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.666110601384603e-06,
+      "loss": 0.2807,
+      "step": 8365
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.66444240553841e-06,
+      "loss": 0.284,
+      "step": 8366
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.662774209692219e-06,
+      "loss": 0.288,
+      "step": 8367
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.661106013846026e-06,
+      "loss": 0.2673,
+      "step": 8368
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.659437817999834e-06,
+      "loss": 0.2882,
+      "step": 8369
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.657769622153642e-06,
+      "loss": 0.2754,
+      "step": 8370
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.656101426307449e-06,
+      "loss": 0.2571,
+      "step": 8371
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.654433230461257e-06,
+      "loss": 0.2594,
+      "step": 8372
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.652765034615064e-06,
+      "loss": 0.2837,
+      "step": 8373
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6510968387688725e-06,
+      "loss": 0.2725,
+      "step": 8374
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.6494286429226795e-06,
+      "loss": 0.2704,
+      "step": 8375
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.647760447076488e-06,
+      "loss": 0.2621,
+      "step": 8376
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.646092251230295e-06,
+      "loss": 0.2829,
+      "step": 8377
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.644424055384102e-06,
+      "loss": 0.2714,
+      "step": 8378
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.64275585953791e-06,
+      "loss": 0.2629,
+      "step": 8379
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.641087663691717e-06,
+      "loss": 0.2638,
+      "step": 8380
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.639419467845526e-06,
+      "loss": 0.2799,
+      "step": 8381
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.637751271999333e-06,
+      "loss": 0.2794,
+      "step": 8382
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.636083076153142e-06,
+      "loss": 0.2756,
+      "step": 8383
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.634414880306949e-06,
+      "loss": 0.2677,
+      "step": 8384
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.632746684460756e-06,
+      "loss": 0.2774,
+      "step": 8385
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.631078488614564e-06,
+      "loss": 0.2911,
+      "step": 8386
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.629410292768372e-06,
+      "loss": 0.2824,
+      "step": 8387
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6277420969221795e-06,
+      "loss": 0.2471,
+      "step": 8388
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6260739010759865e-06,
+      "loss": 0.2964,
+      "step": 8389
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.624405705229795e-06,
+      "loss": 0.278,
+      "step": 8390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.622737509383602e-06,
+      "loss": 0.2719,
+      "step": 8391
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.621069313537409e-06,
+      "loss": 0.2861,
+      "step": 8392
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.619401117691217e-06,
+      "loss": 0.2784,
+      "step": 8393
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.617732921845025e-06,
+      "loss": 0.2592,
+      "step": 8394
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.616064725998833e-06,
+      "loss": 0.2619,
+      "step": 8395
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.61439653015264e-06,
+      "loss": 0.2654,
+      "step": 8396
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.612728334306449e-06,
+      "loss": 0.2704,
+      "step": 8397
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.611060138460256e-06,
+      "loss": 0.2753,
+      "step": 8398
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.609391942614063e-06,
+      "loss": 0.279,
+      "step": 8399
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.607723746767872e-06,
+      "loss": 0.2903,
+      "step": 8400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.606055550921679e-06,
+      "loss": 0.276,
+      "step": 8401
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6043873550754865e-06,
+      "loss": 0.2736,
+      "step": 8402
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.6027191592292936e-06,
+      "loss": 0.2697,
+      "step": 8403
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.601050963383102e-06,
+      "loss": 0.2901,
+      "step": 8404
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.599382767536909e-06,
+      "loss": 0.2719,
+      "step": 8405
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.597714571690716e-06,
+      "loss": 0.2544,
+      "step": 8406
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.596046375844525e-06,
+      "loss": 0.2792,
+      "step": 8407
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.594378179998332e-06,
+      "loss": 0.2683,
+      "step": 8408
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.59270998415214e-06,
+      "loss": 0.2796,
+      "step": 8409
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.591041788305947e-06,
+      "loss": 0.2705,
+      "step": 8410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.589373592459756e-06,
+      "loss": 0.2907,
+      "step": 8411
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.587705396613563e-06,
+      "loss": 0.2763,
+      "step": 8412
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.58603720076737e-06,
+      "loss": 0.2608,
+      "step": 8413
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.584369004921179e-06,
+      "loss": 0.2812,
+      "step": 8414
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.582700809074986e-06,
+      "loss": 0.2818,
+      "step": 8415
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.5810326132287936e-06,
+      "loss": 0.2876,
+      "step": 8416
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.579364417382601e-06,
+      "loss": 0.2922,
+      "step": 8417
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.577696221536409e-06,
+      "loss": 0.2769,
+      "step": 8418
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.576028025690216e-06,
+      "loss": 0.2881,
+      "step": 8419
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.574359829844025e-06,
+      "loss": 0.2836,
+      "step": 8420
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.572691633997832e-06,
+      "loss": 0.2794,
+      "step": 8421
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.571023438151639e-06,
+      "loss": 0.2909,
+      "step": 8422
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.569355242305447e-06,
+      "loss": 0.2977,
+      "step": 8423
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.567687046459255e-06,
+      "loss": 0.2815,
+      "step": 8424
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.566018850613063e-06,
+      "loss": 0.2713,
+      "step": 8425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.56435065476687e-06,
+      "loss": 0.27,
+      "step": 8426
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.562682458920679e-06,
+      "loss": 0.2761,
+      "step": 8427
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.561014263074486e-06,
+      "loss": 0.2838,
+      "step": 8428
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.559346067228293e-06,
+      "loss": 0.2622,
+      "step": 8429
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.557677871382101e-06,
+      "loss": 0.2664,
+      "step": 8430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5560096755359085e-06,
+      "loss": 0.2917,
+      "step": 8431
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.554341479689716e-06,
+      "loss": 0.2829,
+      "step": 8432
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.552673283843523e-06,
+      "loss": 0.2857,
+      "step": 8433
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.551005087997332e-06,
+      "loss": 0.239,
+      "step": 8434
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.549336892151139e-06,
+      "loss": 0.2483,
+      "step": 8435
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.547668696304946e-06,
+      "loss": 0.2697,
+      "step": 8436
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.546000500458755e-06,
+      "loss": 0.2618,
+      "step": 8437
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.544332304612562e-06,
+      "loss": 0.2808,
+      "step": 8438
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.54266410876637e-06,
+      "loss": 0.2521,
+      "step": 8439
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.540995912920177e-06,
+      "loss": 0.2586,
+      "step": 8440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.539327717073986e-06,
+      "loss": 0.2852,
+      "step": 8441
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.537659521227793e-06,
+      "loss": 0.2827,
+      "step": 8442
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5359913253816e-06,
+      "loss": 0.2797,
+      "step": 8443
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5343231295354085e-06,
+      "loss": 0.2839,
+      "step": 8444
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5326549336892155e-06,
+      "loss": 0.2756,
+      "step": 8445
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.530986737843023e-06,
+      "loss": 0.274,
+      "step": 8446
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5293185419968304e-06,
+      "loss": 0.2711,
+      "step": 8447
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.527650346150639e-06,
+      "loss": 0.2623,
+      "step": 8448
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.525982150304446e-06,
+      "loss": 0.2653,
+      "step": 8449
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.524313954458253e-06,
+      "loss": 0.2774,
+      "step": 8450
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.522645758612062e-06,
+      "loss": 0.2888,
+      "step": 8451
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.520977562765869e-06,
+      "loss": 0.2924,
+      "step": 8452
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.519309366919677e-06,
+      "loss": 0.2882,
+      "step": 8453
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.517641171073485e-06,
+      "loss": 0.2893,
+      "step": 8454
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.515972975227293e-06,
+      "loss": 0.2664,
+      "step": 8455
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5143047793811e-06,
+      "loss": 0.2644,
+      "step": 8456
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.512636583534907e-06,
+      "loss": 0.2644,
+      "step": 8457
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5109683876887155e-06,
+      "loss": 0.2654,
+      "step": 8458
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5093001918425225e-06,
+      "loss": 0.2684,
+      "step": 8459
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.50763199599633e-06,
+      "loss": 0.2671,
+      "step": 8460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.505963800150138e-06,
+      "loss": 0.2771,
+      "step": 8461
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.504295604303946e-06,
+      "loss": 0.2795,
+      "step": 8462
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.502627408457753e-06,
+      "loss": 0.299,
+      "step": 8463
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.50095921261156e-06,
+      "loss": 0.2655,
+      "step": 8464
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.499291016765369e-06,
+      "loss": 0.2761,
+      "step": 8465
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.497622820919176e-06,
+      "loss": 0.2732,
+      "step": 8466
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.495954625072984e-06,
+      "loss": 0.2644,
+      "step": 8467
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.494286429226792e-06,
+      "loss": 0.2642,
+      "step": 8468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4926182333806e-06,
+      "loss": 0.2781,
+      "step": 8469
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.490950037534407e-06,
+      "loss": 0.261,
+      "step": 8470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.489281841688214e-06,
+      "loss": 0.2825,
+      "step": 8471
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4876136458420225e-06,
+      "loss": 0.2756,
+      "step": 8472
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4859454499958296e-06,
+      "loss": 0.2655,
+      "step": 8473
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.484277254149638e-06,
+      "loss": 0.2883,
+      "step": 8474
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.482609058303445e-06,
+      "loss": 0.2799,
+      "step": 8475
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.480940862457253e-06,
+      "loss": 0.281,
+      "step": 8476
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.47927266661106e-06,
+      "loss": 0.2804,
+      "step": 8477
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.477604470764869e-06,
+      "loss": 0.2688,
+      "step": 8478
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.475936274918676e-06,
+      "loss": 0.2743,
+      "step": 8479
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.474268079072483e-06,
+      "loss": 0.3062,
+      "step": 8480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.472599883226292e-06,
+      "loss": 0.2968,
+      "step": 8481
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.470931687380099e-06,
+      "loss": 0.2702,
+      "step": 8482
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.469263491533907e-06,
+      "loss": 0.2722,
+      "step": 8483
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.467595295687714e-06,
+      "loss": 0.3124,
+      "step": 8484
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4659270998415225e-06,
+      "loss": 0.2802,
+      "step": 8485
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4642589039953296e-06,
+      "loss": 0.2892,
+      "step": 8486
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.462590708149137e-06,
+      "loss": 0.2644,
+      "step": 8487
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.460922512302945e-06,
+      "loss": 0.2646,
+      "step": 8488
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.459254316456752e-06,
+      "loss": 0.2714,
+      "step": 8489
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.45758612061056e-06,
+      "loss": 0.2812,
+      "step": 8490
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.455917924764368e-06,
+      "loss": 0.271,
+      "step": 8491
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.454249728918176e-06,
+      "loss": 0.2826,
+      "step": 8492
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.452581533071983e-06,
+      "loss": 0.2801,
+      "step": 8493
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.45091333722579e-06,
+      "loss": 0.2827,
+      "step": 8494
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.449245141379599e-06,
+      "loss": 0.2624,
+      "step": 8495
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.447576945533406e-06,
+      "loss": 0.2819,
+      "step": 8496
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.445908749687214e-06,
+      "loss": 0.2789,
+      "step": 8497
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.444240553841022e-06,
+      "loss": 0.2633,
+      "step": 8498
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.4425723579948295e-06,
+      "loss": 0.2696,
+      "step": 8499
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.440904162148637e-06,
+      "loss": 0.2701,
+      "step": 8500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.439235966302444e-06,
+      "loss": 0.2739,
+      "step": 8501
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.437567770456252e-06,
+      "loss": 0.2674,
+      "step": 8502
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.435899574610059e-06,
+      "loss": 0.2954,
+      "step": 8503
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.434231378763868e-06,
+      "loss": 0.2779,
+      "step": 8504
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.432563182917675e-06,
+      "loss": 0.278,
+      "step": 8505
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.430894987071483e-06,
+      "loss": 0.2806,
+      "step": 8506
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.42922679122529e-06,
+      "loss": 0.2708,
+      "step": 8507
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.427558595379097e-06,
+      "loss": 0.2618,
+      "step": 8508
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.425890399532906e-06,
+      "loss": 0.2784,
+      "step": 8509
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.424222203686713e-06,
+      "loss": 0.2721,
+      "step": 8510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.422554007840522e-06,
+      "loss": 0.2685,
+      "step": 8511
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.420885811994329e-06,
+      "loss": 0.2757,
+      "step": 8512
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.419217616148137e-06,
+      "loss": 0.2801,
+      "step": 8513
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.417549420301944e-06,
+      "loss": 0.261,
+      "step": 8514
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.4158812244557515e-06,
+      "loss": 0.2595,
+      "step": 8515
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.414213028609559e-06,
+      "loss": 0.28,
+      "step": 8516
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.4125448327633664e-06,
+      "loss": 0.2533,
+      "step": 8517
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.410876636917175e-06,
+      "loss": 0.2786,
+      "step": 8518
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.409208441070982e-06,
+      "loss": 0.2698,
+      "step": 8519
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.40754024522479e-06,
+      "loss": 0.2755,
+      "step": 8520
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.405872049378597e-06,
+      "loss": 0.2913,
+      "step": 8521
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.404203853532405e-06,
+      "loss": 0.2592,
+      "step": 8522
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.402535657686213e-06,
+      "loss": 0.2666,
+      "step": 8523
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.40086746184002e-06,
+      "loss": 0.2813,
+      "step": 8524
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.399199265993829e-06,
+      "loss": 0.2786,
+      "step": 8525
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.397531070147636e-06,
+      "loss": 0.2513,
+      "step": 8526
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.395862874301444e-06,
+      "loss": 0.2779,
+      "step": 8527
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3941946784552515e-06,
+      "loss": 0.2673,
+      "step": 8528
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3925264826090585e-06,
+      "loss": 0.255,
+      "step": 8529
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.390858286762866e-06,
+      "loss": 0.2931,
+      "step": 8530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3891900909166735e-06,
+      "loss": 0.2436,
+      "step": 8531
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.387521895070482e-06,
+      "loss": 0.2714,
+      "step": 8532
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.385853699224289e-06,
+      "loss": 0.2855,
+      "step": 8533
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.384185503378097e-06,
+      "loss": 0.2734,
+      "step": 8534
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.382517307531905e-06,
+      "loss": 0.262,
+      "step": 8535
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.380849111685713e-06,
+      "loss": 0.2816,
+      "step": 8536
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.37918091583952e-06,
+      "loss": 0.2787,
+      "step": 8537
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.377512719993327e-06,
+      "loss": 0.2973,
+      "step": 8538
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.375844524147136e-06,
+      "loss": 0.2718,
+      "step": 8539
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.374176328300943e-06,
+      "loss": 0.2587,
+      "step": 8540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3725081324547515e-06,
+      "loss": 0.2501,
+      "step": 8541
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3708399366085585e-06,
+      "loss": 0.2709,
+      "step": 8542
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.369171740762366e-06,
+      "loss": 0.2714,
+      "step": 8543
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3675035449161735e-06,
+      "loss": 0.2623,
+      "step": 8544
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.3658353490699805e-06,
+      "loss": 0.2763,
+      "step": 8545
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.364167153223789e-06,
+      "loss": 0.2864,
+      "step": 8546
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.362498957377596e-06,
+      "loss": 0.2672,
+      "step": 8547
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.360830761531405e-06,
+      "loss": 0.2746,
+      "step": 8548
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.359162565685212e-06,
+      "loss": 0.2739,
+      "step": 8549
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.35749436983902e-06,
+      "loss": 0.2654,
+      "step": 8550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.355826173992827e-06,
+      "loss": 0.2888,
+      "step": 8551
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.354157978146635e-06,
+      "loss": 0.2697,
+      "step": 8552
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.352489782300443e-06,
+      "loss": 0.2889,
+      "step": 8553
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.35082158645425e-06,
+      "loss": 0.2871,
+      "step": 8554
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3491533906080585e-06,
+      "loss": 0.2512,
+      "step": 8555
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3474851947618656e-06,
+      "loss": 0.2756,
+      "step": 8556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3458169989156734e-06,
+      "loss": 0.28,
+      "step": 8557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3441488030694805e-06,
+      "loss": 0.2755,
+      "step": 8558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.342480607223288e-06,
+      "loss": 0.2653,
+      "step": 8559
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.340812411377096e-06,
+      "loss": 0.2794,
+      "step": 8560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.339144215530903e-06,
+      "loss": 0.2621,
+      "step": 8561
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.337476019684712e-06,
+      "loss": 0.2602,
+      "step": 8562
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.335807823838519e-06,
+      "loss": 0.2942,
+      "step": 8563
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.334139627992327e-06,
+      "loss": 0.2726,
+      "step": 8564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.332471432146135e-06,
+      "loss": 0.2812,
+      "step": 8565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.330803236299942e-06,
+      "loss": 0.2773,
+      "step": 8566
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.32913504045375e-06,
+      "loss": 0.2619,
+      "step": 8567
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.327466844607557e-06,
+      "loss": 0.2829,
+      "step": 8568
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3257986487613655e-06,
+      "loss": 0.276,
+      "step": 8569
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.324130452915173e-06,
+      "loss": 0.2854,
+      "step": 8570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.3224622570689805e-06,
+      "loss": 0.2721,
+      "step": 8571
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.320794061222788e-06,
+      "loss": 0.2705,
+      "step": 8572
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.319125865376595e-06,
+      "loss": 0.2796,
+      "step": 8573
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.317457669530403e-06,
+      "loss": 0.2845,
+      "step": 8574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 0.2754,
+      "step": 8575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.314121277838019e-06,
+      "loss": 0.2803,
+      "step": 8576
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.312453081991826e-06,
+      "loss": 0.2782,
+      "step": 8577
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.310784886145635e-06,
+      "loss": 0.2665,
+      "step": 8578
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.309116690299442e-06,
+      "loss": 0.262,
+      "step": 8579
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.307448494453249e-06,
+      "loss": 0.291,
+      "step": 8580
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.305780298607057e-06,
+      "loss": 0.268,
+      "step": 8581
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.304112102760865e-06,
+      "loss": 0.2692,
+      "step": 8582
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.302443906914673e-06,
+      "loss": 0.2924,
+      "step": 8583
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.30077571106848e-06,
+      "loss": 0.2626,
+      "step": 8584
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.299107515222288e-06,
+      "loss": 0.2812,
+      "step": 8585
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.297439319376095e-06,
+      "loss": 0.2741,
+      "step": 8586
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.2957711235299024e-06,
+      "loss": 0.2772,
+      "step": 8587
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.29410292768371e-06,
+      "loss": 0.2623,
+      "step": 8588
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.292434731837518e-06,
+      "loss": 0.2852,
+      "step": 8589
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.290766535991326e-06,
+      "loss": 0.2761,
+      "step": 8590
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.289098340145133e-06,
+      "loss": 0.2587,
+      "step": 8591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.287430144298942e-06,
+      "loss": 0.2671,
+      "step": 8592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.285761948452749e-06,
+      "loss": 0.2788,
+      "step": 8593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.284093752606557e-06,
+      "loss": 0.2628,
+      "step": 8594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.282425556760364e-06,
+      "loss": 0.2787,
+      "step": 8595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.280757360914172e-06,
+      "loss": 0.2695,
+      "step": 8596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.27908916506798e-06,
+      "loss": 0.282,
+      "step": 8597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.277420969221787e-06,
+      "loss": 0.2818,
+      "step": 8598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.275752773375595e-06,
+      "loss": 0.2564,
+      "step": 8599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2740845775294024e-06,
+      "loss": 0.3021,
+      "step": 8600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.27241638168321e-06,
+      "loss": 0.2892,
+      "step": 8601
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.270748185837018e-06,
+      "loss": 0.2942,
+      "step": 8602
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.269079989990825e-06,
+      "loss": 0.2794,
+      "step": 8603
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.267411794144633e-06,
+      "loss": 0.2798,
+      "step": 8604
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.26574359829844e-06,
+      "loss": 0.2705,
+      "step": 8605
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.264075402452249e-06,
+      "loss": 0.2462,
+      "step": 8606
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.262407206606056e-06,
+      "loss": 0.2539,
+      "step": 8607
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.260739010759864e-06,
+      "loss": 0.2764,
+      "step": 8608
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.259070814913672e-06,
+      "loss": 0.2969,
+      "step": 8609
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.257402619067479e-06,
+      "loss": 0.2555,
+      "step": 8610
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.255734423221287e-06,
+      "loss": 0.2939,
+      "step": 8611
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.254066227375094e-06,
+      "loss": 0.2936,
+      "step": 8612
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.252398031528902e-06,
+      "loss": 0.2774,
+      "step": 8613
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2507298356827095e-06,
+      "loss": 0.2763,
+      "step": 8614
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.249061639836518e-06,
+      "loss": 0.2702,
+      "step": 8615
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.247393443990325e-06,
+      "loss": 0.2802,
+      "step": 8616
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.245725248144132e-06,
+      "loss": 0.2871,
+      "step": 8617
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.24405705229794e-06,
+      "loss": 0.2593,
+      "step": 8618
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.242388856451748e-06,
+      "loss": 0.2664,
+      "step": 8619
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.240720660605556e-06,
+      "loss": 0.2904,
+      "step": 8620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.239052464759363e-06,
+      "loss": 0.2832,
+      "step": 8621
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.237384268913172e-06,
+      "loss": 0.264,
+      "step": 8622
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.235716073066979e-06,
+      "loss": 0.2676,
+      "step": 8623
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.234047877220786e-06,
+      "loss": 0.2904,
+      "step": 8624
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.232379681374594e-06,
+      "loss": 0.2695,
+      "step": 8625
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2307114855284016e-06,
+      "loss": 0.2794,
+      "step": 8626
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2290432896822094e-06,
+      "loss": 0.2757,
+      "step": 8627
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.2273750938360165e-06,
+      "loss": 0.2776,
+      "step": 8628
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.225706897989825e-06,
+      "loss": 0.2727,
+      "step": 8629
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.224038702143632e-06,
+      "loss": 0.2609,
+      "step": 8630
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.222370506297439e-06,
+      "loss": 0.2586,
+      "step": 8631
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.220702310451248e-06,
+      "loss": 0.2712,
+      "step": 8632
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.219034114605055e-06,
+      "loss": 0.2776,
+      "step": 8633
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.217365918758863e-06,
+      "loss": 0.2811,
+      "step": 8634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.21569772291267e-06,
+      "loss": 0.2769,
+      "step": 8635
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.214029527066479e-06,
+      "loss": 0.2614,
+      "step": 8636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.212361331220286e-06,
+      "loss": 0.2702,
+      "step": 8637
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.210693135374093e-06,
+      "loss": 0.2704,
+      "step": 8638
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2090249395279015e-06,
+      "loss": 0.2899,
+      "step": 8639
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.207356743681709e-06,
+      "loss": 0.2824,
+      "step": 8640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2056885478355165e-06,
+      "loss": 0.2764,
+      "step": 8641
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.2040203519893235e-06,
+      "loss": 0.2759,
+      "step": 8642
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.202352156143132e-06,
+      "loss": 0.277,
+      "step": 8643
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.200683960296939e-06,
+      "loss": 0.2833,
+      "step": 8644
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.199015764450746e-06,
+      "loss": 0.2836,
+      "step": 8645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.197347568604555e-06,
+      "loss": 0.2779,
+      "step": 8646
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.195679372758362e-06,
+      "loss": 0.2566,
+      "step": 8647
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.19401117691217e-06,
+      "loss": 0.2849,
+      "step": 8648
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.192342981065977e-06,
+      "loss": 0.2997,
+      "step": 8649
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.190674785219786e-06,
+      "loss": 0.2778,
+      "step": 8650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.189006589373593e-06,
+      "loss": 0.2805,
+      "step": 8651
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1873383935274015e-06,
+      "loss": 0.2783,
+      "step": 8652
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.185670197681209e-06,
+      "loss": 0.2683,
+      "step": 8653
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.184002001835016e-06,
+      "loss": 0.2612,
+      "step": 8654
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1823338059888235e-06,
+      "loss": 0.2464,
+      "step": 8655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.180665610142631e-06,
+      "loss": 0.2711,
+      "step": 8656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.178997414296439e-06,
+      "loss": 0.2712,
+      "step": 8657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.177329218450246e-06,
+      "loss": 0.2815,
+      "step": 8658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.175661022604055e-06,
+      "loss": 0.2624,
+      "step": 8659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.173992826757862e-06,
+      "loss": 0.266,
+      "step": 8660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.172324630911669e-06,
+      "loss": 0.2701,
+      "step": 8661
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.170656435065477e-06,
+      "loss": 0.2993,
+      "step": 8662
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.168988239219285e-06,
+      "loss": 0.2669,
+      "step": 8663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.167320043373093e-06,
+      "loss": 0.2832,
+      "step": 8664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1656518475269e-06,
+      "loss": 0.2862,
+      "step": 8665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1639836516807086e-06,
+      "loss": 0.2669,
+      "step": 8666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.162315455834516e-06,
+      "loss": 0.2754,
+      "step": 8667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.160647259988323e-06,
+      "loss": 0.2798,
+      "step": 8668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.158979064142131e-06,
+      "loss": 0.2754,
+      "step": 8669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.1573108682959384e-06,
+      "loss": 0.2677,
+      "step": 8670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.155642672449746e-06,
+      "loss": 0.2472,
+      "step": 8671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.153974476603553e-06,
+      "loss": 0.2768,
+      "step": 8672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.152306280757362e-06,
+      "loss": 0.2532,
+      "step": 8673
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.150638084911169e-06,
+      "loss": 0.2851,
+      "step": 8674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.148969889064976e-06,
+      "loss": 0.2776,
+      "step": 8675
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.147301693218785e-06,
+      "loss": 0.2857,
+      "step": 8676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.145633497372592e-06,
+      "loss": 0.2706,
+      "step": 8677
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1439653015264e-06,
+      "loss": 0.2735,
+      "step": 8678
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.142297105680207e-06,
+      "loss": 0.2811,
+      "step": 8679
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.140628909834016e-06,
+      "loss": 0.2605,
+      "step": 8680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.138960713987823e-06,
+      "loss": 0.2657,
+      "step": 8681
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.13729251814163e-06,
+      "loss": 0.2808,
+      "step": 8682
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.135624322295438e-06,
+      "loss": 0.2665,
+      "step": 8683
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1339561264492455e-06,
+      "loss": 0.2874,
+      "step": 8684
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.132287930603053e-06,
+      "loss": 0.2734,
+      "step": 8685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.13061973475686e-06,
+      "loss": 0.2649,
+      "step": 8686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.128951538910669e-06,
+      "loss": 0.2702,
+      "step": 8687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.127283343064476e-06,
+      "loss": 0.2848,
+      "step": 8688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.125615147218283e-06,
+      "loss": 0.2819,
+      "step": 8689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.123946951372092e-06,
+      "loss": 0.2815,
+      "step": 8690
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.122278755525899e-06,
+      "loss": 0.2798,
+      "step": 8691
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.120610559679707e-06,
+      "loss": 0.2871,
+      "step": 8692
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.118942363833515e-06,
+      "loss": 0.2726,
+      "step": 8693
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.117274167987323e-06,
+      "loss": 0.2704,
+      "step": 8694
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.11560597214113e-06,
+      "loss": 0.2507,
+      "step": 8695
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.113937776294937e-06,
+      "loss": 0.2945,
+      "step": 8696
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1122695804487454e-06,
+      "loss": 0.261,
+      "step": 8697
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.1106013846025525e-06,
+      "loss": 0.2748,
+      "step": 8698
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.10893318875636e-06,
+      "loss": 0.2863,
+      "step": 8699
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.107264992910168e-06,
+      "loss": 0.276,
+      "step": 8700
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.105596797063976e-06,
+      "loss": 0.288,
+      "step": 8701
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.103928601217783e-06,
+      "loss": 0.2692,
+      "step": 8702
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.10226040537159e-06,
+      "loss": 0.2812,
+      "step": 8703
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.100592209525399e-06,
+      "loss": 0.2803,
+      "step": 8704
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.098924013679206e-06,
+      "loss": 0.2537,
+      "step": 8705
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.097255817833015e-06,
+      "loss": 0.2728,
+      "step": 8706
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.095587621986822e-06,
+      "loss": 0.2759,
+      "step": 8707
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.09391942614063e-06,
+      "loss": 0.2698,
+      "step": 8708
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.092251230294437e-06,
+      "loss": 0.2969,
+      "step": 8709
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.090583034448245e-06,
+      "loss": 0.2585,
+      "step": 8710
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.0889148386020525e-06,
+      "loss": 0.2905,
+      "step": 8711
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.0872466427558595e-06,
+      "loss": 0.2679,
+      "step": 8712
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.085578446909668e-06,
+      "loss": 0.2667,
+      "step": 8713
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.083910251063475e-06,
+      "loss": 0.2707,
+      "step": 8714
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.082242055217283e-06,
+      "loss": 0.2954,
+      "step": 8715
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.08057385937109e-06,
+      "loss": 0.2796,
+      "step": 8716
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.078905663524899e-06,
+      "loss": 0.2733,
+      "step": 8717
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.077237467678706e-06,
+      "loss": 0.2621,
+      "step": 8718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.075569271832513e-06,
+      "loss": 0.269,
+      "step": 8719
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.073901075986322e-06,
+      "loss": 0.2632,
+      "step": 8720
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.072232880140129e-06,
+      "loss": 0.2602,
+      "step": 8721
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.070564684293937e-06,
+      "loss": 0.2876,
+      "step": 8722
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.068896488447744e-06,
+      "loss": 0.2846,
+      "step": 8723
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0672282926015525e-06,
+      "loss": 0.2621,
+      "step": 8724
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0655600967553595e-06,
+      "loss": 0.2753,
+      "step": 8725
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0638919009091665e-06,
+      "loss": 0.2692,
+      "step": 8726
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.062223705062975e-06,
+      "loss": 0.2815,
+      "step": 8727
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.060555509216782e-06,
+      "loss": 0.285,
+      "step": 8728
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.05888731337059e-06,
+      "loss": 0.256,
+      "step": 8729
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.057219117524398e-06,
+      "loss": 0.2607,
+      "step": 8730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.055550921678206e-06,
+      "loss": 0.2701,
+      "step": 8731
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.053882725832013e-06,
+      "loss": 0.2689,
+      "step": 8732
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.05221452998582e-06,
+      "loss": 0.2731,
+      "step": 8733
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.050546334139629e-06,
+      "loss": 0.2638,
+      "step": 8734
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.048878138293436e-06,
+      "loss": 0.2713,
+      "step": 8735
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.047209942447244e-06,
+      "loss": 0.286,
+      "step": 8736
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.045541746601052e-06,
+      "loss": 0.2797,
+      "step": 8737
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0438735507548595e-06,
+      "loss": 0.2685,
+      "step": 8738
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0422053549086665e-06,
+      "loss": 0.2631,
+      "step": 8739
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.040537159062474e-06,
+      "loss": 0.2778,
+      "step": 8740
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.038868963216282e-06,
+      "loss": 0.2601,
+      "step": 8741
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.037200767370089e-06,
+      "loss": 0.2712,
+      "step": 8742
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.035532571523898e-06,
+      "loss": 0.2755,
+      "step": 8743
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.033864375677705e-06,
+      "loss": 0.2643,
+      "step": 8744
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.032196179831513e-06,
+      "loss": 0.2693,
+      "step": 8745
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.03052798398532e-06,
+      "loss": 0.2964,
+      "step": 8746
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.028859788139128e-06,
+      "loss": 0.2775,
+      "step": 8747
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.027191592292936e-06,
+      "loss": 0.2717,
+      "step": 8748
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.025523396446743e-06,
+      "loss": 0.2863,
+      "step": 8749
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.023855200600552e-06,
+      "loss": 0.2649,
+      "step": 8750
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.022187004754359e-06,
+      "loss": 0.2653,
+      "step": 8751
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0205188089081665e-06,
+      "loss": 0.2634,
+      "step": 8752
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0188506130619736e-06,
+      "loss": 0.2824,
+      "step": 8753
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.0171824172157815e-06,
+      "loss": 0.264,
+      "step": 8754
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.015514221369589e-06,
+      "loss": 0.2626,
+      "step": 8755
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.013846025523396e-06,
+      "loss": 0.2547,
+      "step": 8756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.012177829677205e-06,
+      "loss": 0.2791,
+      "step": 8757
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.010509633831012e-06,
+      "loss": 0.2791,
+      "step": 8758
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.00884143798482e-06,
+      "loss": 0.2843,
+      "step": 8759
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.007173242138628e-06,
+      "loss": 0.2736,
+      "step": 8760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.005505046292435e-06,
+      "loss": 0.2596,
+      "step": 8761
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.003836850446243e-06,
+      "loss": 0.2689,
+      "step": 8762
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.00216865460005e-06,
+      "loss": 0.2611,
+      "step": 8763
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.000500458753859e-06,
+      "loss": 0.2733,
+      "step": 8764
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.998832262907666e-06,
+      "loss": 0.2769,
+      "step": 8765
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9971640670614736e-06,
+      "loss": 0.2722,
+      "step": 8766
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9954958712152814e-06,
+      "loss": 0.2694,
+      "step": 8767
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9938276753690885e-06,
+      "loss": 0.2486,
+      "step": 8768
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.992159479522896e-06,
+      "loss": 0.2948,
+      "step": 8769
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.990491283676703e-06,
+      "loss": 0.2888,
+      "step": 8770
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.988823087830512e-06,
+      "loss": 0.2733,
+      "step": 8771
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.987154891984319e-06,
+      "loss": 0.2807,
+      "step": 8772
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.985486696138127e-06,
+      "loss": 0.2728,
+      "step": 8773
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.983818500291935e-06,
+      "loss": 0.2751,
+      "step": 8774
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.982150304445743e-06,
+      "loss": 0.2861,
+      "step": 8775
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.98048210859955e-06,
+      "loss": 0.2796,
+      "step": 8776
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.978813912753357e-06,
+      "loss": 0.2608,
+      "step": 8777
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.977145716907166e-06,
+      "loss": 0.2714,
+      "step": 8778
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.975477521060973e-06,
+      "loss": 0.292,
+      "step": 8779
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9738093252147814e-06,
+      "loss": 0.2685,
+      "step": 8780
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9721411293685885e-06,
+      "loss": 0.2913,
+      "step": 8781
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.970472933522396e-06,
+      "loss": 0.2776,
+      "step": 8782
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.968804737676203e-06,
+      "loss": 0.2758,
+      "step": 8783
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.967136541830011e-06,
+      "loss": 0.2672,
+      "step": 8784
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.965468345983819e-06,
+      "loss": 0.274,
+      "step": 8785
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.963800150137626e-06,
+      "loss": 0.2625,
+      "step": 8786
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.962131954291435e-06,
+      "loss": 0.2881,
+      "step": 8787
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.960463758445242e-06,
+      "loss": 0.2963,
+      "step": 8788
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.95879556259905e-06,
+      "loss": 0.27,
+      "step": 8789
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.957127366752857e-06,
+      "loss": 0.2639,
+      "step": 8790
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.955459170906665e-06,
+      "loss": 0.2877,
+      "step": 8791
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.953790975060473e-06,
+      "loss": 0.2728,
+      "step": 8792
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.95212277921428e-06,
+      "loss": 0.2739,
+      "step": 8793
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9504545833680885e-06,
+      "loss": 0.2776,
+      "step": 8794
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.9487863875218955e-06,
+      "loss": 0.3004,
+      "step": 8795
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.947118191675703e-06,
+      "loss": 0.2698,
+      "step": 8796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.945449995829511e-06,
+      "loss": 0.261,
+      "step": 8797
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.943781799983318e-06,
+      "loss": 0.2653,
+      "step": 8798
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.942113604137126e-06,
+      "loss": 0.2747,
+      "step": 8799
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.940445408290933e-06,
+      "loss": 0.2641,
+      "step": 8800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.938777212444742e-06,
+      "loss": 0.298,
+      "step": 8801
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.937109016598549e-06,
+      "loss": 0.2618,
+      "step": 8802
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.935440820752357e-06,
+      "loss": 0.269,
+      "step": 8803
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.933772624906165e-06,
+      "loss": 0.2863,
+      "step": 8804
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.932104429059972e-06,
+      "loss": 0.2809,
+      "step": 8805
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.93043623321378e-06,
+      "loss": 0.2731,
+      "step": 8806
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.928768037367587e-06,
+      "loss": 0.2818,
+      "step": 8807
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9270998415213955e-06,
+      "loss": 0.2869,
+      "step": 8808
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9254316456752025e-06,
+      "loss": 0.2633,
+      "step": 8809
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.923763449829011e-06,
+      "loss": 0.2736,
+      "step": 8810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.922095253982818e-06,
+      "loss": 0.2741,
+      "step": 8811
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.920427058136625e-06,
+      "loss": 0.2666,
+      "step": 8812
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.918758862290433e-06,
+      "loss": 0.2664,
+      "step": 8813
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.91709066644424e-06,
+      "loss": 0.289,
+      "step": 8814
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.915422470598049e-06,
+      "loss": 0.2872,
+      "step": 8815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.913754274751856e-06,
+      "loss": 0.2559,
+      "step": 8816
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.912086078905665e-06,
+      "loss": 0.2837,
+      "step": 8817
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.910417883059472e-06,
+      "loss": 0.2948,
+      "step": 8818
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.908749687213279e-06,
+      "loss": 0.2592,
+      "step": 8819
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.907081491367087e-06,
+      "loss": 0.2698,
+      "step": 8820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.905413295520895e-06,
+      "loss": 0.2763,
+      "step": 8821
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9037450996747025e-06,
+      "loss": 0.2826,
+      "step": 8822
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.9020769038285096e-06,
+      "loss": 0.2664,
+      "step": 8823
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.900408707982318e-06,
+      "loss": 0.2568,
+      "step": 8824
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.898740512136125e-06,
+      "loss": 0.2603,
+      "step": 8825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.897072316289932e-06,
+      "loss": 0.2663,
+      "step": 8826
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.89540412044374e-06,
+      "loss": 0.2641,
+      "step": 8827
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.893735924597548e-06,
+      "loss": 0.2609,
+      "step": 8828
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.892067728751356e-06,
+      "loss": 0.2853,
+      "step": 8829
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.890399532905163e-06,
+      "loss": 0.251,
+      "step": 8830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.888731337058972e-06,
+      "loss": 0.3004,
+      "step": 8831
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.887063141212779e-06,
+      "loss": 0.2808,
+      "step": 8832
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.885394945366587e-06,
+      "loss": 0.2965,
+      "step": 8833
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.883726749520395e-06,
+      "loss": 0.2666,
+      "step": 8834
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.882058553674202e-06,
+      "loss": 0.2789,
+      "step": 8835
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.8803903578280096e-06,
+      "loss": 0.2721,
+      "step": 8836
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.878722161981817e-06,
+      "loss": 0.2889,
+      "step": 8837
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.877053966135625e-06,
+      "loss": 0.2844,
+      "step": 8838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.875385770289432e-06,
+      "loss": 0.2856,
+      "step": 8839
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.87371757444324e-06,
+      "loss": 0.2775,
+      "step": 8840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.872049378597048e-06,
+      "loss": 0.2653,
+      "step": 8841
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.870381182750855e-06,
+      "loss": 0.2745,
+      "step": 8842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.868712986904663e-06,
+      "loss": 0.2914,
+      "step": 8843
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.86704479105847e-06,
+      "loss": 0.2922,
+      "step": 8844
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.865376595212279e-06,
+      "loss": 0.2928,
+      "step": 8845
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.863708399366086e-06,
+      "loss": 0.2708,
+      "step": 8846
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.862040203519895e-06,
+      "loss": 0.2835,
+      "step": 8847
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.860372007673702e-06,
+      "loss": 0.2513,
+      "step": 8848
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.858703811827509e-06,
+      "loss": 0.2841,
+      "step": 8849
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.857035615981317e-06,
+      "loss": 0.2777,
+      "step": 8850
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.855367420135124e-06,
+      "loss": 0.2695,
+      "step": 8851
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.853699224288932e-06,
+      "loss": 0.2886,
+      "step": 8852
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.852031028442739e-06,
+      "loss": 0.2702,
+      "step": 8853
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.850362832596548e-06,
+      "loss": 0.2863,
+      "step": 8854
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.848694636750355e-06,
+      "loss": 0.2821,
+      "step": 8855
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.847026440904162e-06,
+      "loss": 0.2779,
+      "step": 8856
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.84535824505797e-06,
+      "loss": 0.2705,
+      "step": 8857
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.843690049211778e-06,
+      "loss": 0.28,
+      "step": 8858
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.842021853365586e-06,
+      "loss": 0.2841,
+      "step": 8859
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.840353657519393e-06,
+      "loss": 0.2781,
+      "step": 8860
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.838685461673202e-06,
+      "loss": 0.2939,
+      "step": 8861
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.837017265827009e-06,
+      "loss": 0.2893,
+      "step": 8862
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.835349069980816e-06,
+      "loss": 0.268,
+      "step": 8863
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.833680874134624e-06,
+      "loss": 0.2846,
+      "step": 8864
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8320126782884315e-06,
+      "loss": 0.2726,
+      "step": 8865
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.830344482442239e-06,
+      "loss": 0.295,
+      "step": 8866
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8286762865960464e-06,
+      "loss": 0.285,
+      "step": 8867
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.827008090749855e-06,
+      "loss": 0.2643,
+      "step": 8868
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.825339894903662e-06,
+      "loss": 0.2726,
+      "step": 8869
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.823671699057469e-06,
+      "loss": 0.27,
+      "step": 8870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.822003503211278e-06,
+      "loss": 0.2884,
+      "step": 8871
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.820335307365085e-06,
+      "loss": 0.275,
+      "step": 8872
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.818667111518893e-06,
+      "loss": 0.273,
+      "step": 8873
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8169989156727e-06,
+      "loss": 0.2838,
+      "step": 8874
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.815330719826509e-06,
+      "loss": 0.2745,
+      "step": 8875
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.813662523980316e-06,
+      "loss": 0.2748,
+      "step": 8876
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.811994328134123e-06,
+      "loss": 0.267,
+      "step": 8877
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8103261322879315e-06,
+      "loss": 0.2717,
+      "step": 8878
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.8086579364417385e-06,
+      "loss": 0.2585,
+      "step": 8879
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.8069897405955464e-06,
+      "loss": 0.2672,
+      "step": 8880
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.8053215447493535e-06,
+      "loss": 0.2708,
+      "step": 8881
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.803653348903162e-06,
+      "loss": 0.2724,
+      "step": 8882
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.801985153056969e-06,
+      "loss": 0.2699,
+      "step": 8883
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.800316957210776e-06,
+      "loss": 0.2607,
+      "step": 8884
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.798648761364585e-06,
+      "loss": 0.2727,
+      "step": 8885
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.796980565518392e-06,
+      "loss": 0.2662,
+      "step": 8886
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7953123696722e-06,
+      "loss": 0.2588,
+      "step": 8887
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.793644173826008e-06,
+      "loss": 0.2891,
+      "step": 8888
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.791975977979816e-06,
+      "loss": 0.2812,
+      "step": 8889
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.790307782133623e-06,
+      "loss": 0.2906,
+      "step": 8890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7886395862874315e-06,
+      "loss": 0.3013,
+      "step": 8891
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7869713904412385e-06,
+      "loss": 0.2841,
+      "step": 8892
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7853031945950456e-06,
+      "loss": 0.2863,
+      "step": 8893
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7836349987488535e-06,
+      "loss": 0.2882,
+      "step": 8894
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.781966802902661e-06,
+      "loss": 0.2979,
+      "step": 8895
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.780298607056469e-06,
+      "loss": 0.2534,
+      "step": 8896
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.778630411210276e-06,
+      "loss": 0.2579,
+      "step": 8897
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.776962215364085e-06,
+      "loss": 0.2718,
+      "step": 8898
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.775294019517892e-06,
+      "loss": 0.283,
+      "step": 8899
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.773625823671699e-06,
+      "loss": 0.2637,
+      "step": 8900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.771957627825507e-06,
+      "loss": 0.2812,
+      "step": 8901
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.770289431979315e-06,
+      "loss": 0.2815,
+      "step": 8902
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.768621236133123e-06,
+      "loss": 0.2865,
+      "step": 8903
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.76695304028693e-06,
+      "loss": 0.2845,
+      "step": 8904
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7652848444407385e-06,
+      "loss": 0.2638,
+      "step": 8905
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7636166485945456e-06,
+      "loss": 0.285,
+      "step": 8906
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.761948452748353e-06,
+      "loss": 0.293,
+      "step": 8907
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.760280256902161e-06,
+      "loss": 0.2828,
+      "step": 8908
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.758612061055968e-06,
+      "loss": 0.274,
+      "step": 8909
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.756943865209776e-06,
+      "loss": 0.2841,
+      "step": 8910
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.755275669363583e-06,
+      "loss": 0.2706,
+      "step": 8911
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.753607473517392e-06,
+      "loss": 0.2532,
+      "step": 8912
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.751939277671199e-06,
+      "loss": 0.2809,
+      "step": 8913
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.750271081825006e-06,
+      "loss": 0.2745,
+      "step": 8914
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.748602885978815e-06,
+      "loss": 0.2779,
+      "step": 8915
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.746934690132622e-06,
+      "loss": 0.2833,
+      "step": 8916
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.74526649428643e-06,
+      "loss": 0.2844,
+      "step": 8917
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.743598298440237e-06,
+      "loss": 0.2789,
+      "step": 8918
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7419301025940456e-06,
+      "loss": 0.2904,
+      "step": 8919
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740261906747853e-06,
+      "loss": 0.2838,
+      "step": 8920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.73859371090166e-06,
+      "loss": 0.2655,
+      "step": 8921
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.736925515055468e-06,
+      "loss": 0.2619,
+      "step": 8922
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.735257319209275e-06,
+      "loss": 0.2719,
+      "step": 8923
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.733589123363083e-06,
+      "loss": 0.2745,
+      "step": 8924
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.731920927516891e-06,
+      "loss": 0.2699,
+      "step": 8925
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.730252731670699e-06,
+      "loss": 0.2805,
+      "step": 8926
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.728584535824506e-06,
+      "loss": 0.2809,
+      "step": 8927
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.726916339978313e-06,
+      "loss": 0.2595,
+      "step": 8928
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.725248144132122e-06,
+      "loss": 0.2545,
+      "step": 8929
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.723579948285929e-06,
+      "loss": 0.2729,
+      "step": 8930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.721911752439737e-06,
+      "loss": 0.2699,
+      "step": 8931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.720243556593545e-06,
+      "loss": 0.2691,
+      "step": 8932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.718575360747353e-06,
+      "loss": 0.2675,
+      "step": 8933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.71690716490116e-06,
+      "loss": 0.2914,
+      "step": 8934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.715238969054967e-06,
+      "loss": 0.3119,
+      "step": 8935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.713570773208775e-06,
+      "loss": 0.2824,
+      "step": 8936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7119025773625824e-06,
+      "loss": 0.2743,
+      "step": 8937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.710234381516391e-06,
+      "loss": 0.2709,
+      "step": 8938
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.708566185670198e-06,
+      "loss": 0.2662,
+      "step": 8939
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.706897989824006e-06,
+      "loss": 0.2704,
+      "step": 8940
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.705229793977813e-06,
+      "loss": 0.2781,
+      "step": 8941
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.70356159813162e-06,
+      "loss": 0.2847,
+      "step": 8942
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.701893402285429e-06,
+      "loss": 0.2704,
+      "step": 8943
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.700225206439236e-06,
+      "loss": 0.2773,
+      "step": 8944
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.698557010593045e-06,
+      "loss": 0.277,
+      "step": 8945
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.696888814746852e-06,
+      "loss": 0.2794,
+      "step": 8946
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.69522061890066e-06,
+      "loss": 0.2725,
+      "step": 8947
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.693552423054467e-06,
+      "loss": 0.275,
+      "step": 8948
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.691884227208275e-06,
+      "loss": 0.2987,
+      "step": 8949
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6902160313620824e-06,
+      "loss": 0.2625,
+      "step": 8950
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6885478355158895e-06,
+      "loss": 0.2695,
+      "step": 8951
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.686879639669698e-06,
+      "loss": 0.2881,
+      "step": 8952
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.685211443823505e-06,
+      "loss": 0.2772,
+      "step": 8953
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.683543247977313e-06,
+      "loss": 0.2806,
+      "step": 8954
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.68187505213112e-06,
+      "loss": 0.2626,
+      "step": 8955
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.680206856284929e-06,
+      "loss": 0.2723,
+      "step": 8956
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.678538660438736e-06,
+      "loss": 0.3026,
+      "step": 8957
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.676870464592543e-06,
+      "loss": 0.2595,
+      "step": 8958
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.675202268746352e-06,
+      "loss": 0.2705,
+      "step": 8959
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.673534072900159e-06,
+      "loss": 0.2884,
+      "step": 8960
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.671865877053967e-06,
+      "loss": 0.277,
+      "step": 8961
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.6701976812077745e-06,
+      "loss": 0.2739,
+      "step": 8962
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.668529485361582e-06,
+      "loss": 0.2763,
+      "step": 8963
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6668612895153895e-06,
+      "loss": 0.29,
+      "step": 8964
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6651930936691965e-06,
+      "loss": 0.2636,
+      "step": 8965
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.663524897823005e-06,
+      "loss": 0.2803,
+      "step": 8966
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.661856701976812e-06,
+      "loss": 0.2726,
+      "step": 8967
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.66018850613062e-06,
+      "loss": 0.2647,
+      "step": 8968
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.658520310284428e-06,
+      "loss": 0.2836,
+      "step": 8969
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.656852114438236e-06,
+      "loss": 0.2857,
+      "step": 8970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.655183918592043e-06,
+      "loss": 0.2665,
+      "step": 8971
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.65351572274585e-06,
+      "loss": 0.2901,
+      "step": 8972
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.651847526899659e-06,
+      "loss": 0.2778,
+      "step": 8973
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.650179331053466e-06,
+      "loss": 0.2675,
+      "step": 8974
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6485111352072745e-06,
+      "loss": 0.2714,
+      "step": 8975
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6468429393610816e-06,
+      "loss": 0.261,
+      "step": 8976
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6451747435148894e-06,
+      "loss": 0.2836,
+      "step": 8977
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6435065476686965e-06,
+      "loss": 0.2583,
+      "step": 8978
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6418383518225035e-06,
+      "loss": 0.2987,
+      "step": 8979
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.640170155976312e-06,
+      "loss": 0.2757,
+      "step": 8980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.638501960130119e-06,
+      "loss": 0.2788,
+      "step": 8981
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.636833764283928e-06,
+      "loss": 0.2569,
+      "step": 8982
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.635165568437735e-06,
+      "loss": 0.267,
+      "step": 8983
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.633497372591543e-06,
+      "loss": 0.2709,
+      "step": 8984
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.63182917674535e-06,
+      "loss": 0.247,
+      "step": 8985
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.630160980899158e-06,
+      "loss": 0.2665,
+      "step": 8986
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.628492785052966e-06,
+      "loss": 0.275,
+      "step": 8987
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.626824589206773e-06,
+      "loss": 0.2756,
+      "step": 8988
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6251563933605816e-06,
+      "loss": 0.3034,
+      "step": 8989
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.623488197514389e-06,
+      "loss": 0.2742,
+      "step": 8990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6218200016681965e-06,
+      "loss": 0.2649,
+      "step": 8991
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.6201518058220035e-06,
+      "loss": 0.2614,
+      "step": 8992
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.618483609975811e-06,
+      "loss": 0.2731,
+      "step": 8993
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.616815414129619e-06,
+      "loss": 0.2824,
+      "step": 8994
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.615147218283426e-06,
+      "loss": 0.2917,
+      "step": 8995
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.613479022437235e-06,
+      "loss": 0.276,
+      "step": 8996
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.611810826591042e-06,
+      "loss": 0.2701,
+      "step": 8997
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.61014263074485e-06,
+      "loss": 0.2653,
+      "step": 8998
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.608474434898658e-06,
+      "loss": 0.2747,
+      "step": 8999
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.606806239052465e-06,
+      "loss": 0.2729,
+      "step": 9000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.605138043206273e-06,
+      "loss": 0.2819,
+      "step": 9001
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.60346984736008e-06,
+      "loss": 0.2873,
+      "step": 9002
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.601801651513889e-06,
+      "loss": 0.2665,
+      "step": 9003
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.600133455667696e-06,
+      "loss": 0.281,
+      "step": 9004
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5984652598215035e-06,
+      "loss": 0.2708,
+      "step": 9005
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.596797063975311e-06,
+      "loss": 0.2965,
+      "step": 9006
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5951288681291184e-06,
+      "loss": 0.279,
+      "step": 9007
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.593460672282926e-06,
+      "loss": 0.2682,
+      "step": 9008
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.591792476436733e-06,
+      "loss": 0.2828,
+      "step": 9009
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.590124280590542e-06,
+      "loss": 0.2861,
+      "step": 9010
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.588456084744349e-06,
+      "loss": 0.2512,
+      "step": 9011
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.586787888898158e-06,
+      "loss": 0.2797,
+      "step": 9012
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.585119693051965e-06,
+      "loss": 0.2643,
+      "step": 9013
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.583451497205773e-06,
+      "loss": 0.2727,
+      "step": 9014
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.58178330135958e-06,
+      "loss": 0.2745,
+      "step": 9015
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.580115105513388e-06,
+      "loss": 0.2753,
+      "step": 9016
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.578446909667196e-06,
+      "loss": 0.2685,
+      "step": 9017
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.576778713821003e-06,
+      "loss": 0.2875,
+      "step": 9018
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.575110517974811e-06,
+      "loss": 0.2793,
+      "step": 9019
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5734423221286184e-06,
+      "loss": 0.2883,
+      "step": 9020
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.571774126282426e-06,
+      "loss": 0.2652,
+      "step": 9021
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.570105930436233e-06,
+      "loss": 0.2752,
+      "step": 9022
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.568437734590041e-06,
+      "loss": 0.2771,
+      "step": 9023
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.566769538743849e-06,
+      "loss": 0.2632,
+      "step": 9024
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.565101342897656e-06,
+      "loss": 0.2647,
+      "step": 9025
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.563433147051465e-06,
+      "loss": 0.2709,
+      "step": 9026
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.561764951205272e-06,
+      "loss": 0.2497,
+      "step": 9027
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.56009675535908e-06,
+      "loss": 0.2769,
+      "step": 9028
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.558428559512887e-06,
+      "loss": 0.2854,
+      "step": 9029
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.556760363666695e-06,
+      "loss": 0.2664,
+      "step": 9030
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.555092167820503e-06,
+      "loss": 0.2722,
+      "step": 9031
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.55342397197431e-06,
+      "loss": 0.2895,
+      "step": 9032
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.551755776128118e-06,
+      "loss": 0.2597,
+      "step": 9033
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.5500875802819255e-06,
+      "loss": 0.2594,
+      "step": 9034
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.548419384435733e-06,
+      "loss": 0.2698,
+      "step": 9035
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.546751188589541e-06,
+      "loss": 0.2764,
+      "step": 9036
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.545082992743348e-06,
+      "loss": 0.2661,
+      "step": 9037
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.543414796897156e-06,
+      "loss": 0.2636,
+      "step": 9038
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.541746601050963e-06,
+      "loss": 0.2864,
+      "step": 9039
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.540078405204772e-06,
+      "loss": 0.2608,
+      "step": 9040
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.538410209358579e-06,
+      "loss": 0.2841,
+      "step": 9041
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.536742013512387e-06,
+      "loss": 0.2904,
+      "step": 9042
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.535073817666195e-06,
+      "loss": 0.273,
+      "step": 9043
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.533405621820002e-06,
+      "loss": 0.2789,
+      "step": 9044
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.53173742597381e-06,
+      "loss": 0.2703,
+      "step": 9045
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.530069230127617e-06,
+      "loss": 0.2834,
+      "step": 9046
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5284010342814255e-06,
+      "loss": 0.2473,
+      "step": 9047
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5267328384352325e-06,
+      "loss": 0.2671,
+      "step": 9048
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.525064642589041e-06,
+      "loss": 0.2647,
+      "step": 9049
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.523396446742848e-06,
+      "loss": 0.2759,
+      "step": 9050
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.521728250896655e-06,
+      "loss": 0.2622,
+      "step": 9051
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.520060055050463e-06,
+      "loss": 0.2719,
+      "step": 9052
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.518391859204271e-06,
+      "loss": 0.2803,
+      "step": 9053
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.516723663358079e-06,
+      "loss": 0.2784,
+      "step": 9054
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.515055467511886e-06,
+      "loss": 0.2712,
+      "step": 9055
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.513387271665695e-06,
+      "loss": 0.2662,
+      "step": 9056
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.511719075819502e-06,
+      "loss": 0.2675,
+      "step": 9057
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.510050879973309e-06,
+      "loss": 0.2677,
+      "step": 9058
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.508382684127117e-06,
+      "loss": 0.2516,
+      "step": 9059
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.506714488280925e-06,
+      "loss": 0.2899,
+      "step": 9060
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5050462924347325e-06,
+      "loss": 0.2742,
+      "step": 9061
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.5033780965885395e-06,
+      "loss": 0.2802,
+      "step": 9062
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.501709900742348e-06,
+      "loss": 0.2504,
+      "step": 9063
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.500041704896155e-06,
+      "loss": 0.2875,
+      "step": 9064
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.498373509049962e-06,
+      "loss": 0.2537,
+      "step": 9065
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.496705313203771e-06,
+      "loss": 0.2762,
+      "step": 9066
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.495037117357578e-06,
+      "loss": 0.2594,
+      "step": 9067
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.493368921511386e-06,
+      "loss": 0.2828,
+      "step": 9068
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.491700725665193e-06,
+      "loss": 0.2841,
+      "step": 9069
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.490032529819002e-06,
+      "loss": 0.261,
+      "step": 9070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.488364333972809e-06,
+      "loss": 0.2771,
+      "step": 9071
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.486696138126617e-06,
+      "loss": 0.2952,
+      "step": 9072
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.485027942280425e-06,
+      "loss": 0.287,
+      "step": 9073
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.483359746434232e-06,
+      "loss": 0.2647,
+      "step": 9074
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4816915505880395e-06,
+      "loss": 0.2646,
+      "step": 9075
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4800233547418466e-06,
+      "loss": 0.2732,
+      "step": 9076
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.478355158895655e-06,
+      "loss": 0.2903,
+      "step": 9077
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.476686963049462e-06,
+      "loss": 0.2815,
+      "step": 9078
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.47501876720327e-06,
+      "loss": 0.2726,
+      "step": 9079
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.473350571357078e-06,
+      "loss": 0.2714,
+      "step": 9080
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.471682375510885e-06,
+      "loss": 0.272,
+      "step": 9081
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.470014179664693e-06,
+      "loss": 0.2752,
+      "step": 9082
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.4683459838185e-06,
+      "loss": 0.2969,
+      "step": 9083
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.466677787972309e-06,
+      "loss": 0.2748,
+      "step": 9084
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.465009592126116e-06,
+      "loss": 0.2755,
+      "step": 9085
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4633413962799246e-06,
+      "loss": 0.2765,
+      "step": 9086
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.461673200433732e-06,
+      "loss": 0.2631,
+      "step": 9087
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.460005004587539e-06,
+      "loss": 0.2686,
+      "step": 9088
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4583368087413465e-06,
+      "loss": 0.258,
+      "step": 9089
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4566686128951544e-06,
+      "loss": 0.2659,
+      "step": 9090
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.455000417048962e-06,
+      "loss": 0.2864,
+      "step": 9091
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.453332221202769e-06,
+      "loss": 0.276,
+      "step": 9092
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.451664025356578e-06,
+      "loss": 0.2708,
+      "step": 9093
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.449995829510385e-06,
+      "loss": 0.2544,
+      "step": 9094
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.448327633664192e-06,
+      "loss": 0.2878,
+      "step": 9095
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.446659437818e-06,
+      "loss": 0.2681,
+      "step": 9096
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.444991241971808e-06,
+      "loss": 0.2817,
+      "step": 9097
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.443323046125616e-06,
+      "loss": 0.275,
+      "step": 9098
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.441654850279423e-06,
+      "loss": 0.2657,
+      "step": 9099
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.439986654433232e-06,
+      "loss": 0.2728,
+      "step": 9100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.438318458587039e-06,
+      "loss": 0.2714,
+      "step": 9101
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.436650262740846e-06,
+      "loss": 0.2779,
+      "step": 9102
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.434982066894654e-06,
+      "loss": 0.277,
+      "step": 9103
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4333138710484615e-06,
+      "loss": 0.2713,
+      "step": 9104
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.431645675202269e-06,
+      "loss": 0.2637,
+      "step": 9105
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.429977479356076e-06,
+      "loss": 0.2824,
+      "step": 9106
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.428309283509885e-06,
+      "loss": 0.2711,
+      "step": 9107
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.426641087663692e-06,
+      "loss": 0.2703,
+      "step": 9108
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.424972891817499e-06,
+      "loss": 0.2664,
+      "step": 9109
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.423304695971308e-06,
+      "loss": 0.2778,
+      "step": 9110
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.421636500125115e-06,
+      "loss": 0.2825,
+      "step": 9111
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.419968304278923e-06,
+      "loss": 0.2829,
+      "step": 9112
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.41830010843273e-06,
+      "loss": 0.2642,
+      "step": 9113
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.416631912586539e-06,
+      "loss": 0.2645,
+      "step": 9114
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.414963716740346e-06,
+      "loss": 0.2768,
+      "step": 9115
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.413295520894153e-06,
+      "loss": 0.2718,
+      "step": 9116
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4116273250479615e-06,
+      "loss": 0.248,
+      "step": 9117
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.4099591292017685e-06,
+      "loss": 0.2606,
+      "step": 9118
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.408290933355576e-06,
+      "loss": 0.2633,
+      "step": 9119
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.406622737509383e-06,
+      "loss": 0.2774,
+      "step": 9120
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.404954541663192e-06,
+      "loss": 0.2648,
+      "step": 9121
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.403286345816999e-06,
+      "loss": 0.286,
+      "step": 9122
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.401618149970806e-06,
+      "loss": 0.2895,
+      "step": 9123
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.399949954124615e-06,
+      "loss": 0.2565,
+      "step": 9124
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.398281758278422e-06,
+      "loss": 0.2755,
+      "step": 9125
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.39661356243223e-06,
+      "loss": 0.2685,
+      "step": 9126
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.394945366586038e-06,
+      "loss": 0.2996,
+      "step": 9127
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.393277170739846e-06,
+      "loss": 0.2624,
+      "step": 9128
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.391608974893653e-06,
+      "loss": 0.2483,
+      "step": 9129
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3899407790474614e-06,
+      "loss": 0.2557,
+      "step": 9130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3882725832012685e-06,
+      "loss": 0.2699,
+      "step": 9131
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3866043873550755e-06,
+      "loss": 0.2815,
+      "step": 9132
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.384936191508883e-06,
+      "loss": 0.2646,
+      "step": 9133
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.383267995662691e-06,
+      "loss": 0.2679,
+      "step": 9134
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.381599799816499e-06,
+      "loss": 0.2825,
+      "step": 9135
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.379931603970306e-06,
+      "loss": 0.2752,
+      "step": 9136
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.378263408124115e-06,
+      "loss": 0.2847,
+      "step": 9137
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.376595212277922e-06,
+      "loss": 0.2977,
+      "step": 9138
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.374927016431729e-06,
+      "loss": 0.2883,
+      "step": 9139
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.373258820585538e-06,
+      "loss": 0.283,
+      "step": 9140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.371590624739345e-06,
+      "loss": 0.2779,
+      "step": 9141
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.369922428893153e-06,
+      "loss": 0.2669,
+      "step": 9142
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.36825423304696e-06,
+      "loss": 0.2789,
+      "step": 9143
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3665860372007685e-06,
+      "loss": 0.2468,
+      "step": 9144
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3649178413545755e-06,
+      "loss": 0.2613,
+      "step": 9145
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3632496455083826e-06,
+      "loss": 0.2981,
+      "step": 9146
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.361581449662191e-06,
+      "loss": 0.2853,
+      "step": 9147
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.359913253815998e-06,
+      "loss": 0.2751,
+      "step": 9148
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.358245057969806e-06,
+      "loss": 0.2865,
+      "step": 9149
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.356576862123613e-06,
+      "loss": 0.2633,
+      "step": 9150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.354908666277422e-06,
+      "loss": 0.2905,
+      "step": 9151
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.353240470431229e-06,
+      "loss": 0.2748,
+      "step": 9152
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.351572274585036e-06,
+      "loss": 0.2825,
+      "step": 9153
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.349904078738845e-06,
+      "loss": 0.2809,
+      "step": 9154
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.348235882892652e-06,
+      "loss": 0.2578,
+      "step": 9155
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.34656768704646e-06,
+      "loss": 0.2759,
+      "step": 9156
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.344899491200267e-06,
+      "loss": 0.2693,
+      "step": 9157
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3432312953540755e-06,
+      "loss": 0.2932,
+      "step": 9158
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.3415630995078825e-06,
+      "loss": 0.2662,
+      "step": 9159
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.33989490366169e-06,
+      "loss": 0.2582,
+      "step": 9160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.338226707815498e-06,
+      "loss": 0.2617,
+      "step": 9161
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.336558511969305e-06,
+      "loss": 0.2835,
+      "step": 9162
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.334890316123113e-06,
+      "loss": 0.2762,
+      "step": 9163
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.333222120276921e-06,
+      "loss": 0.2795,
+      "step": 9164
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.331553924430729e-06,
+      "loss": 0.2781,
+      "step": 9165
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.329885728584536e-06,
+      "loss": 0.2768,
+      "step": 9166
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.328217532738343e-06,
+      "loss": 0.2782,
+      "step": 9167
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.326549336892152e-06,
+      "loss": 0.2666,
+      "step": 9168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.324881141045959e-06,
+      "loss": 0.2774,
+      "step": 9169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.323212945199767e-06,
+      "loss": 0.2757,
+      "step": 9170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.321544749353575e-06,
+      "loss": 0.2861,
+      "step": 9171
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.3198765535073825e-06,
+      "loss": 0.2623,
+      "step": 9172
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.3182083576611896e-06,
+      "loss": 0.2567,
+      "step": 9173
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.316540161814997e-06,
+      "loss": 0.2744,
+      "step": 9174
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.314871965968805e-06,
+      "loss": 0.2715,
+      "step": 9175
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.313203770122612e-06,
+      "loss": 0.2856,
+      "step": 9176
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.311535574276421e-06,
+      "loss": 0.2772,
+      "step": 9177
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.309867378430228e-06,
+      "loss": 0.2755,
+      "step": 9178
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.308199182584036e-06,
+      "loss": 0.2841,
+      "step": 9179
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.306530986737843e-06,
+      "loss": 0.2848,
+      "step": 9180
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.304862790891651e-06,
+      "loss": 0.2722,
+      "step": 9181
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.303194595045459e-06,
+      "loss": 0.2866,
+      "step": 9182
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.301526399199266e-06,
+      "loss": 0.2758,
+      "step": 9183
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.299858203353075e-06,
+      "loss": 0.272,
+      "step": 9184
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.298190007506882e-06,
+      "loss": 0.2649,
+      "step": 9185
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2965218116606896e-06,
+      "loss": 0.2871,
+      "step": 9186
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.294853615814497e-06,
+      "loss": 0.2543,
+      "step": 9187
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.293185419968305e-06,
+      "loss": 0.2593,
+      "step": 9188
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.291517224122112e-06,
+      "loss": 0.2511,
+      "step": 9189
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.289849028275919e-06,
+      "loss": 0.2458,
+      "step": 9190
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.288180832429728e-06,
+      "loss": 0.2779,
+      "step": 9191
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.286512636583535e-06,
+      "loss": 0.2658,
+      "step": 9192
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.284844440737343e-06,
+      "loss": 0.2759,
+      "step": 9193
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.283176244891151e-06,
+      "loss": 0.2512,
+      "step": 9194
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.281508049044959e-06,
+      "loss": 0.2735,
+      "step": 9195
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.279839853198766e-06,
+      "loss": 0.2874,
+      "step": 9196
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.278171657352573e-06,
+      "loss": 0.2615,
+      "step": 9197
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.276503461506382e-06,
+      "loss": 0.2803,
+      "step": 9198
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.274835265660189e-06,
+      "loss": 0.258,
+      "step": 9199
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.273167069813997e-06,
+      "loss": 0.2862,
+      "step": 9200
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2714988739678045e-06,
+      "loss": 0.2698,
+      "step": 9201
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.269830678121612e-06,
+      "loss": 0.2487,
+      "step": 9202
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.268162482275419e-06,
+      "loss": 0.2696,
+      "step": 9203
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.2664942864292264e-06,
+      "loss": 0.2522,
+      "step": 9204
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.264826090583035e-06,
+      "loss": 0.2921,
+      "step": 9205
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 0.2674,
+      "step": 9206
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.26148969889065e-06,
+      "loss": 0.2903,
+      "step": 9207
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.259821503044458e-06,
+      "loss": 0.287,
+      "step": 9208
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.258153307198266e-06,
+      "loss": 0.2801,
+      "step": 9209
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.256485111352073e-06,
+      "loss": 0.2738,
+      "step": 9210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.25481691550588e-06,
+      "loss": 0.2717,
+      "step": 9211
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.253148719659689e-06,
+      "loss": 0.2814,
+      "step": 9212
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.251480523813496e-06,
+      "loss": 0.2821,
+      "step": 9213
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2498123279673045e-06,
+      "loss": 0.2559,
+      "step": 9214
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2481441321211115e-06,
+      "loss": 0.2895,
+      "step": 9215
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.246475936274919e-06,
+      "loss": 0.2826,
+      "step": 9216
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2448077404287264e-06,
+      "loss": 0.2738,
+      "step": 9217
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.243139544582534e-06,
+      "loss": 0.2878,
+      "step": 9218
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.241471348736342e-06,
+      "loss": 0.2608,
+      "step": 9219
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.239803152890149e-06,
+      "loss": 0.2825,
+      "step": 9220
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.238134957043958e-06,
+      "loss": 0.2502,
+      "step": 9221
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.236466761197765e-06,
+      "loss": 0.2924,
+      "step": 9222
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.234798565351573e-06,
+      "loss": 0.2764,
+      "step": 9223
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.23313036950538e-06,
+      "loss": 0.2777,
+      "step": 9224
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.231462173659188e-06,
+      "loss": 0.2705,
+      "step": 9225
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.229793977812996e-06,
+      "loss": 0.2647,
+      "step": 9226
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.228125781966803e-06,
+      "loss": 0.2674,
+      "step": 9227
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2264575861206115e-06,
+      "loss": 0.2749,
+      "step": 9228
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2247893902744185e-06,
+      "loss": 0.2869,
+      "step": 9229
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2231211944282264e-06,
+      "loss": 0.2926,
+      "step": 9230
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.221452998582034e-06,
+      "loss": 0.2795,
+      "step": 9231
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.219784802735841e-06,
+      "loss": 0.28,
+      "step": 9232
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.218116606889649e-06,
+      "loss": 0.2435,
+      "step": 9233
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.216448411043456e-06,
+      "loss": 0.2832,
+      "step": 9234
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.214780215197265e-06,
+      "loss": 0.2692,
+      "step": 9235
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.213112019351072e-06,
+      "loss": 0.2614,
+      "step": 9236
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.21144382350488e-06,
+      "loss": 0.2786,
+      "step": 9237
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.209775627658688e-06,
+      "loss": 0.2699,
+      "step": 9238
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.208107431812495e-06,
+      "loss": 0.2664,
+      "step": 9239
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.206439235966303e-06,
+      "loss": 0.2681,
+      "step": 9240
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.20477104012011e-06,
+      "loss": 0.2593,
+      "step": 9241
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2031028442739185e-06,
+      "loss": 0.2822,
+      "step": 9242
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.2014346484277256e-06,
+      "loss": 0.2457,
+      "step": 9243
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.199766452581534e-06,
+      "loss": 0.2678,
+      "step": 9244
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.198098256735341e-06,
+      "loss": 0.2813,
+      "step": 9245
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.196430060889149e-06,
+      "loss": 0.2752,
+      "step": 9246
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.194761865042956e-06,
+      "loss": 0.2693,
+      "step": 9247
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.193093669196763e-06,
+      "loss": 0.2712,
+      "step": 9248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.191425473350572e-06,
+      "loss": 0.2658,
+      "step": 9249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.189757277504379e-06,
+      "loss": 0.2669,
+      "step": 9250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.188089081658188e-06,
+      "loss": 0.2733,
+      "step": 9251
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.186420885811995e-06,
+      "loss": 0.2732,
+      "step": 9252
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.184752689965803e-06,
+      "loss": 0.2772,
+      "step": 9253
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.18308449411961e-06,
+      "loss": 0.2866,
+      "step": 9254
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.181416298273418e-06,
+      "loss": 0.2707,
+      "step": 9255
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1797481024272256e-06,
+      "loss": 0.2717,
+      "step": 9256
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.178079906581033e-06,
+      "loss": 0.2835,
+      "step": 9257
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.176411710734841e-06,
+      "loss": 0.2865,
+      "step": 9258
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.174743514888648e-06,
+      "loss": 0.2736,
+      "step": 9259
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.173075319042456e-06,
+      "loss": 0.2747,
+      "step": 9260
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.171407123196263e-06,
+      "loss": 0.2923,
+      "step": 9261
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.169738927350071e-06,
+      "loss": 0.2973,
+      "step": 9262
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.168070731503879e-06,
+      "loss": 0.2898,
+      "step": 9263
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.166402535657686e-06,
+      "loss": 0.2742,
+      "step": 9264
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.164734339811495e-06,
+      "loss": 0.2669,
+      "step": 9265
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.163066143965302e-06,
+      "loss": 0.2811,
+      "step": 9266
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.16139794811911e-06,
+      "loss": 0.2947,
+      "step": 9267
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.159729752272918e-06,
+      "loss": 0.2847,
+      "step": 9268
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.158061556426725e-06,
+      "loss": 0.267,
+      "step": 9269
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.156393360580533e-06,
+      "loss": 0.2794,
+      "step": 9270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.15472516473434e-06,
+      "loss": 0.2604,
+      "step": 9271
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.153056968888148e-06,
+      "loss": 0.2651,
+      "step": 9272
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.151388773041955e-06,
+      "loss": 0.2546,
+      "step": 9273
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.149720577195763e-06,
+      "loss": 0.2644,
+      "step": 9274
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.148052381349571e-06,
+      "loss": 0.2951,
+      "step": 9275
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.146384185503378e-06,
+      "loss": 0.2773,
+      "step": 9276
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.144715989657186e-06,
+      "loss": 0.2719,
+      "step": 9277
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.143047793810993e-06,
+      "loss": 0.2673,
+      "step": 9278
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.141379597964802e-06,
+      "loss": 0.2694,
+      "step": 9279
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.139711402118609e-06,
+      "loss": 0.2726,
+      "step": 9280
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.138043206272418e-06,
+      "loss": 0.2864,
+      "step": 9281
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.136375010426225e-06,
+      "loss": 0.2672,
+      "step": 9282
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.134706814580032e-06,
+      "loss": 0.2605,
+      "step": 9283
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.13303861873384e-06,
+      "loss": 0.2738,
+      "step": 9284
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.131370422887647e-06,
+      "loss": 0.2575,
+      "step": 9285
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.129702227041455e-06,
+      "loss": 0.279,
+      "step": 9286
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.1280340311952624e-06,
+      "loss": 0.2793,
+      "step": 9287
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.126365835349071e-06,
+      "loss": 0.2695,
+      "step": 9288
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.124697639502878e-06,
+      "loss": 0.2672,
+      "step": 9289
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.123029443656685e-06,
+      "loss": 0.2519,
+      "step": 9290
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.121361247810493e-06,
+      "loss": 0.278,
+      "step": 9291
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.119693051964301e-06,
+      "loss": 0.2659,
+      "step": 9292
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.118024856118109e-06,
+      "loss": 0.2766,
+      "step": 9293
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.116356660271916e-06,
+      "loss": 0.277,
+      "step": 9294
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.114688464425725e-06,
+      "loss": 0.2756,
+      "step": 9295
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.113020268579532e-06,
+      "loss": 0.3091,
+      "step": 9296
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.111352072733339e-06,
+      "loss": 0.2854,
+      "step": 9297
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.109683876887147e-06,
+      "loss": 0.2551,
+      "step": 9298
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.1080156810409545e-06,
+      "loss": 0.2928,
+      "step": 9299
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.1063474851947624e-06,
+      "loss": 0.264,
+      "step": 9300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.1046792893485695e-06,
+      "loss": 0.273,
+      "step": 9301
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.103011093502378e-06,
+      "loss": 0.289,
+      "step": 9302
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.101342897656185e-06,
+      "loss": 0.2919,
+      "step": 9303
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.099674701809992e-06,
+      "loss": 0.2463,
+      "step": 9304
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.098006505963801e-06,
+      "loss": 0.294,
+      "step": 9305
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.096338310117608e-06,
+      "loss": 0.2639,
+      "step": 9306
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.094670114271416e-06,
+      "loss": 0.2782,
+      "step": 9307
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.093001918425223e-06,
+      "loss": 0.2478,
+      "step": 9308
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.091333722579032e-06,
+      "loss": 0.2714,
+      "step": 9309
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.089665526732839e-06,
+      "loss": 0.2586,
+      "step": 9310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.087997330886647e-06,
+      "loss": 0.2731,
+      "step": 9311
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0863291350404545e-06,
+      "loss": 0.2825,
+      "step": 9312
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0846609391942616e-06,
+      "loss": 0.2818,
+      "step": 9313
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0829927433480695e-06,
+      "loss": 0.2649,
+      "step": 9314
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0813245475018765e-06,
+      "loss": 0.2784,
+      "step": 9315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.079656351655685e-06,
+      "loss": 0.2782,
+      "step": 9316
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.077988155809492e-06,
+      "loss": 0.2708,
+      "step": 9317
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.076319959963301e-06,
+      "loss": 0.2705,
+      "step": 9318
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.074651764117108e-06,
+      "loss": 0.281,
+      "step": 9319
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.072983568270915e-06,
+      "loss": 0.2936,
+      "step": 9320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.071315372424723e-06,
+      "loss": 0.2598,
+      "step": 9321
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.069647176578531e-06,
+      "loss": 0.2529,
+      "step": 9322
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.067978980732339e-06,
+      "loss": 0.2852,
+      "step": 9323
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.066310784886146e-06,
+      "loss": 0.2628,
+      "step": 9324
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0646425890399545e-06,
+      "loss": 0.2757,
+      "step": 9325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0629743931937616e-06,
+      "loss": 0.2836,
+      "step": 9326
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.061306197347569e-06,
+      "loss": 0.2753,
+      "step": 9327
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.0596380015013765e-06,
+      "loss": 0.2801,
+      "step": 9328
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.057969805655184e-06,
+      "loss": 0.2749,
+      "step": 9329
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.056301609808992e-06,
+      "loss": 0.293,
+      "step": 9330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.054633413962799e-06,
+      "loss": 0.2661,
+      "step": 9331
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.052965218116608e-06,
+      "loss": 0.2834,
+      "step": 9332
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.051297022270415e-06,
+      "loss": 0.2877,
+      "step": 9333
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.049628826424222e-06,
+      "loss": 0.2829,
+      "step": 9334
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.04796063057803e-06,
+      "loss": 0.2755,
+      "step": 9335
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.046292434731838e-06,
+      "loss": 0.2785,
+      "step": 9336
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.044624238885646e-06,
+      "loss": 0.2943,
+      "step": 9337
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.042956043039453e-06,
+      "loss": 0.2798,
+      "step": 9338
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0412878471932616e-06,
+      "loss": 0.2867,
+      "step": 9339
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.039619651347069e-06,
+      "loss": 0.2667,
+      "step": 9340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.037951455500876e-06,
+      "loss": 0.2562,
+      "step": 9341
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.036283259654684e-06,
+      "loss": 0.2619,
+      "step": 9342
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.034615063808491e-06,
+      "loss": 0.2642,
+      "step": 9343
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.032946867962299e-06,
+      "loss": 0.2956,
+      "step": 9344
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.031278672116106e-06,
+      "loss": 0.262,
+      "step": 9345
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.029610476269915e-06,
+      "loss": 0.2572,
+      "step": 9346
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.027942280423722e-06,
+      "loss": 0.2612,
+      "step": 9347
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.026274084577529e-06,
+      "loss": 0.2772,
+      "step": 9348
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.024605888731338e-06,
+      "loss": 0.2724,
+      "step": 9349
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.022937692885145e-06,
+      "loss": 0.2857,
+      "step": 9350
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.021269497038953e-06,
+      "loss": 0.2718,
+      "step": 9351
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.01960130119276e-06,
+      "loss": 0.2852,
+      "step": 9352
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.017933105346569e-06,
+      "loss": 0.2765,
+      "step": 9353
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.016264909500376e-06,
+      "loss": 0.2645,
+      "step": 9354
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.014596713654183e-06,
+      "loss": 0.2619,
+      "step": 9355
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.012928517807991e-06,
+      "loss": 0.2765,
+      "step": 9356
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.0112603219617984e-06,
+      "loss": 0.2722,
+      "step": 9357
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.009592126115606e-06,
+      "loss": 0.2755,
+      "step": 9358
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.007923930269414e-06,
+      "loss": 0.2824,
+      "step": 9359
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.006255734423222e-06,
+      "loss": 0.2751,
+      "step": 9360
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.004587538577029e-06,
+      "loss": 0.2641,
+      "step": 9361
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.002919342730836e-06,
+      "loss": 0.2667,
+      "step": 9362
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.001251146884645e-06,
+      "loss": 0.2827,
+      "step": 9363
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.999582951038453e-06,
+      "loss": 0.2761,
+      "step": 9364
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.99791475519226e-06,
+      "loss": 0.2637,
+      "step": 9365
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.996246559346068e-06,
+      "loss": 0.2711,
+      "step": 9366
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.994578363499875e-06,
+      "loss": 0.2659,
+      "step": 9367
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.992910167653683e-06,
+      "loss": 0.2758,
+      "step": 9368
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.9912419718074905e-06,
+      "loss": 0.276,
+      "step": 9369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.9895737759612984e-06,
+      "loss": 0.2659,
+      "step": 9370
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.987905580115106e-06,
+      "loss": 0.2813,
+      "step": 9371
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.986237384268914e-06,
+      "loss": 0.2543,
+      "step": 9372
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.984569188422721e-06,
+      "loss": 0.253,
+      "step": 9373
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.982900992576528e-06,
+      "loss": 0.2807,
+      "step": 9374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.981232796730336e-06,
+      "loss": 0.2807,
+      "step": 9375
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.979564600884144e-06,
+      "loss": 0.2814,
+      "step": 9376
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.977896405037952e-06,
+      "loss": 0.2857,
+      "step": 9377
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.97622820919176e-06,
+      "loss": 0.2799,
+      "step": 9378
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.974560013345568e-06,
+      "loss": 0.2391,
+      "step": 9379
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.972891817499375e-06,
+      "loss": 0.259,
+      "step": 9380
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.971223621653183e-06,
+      "loss": 0.283,
+      "step": 9381
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.96955542580699e-06,
+      "loss": 0.2646,
+      "step": 9382
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9678872299607976e-06,
+      "loss": 0.2654,
+      "step": 9383
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9662190341146055e-06,
+      "loss": 0.2841,
+      "step": 9384
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.964550838268413e-06,
+      "loss": 0.2679,
+      "step": 9385
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.962882642422221e-06,
+      "loss": 0.2748,
+      "step": 9386
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.961214446576028e-06,
+      "loss": 0.2772,
+      "step": 9387
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.959546250729836e-06,
+      "loss": 0.2753,
+      "step": 9388
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.957878054883643e-06,
+      "loss": 0.2845,
+      "step": 9389
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.956209859037451e-06,
+      "loss": 0.2802,
+      "step": 9390
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.954541663191259e-06,
+      "loss": 0.2785,
+      "step": 9391
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.952873467345067e-06,
+      "loss": 0.2724,
+      "step": 9392
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.951205271498875e-06,
+      "loss": 0.2813,
+      "step": 9393
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.949537075652683e-06,
+      "loss": 0.2771,
+      "step": 9394
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.94786887980649e-06,
+      "loss": 0.2885,
+      "step": 9395
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9462006839602976e-06,
+      "loss": 0.2629,
+      "step": 9396
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.944532488114105e-06,
+      "loss": 0.278,
+      "step": 9397
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9428642922679125e-06,
+      "loss": 0.2413,
+      "step": 9398
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.94119609642172e-06,
+      "loss": 0.2826,
+      "step": 9399
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.939527900575528e-06,
+      "loss": 0.2752,
+      "step": 9400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.937859704729336e-06,
+      "loss": 0.281,
+      "step": 9401
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.936191508883143e-06,
+      "loss": 0.2889,
+      "step": 9402
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.934523313036951e-06,
+      "loss": 0.2701,
+      "step": 9403
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.932855117190758e-06,
+      "loss": 0.2753,
+      "step": 9404
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.931186921344566e-06,
+      "loss": 0.2818,
+      "step": 9405
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.929518725498374e-06,
+      "loss": 0.2852,
+      "step": 9406
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.927850529652182e-06,
+      "loss": 0.2781,
+      "step": 9407
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.92618233380599e-06,
+      "loss": 0.2714,
+      "step": 9408
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9245141379597976e-06,
+      "loss": 0.2682,
+      "step": 9409
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.922845942113605e-06,
+      "loss": 0.2629,
+      "step": 9410
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9211777462674125e-06,
+      "loss": 0.2778,
+      "step": 9411
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.9195095504212195e-06,
+      "loss": 0.2461,
+      "step": 9412
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.917841354575027e-06,
+      "loss": 0.2904,
+      "step": 9413
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.916173158728835e-06,
+      "loss": 0.2553,
+      "step": 9414
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 4.914504962882643e-06,
+      "loss": 0.2717,
+      "step": 9415
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.912836767036451e-06,
+      "loss": 0.2834,
+      "step": 9416
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.911168571190258e-06,
+      "loss": 0.2711,
+      "step": 9417
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.909500375344066e-06,
+      "loss": 0.2746,
+      "step": 9418
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.907832179497873e-06,
+      "loss": 0.2701,
+      "step": 9419
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.906163983651681e-06,
+      "loss": 0.2722,
+      "step": 9420
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.904495787805489e-06,
+      "loss": 0.2635,
+      "step": 9421
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.902827591959297e-06,
+      "loss": 0.2826,
+      "step": 9422
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.901159396113105e-06,
+      "loss": 0.279,
+      "step": 9423
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.899491200266912e-06,
+      "loss": 0.2677,
+      "step": 9424
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8978230044207195e-06,
+      "loss": 0.2689,
+      "step": 9425
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8961548085745266e-06,
+      "loss": 0.2952,
+      "step": 9426
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8944866127283344e-06,
+      "loss": 0.2825,
+      "step": 9427
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.892818416882142e-06,
+      "loss": 0.2834,
+      "step": 9428
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.89115022103595e-06,
+      "loss": 0.2679,
+      "step": 9429
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.889482025189758e-06,
+      "loss": 0.2752,
+      "step": 9430
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.887813829343565e-06,
+      "loss": 0.2787,
+      "step": 9431
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.886145633497373e-06,
+      "loss": 0.2739,
+      "step": 9432
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.884477437651181e-06,
+      "loss": 0.2707,
+      "step": 9433
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.882809241804988e-06,
+      "loss": 0.2856,
+      "step": 9434
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.881141045958796e-06,
+      "loss": 0.2861,
+      "step": 9435
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.879472850112604e-06,
+      "loss": 0.2559,
+      "step": 9436
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.877804654266412e-06,
+      "loss": 0.2699,
+      "step": 9437
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8761364584202195e-06,
+      "loss": 0.286,
+      "step": 9438
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8744682625740265e-06,
+      "loss": 0.2797,
+      "step": 9439
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8728000667278344e-06,
+      "loss": 0.285,
+      "step": 9440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8711318708816415e-06,
+      "loss": 0.2615,
+      "step": 9441
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.869463675035449e-06,
+      "loss": 0.2877,
+      "step": 9442
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.867795479189257e-06,
+      "loss": 0.2706,
+      "step": 9443
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.866127283343065e-06,
+      "loss": 0.2667,
+      "step": 9444
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.864459087496873e-06,
+      "loss": 0.2671,
+      "step": 9445
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.86279089165068e-06,
+      "loss": 0.2761,
+      "step": 9446
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.861122695804488e-06,
+      "loss": 0.2973,
+      "step": 9447
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.859454499958296e-06,
+      "loss": 0.2891,
+      "step": 9448
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.857786304112103e-06,
+      "loss": 0.2756,
+      "step": 9449
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.856118108265911e-06,
+      "loss": 0.2689,
+      "step": 9450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.854449912419719e-06,
+      "loss": 0.2697,
+      "step": 9451
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8527817165735265e-06,
+      "loss": 0.2726,
+      "step": 9452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.851113520727334e-06,
+      "loss": 0.2691,
+      "step": 9453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.8494453248811415e-06,
+      "loss": 0.2643,
+      "step": 9454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.847777129034949e-06,
+      "loss": 0.2729,
+      "step": 9455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.846108933188756e-06,
+      "loss": 0.2643,
+      "step": 9456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.844440737342564e-06,
+      "loss": 0.2699,
+      "step": 9457
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.842772541496372e-06,
+      "loss": 0.2681,
+      "step": 9458
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.84110434565018e-06,
+      "loss": 0.2891,
+      "step": 9459
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.839436149803987e-06,
+      "loss": 0.2708,
+      "step": 9460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.837767953957795e-06,
+      "loss": 0.2689,
+      "step": 9461
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.836099758111603e-06,
+      "loss": 0.2794,
+      "step": 9462
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.83443156226541e-06,
+      "loss": 0.2653,
+      "step": 9463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.832763366419218e-06,
+      "loss": 0.2711,
+      "step": 9464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.831095170573026e-06,
+      "loss": 0.2774,
+      "step": 9465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8294269747268336e-06,
+      "loss": 0.2719,
+      "step": 9466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.827758778880641e-06,
+      "loss": 0.2989,
+      "step": 9467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8260905830344485e-06,
+      "loss": 0.2746,
+      "step": 9468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.824422387188256e-06,
+      "loss": 0.2749,
+      "step": 9469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.822754191342064e-06,
+      "loss": 0.2748,
+      "step": 9470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.821085995495871e-06,
+      "loss": 0.2633,
+      "step": 9471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.819417799649679e-06,
+      "loss": 0.2953,
+      "step": 9472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.817749603803487e-06,
+      "loss": 0.283,
+      "step": 9473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.816081407957295e-06,
+      "loss": 0.2649,
+      "step": 9474
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.814413212111102e-06,
+      "loss": 0.2776,
+      "step": 9475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.81274501626491e-06,
+      "loss": 0.2772,
+      "step": 9476
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.811076820418718e-06,
+      "loss": 0.2775,
+      "step": 9477
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.809408624572525e-06,
+      "loss": 0.2732,
+      "step": 9478
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.807740428726333e-06,
+      "loss": 0.2773,
+      "step": 9479
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.806072232880141e-06,
+      "loss": 0.2706,
+      "step": 9480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8044040370339485e-06,
+      "loss": 0.2699,
+      "step": 9481
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.8027358411877555e-06,
+      "loss": 0.2607,
+      "step": 9482
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.801067645341563e-06,
+      "loss": 0.2943,
+      "step": 9483
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.799399449495371e-06,
+      "loss": 0.298,
+      "step": 9484
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.797731253649179e-06,
+      "loss": 0.2661,
+      "step": 9485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.796063057802986e-06,
+      "loss": 0.2723,
+      "step": 9486
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.794394861956794e-06,
+      "loss": 0.2665,
+      "step": 9487
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.792726666110602e-06,
+      "loss": 0.2743,
+      "step": 9488
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.791058470264409e-06,
+      "loss": 0.2745,
+      "step": 9489
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.789390274418217e-06,
+      "loss": 0.2847,
+      "step": 9490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.787722078572025e-06,
+      "loss": 0.2877,
+      "step": 9491
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.786053882725833e-06,
+      "loss": 0.2913,
+      "step": 9492
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.78438568687964e-06,
+      "loss": 0.2537,
+      "step": 9493
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.782717491033448e-06,
+      "loss": 0.2644,
+      "step": 9494
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.7810492951872555e-06,
+      "loss": 0.2498,
+      "step": 9495
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.7793810993410626e-06,
+      "loss": 0.2792,
+      "step": 9496
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.7777129034948704e-06,
+      "loss": 0.282,
+      "step": 9497
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.776044707648678e-06,
+      "loss": 0.2739,
+      "step": 9498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.774376511802486e-06,
+      "loss": 0.2697,
+      "step": 9499
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.772708315956294e-06,
+      "loss": 0.2867,
+      "step": 9500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.771040120110101e-06,
+      "loss": 0.2529,
+      "step": 9501
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.769371924263909e-06,
+      "loss": 0.2851,
+      "step": 9502
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.767703728417717e-06,
+      "loss": 0.2515,
+      "step": 9503
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.766035532571524e-06,
+      "loss": 0.2688,
+      "step": 9504
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.764367336725332e-06,
+      "loss": 0.2616,
+      "step": 9505
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.76269914087914e-06,
+      "loss": 0.2471,
+      "step": 9506
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.761030945032948e-06,
+      "loss": 0.2807,
+      "step": 9507
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.759362749186755e-06,
+      "loss": 0.272,
+      "step": 9508
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7576945533405625e-06,
+      "loss": 0.2882,
+      "step": 9509
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7560263574943704e-06,
+      "loss": 0.2816,
+      "step": 9510
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7543581616481775e-06,
+      "loss": 0.2797,
+      "step": 9511
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.752689965801985e-06,
+      "loss": 0.2755,
+      "step": 9512
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.751021769955793e-06,
+      "loss": 0.2613,
+      "step": 9513
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.749353574109601e-06,
+      "loss": 0.2706,
+      "step": 9514
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.747685378263408e-06,
+      "loss": 0.2662,
+      "step": 9515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.746017182417216e-06,
+      "loss": 0.2628,
+      "step": 9516
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.744348986571024e-06,
+      "loss": 0.2553,
+      "step": 9517
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.742680790724831e-06,
+      "loss": 0.2597,
+      "step": 9518
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.741012594878639e-06,
+      "loss": 0.2672,
+      "step": 9519
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.739344399032447e-06,
+      "loss": 0.2741,
+      "step": 9520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.737676203186255e-06,
+      "loss": 0.2728,
+      "step": 9521
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7360080073400625e-06,
+      "loss": 0.2554,
+      "step": 9522
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.73433981149387e-06,
+      "loss": 0.2646,
+      "step": 9523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7326716156476775e-06,
+      "loss": 0.2991,
+      "step": 9524
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7310034198014845e-06,
+      "loss": 0.2934,
+      "step": 9525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.729335223955292e-06,
+      "loss": 0.2552,
+      "step": 9526
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7276670281091e-06,
+      "loss": 0.2705,
+      "step": 9527
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.725998832262908e-06,
+      "loss": 0.2847,
+      "step": 9528
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.724330636416716e-06,
+      "loss": 0.2748,
+      "step": 9529
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.722662440570523e-06,
+      "loss": 0.2827,
+      "step": 9530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.720994244724331e-06,
+      "loss": 0.2663,
+      "step": 9531
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.719326048878139e-06,
+      "loss": 0.2523,
+      "step": 9532
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.717657853031946e-06,
+      "loss": 0.2628,
+      "step": 9533
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.715989657185754e-06,
+      "loss": 0.2876,
+      "step": 9534
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.714321461339562e-06,
+      "loss": 0.2587,
+      "step": 9535
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7126532654933696e-06,
+      "loss": 0.2623,
+      "step": 9536
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7109850696471775e-06,
+      "loss": 0.2943,
+      "step": 9537
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.7093168738009845e-06,
+      "loss": 0.2539,
+      "step": 9538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.707648677954792e-06,
+      "loss": 0.293,
+      "step": 9539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.705980482108599e-06,
+      "loss": 0.2751,
+      "step": 9540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.704312286262407e-06,
+      "loss": 0.2702,
+      "step": 9541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.702644090416215e-06,
+      "loss": 0.2787,
+      "step": 9542
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.700975894570023e-06,
+      "loss": 0.2531,
+      "step": 9543
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.699307698723831e-06,
+      "loss": 0.269,
+      "step": 9544
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.697639502877638e-06,
+      "loss": 0.2416,
+      "step": 9545
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.695971307031446e-06,
+      "loss": 0.2887,
+      "step": 9546
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.694303111185253e-06,
+      "loss": 0.2803,
+      "step": 9547
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.692634915339061e-06,
+      "loss": 0.2668,
+      "step": 9548
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.690966719492869e-06,
+      "loss": 0.2741,
+      "step": 9549
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.689298523646677e-06,
+      "loss": 0.2724,
+      "step": 9550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6876303278004845e-06,
+      "loss": 0.2613,
+      "step": 9551
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6859621319542915e-06,
+      "loss": 0.2752,
+      "step": 9552
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.684293936108099e-06,
+      "loss": 0.2763,
+      "step": 9553
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6826257402619065e-06,
+      "loss": 0.2734,
+      "step": 9554
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.680957544415714e-06,
+      "loss": 0.2446,
+      "step": 9555
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.679289348569522e-06,
+      "loss": 0.2737,
+      "step": 9556
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.67762115272333e-06,
+      "loss": 0.2725,
+      "step": 9557
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.675952956877138e-06,
+      "loss": 0.2826,
+      "step": 9558
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.674284761030946e-06,
+      "loss": 0.2639,
+      "step": 9559
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.672616565184753e-06,
+      "loss": 0.2853,
+      "step": 9560
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.670948369338561e-06,
+      "loss": 0.274,
+      "step": 9561
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.669280173492368e-06,
+      "loss": 0.2561,
+      "step": 9562
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.667611977646176e-06,
+      "loss": 0.2639,
+      "step": 9563
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.665943781799984e-06,
+      "loss": 0.2841,
+      "step": 9564
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6642755859537915e-06,
+      "loss": 0.2786,
+      "step": 9565
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.662607390107599e-06,
+      "loss": 0.2701,
+      "step": 9566
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6609391942614064e-06,
+      "loss": 0.2578,
+      "step": 9567
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.659270998415214e-06,
+      "loss": 0.2628,
+      "step": 9568
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.657602802569021e-06,
+      "loss": 0.2716,
+      "step": 9569
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.655934606722829e-06,
+      "loss": 0.2747,
+      "step": 9570
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.654266410876637e-06,
+      "loss": 0.2751,
+      "step": 9571
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.652598215030445e-06,
+      "loss": 0.2669,
+      "step": 9572
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.650930019184253e-06,
+      "loss": 0.2712,
+      "step": 9573
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.649261823338061e-06,
+      "loss": 0.2506,
+      "step": 9574
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.647593627491868e-06,
+      "loss": 0.2621,
+      "step": 9575
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.645925431645676e-06,
+      "loss": 0.3043,
+      "step": 9576
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.644257235799483e-06,
+      "loss": 0.2493,
+      "step": 9577
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.642589039953291e-06,
+      "loss": 0.2507,
+      "step": 9578
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6409208441070985e-06,
+      "loss": 0.2709,
+      "step": 9579
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.6392526482609064e-06,
+      "loss": 0.2771,
+      "step": 9580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.637584452414714e-06,
+      "loss": 0.2758,
+      "step": 9581
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.635916256568521e-06,
+      "loss": 0.273,
+      "step": 9582
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.634248060722329e-06,
+      "loss": 0.2842,
+      "step": 9583
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.632579864876136e-06,
+      "loss": 0.2632,
+      "step": 9584
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.630911669029944e-06,
+      "loss": 0.2607,
+      "step": 9585
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.629243473183752e-06,
+      "loss": 0.265,
+      "step": 9586
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.62757527733756e-06,
+      "loss": 0.2646,
+      "step": 9587
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.625907081491368e-06,
+      "loss": 0.2982,
+      "step": 9588
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.624238885645176e-06,
+      "loss": 0.2761,
+      "step": 9589
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.622570689798983e-06,
+      "loss": 0.2748,
+      "step": 9590
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.62090249395279e-06,
+      "loss": 0.2933,
+      "step": 9591
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.619234298106598e-06,
+      "loss": 0.2748,
+      "step": 9592
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.617566102260406e-06,
+      "loss": 0.2565,
+      "step": 9593
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.6158979064142135e-06,
+      "loss": 0.276,
+      "step": 9594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.614229710568021e-06,
+      "loss": 0.2787,
+      "step": 9595
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.612561514721829e-06,
+      "loss": 0.2589,
+      "step": 9596
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.610893318875636e-06,
+      "loss": 0.2674,
+      "step": 9597
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.609225123029444e-06,
+      "loss": 0.2907,
+      "step": 9598
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.607556927183251e-06,
+      "loss": 0.2606,
+      "step": 9599
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.605888731337059e-06,
+      "loss": 0.274,
+      "step": 9600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.604220535490867e-06,
+      "loss": 0.2733,
+      "step": 9601
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.602552339644675e-06,
+      "loss": 0.2654,
+      "step": 9602
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.600884143798483e-06,
+      "loss": 0.2764,
+      "step": 9603
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.59921594795229e-06,
+      "loss": 0.2689,
+      "step": 9604
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.597547752106098e-06,
+      "loss": 0.2887,
+      "step": 9605
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.595879556259905e-06,
+      "loss": 0.2789,
+      "step": 9606
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.594211360413713e-06,
+      "loss": 0.3082,
+      "step": 9607
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.5925431645675205e-06,
+      "loss": 0.2634,
+      "step": 9608
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.590874968721328e-06,
+      "loss": 0.265,
+      "step": 9609
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.589206772875136e-06,
+      "loss": 0.2613,
+      "step": 9610
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.587538577028944e-06,
+      "loss": 0.2644,
+      "step": 9611
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.585870381182751e-06,
+      "loss": 0.2917,
+      "step": 9612
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.584202185336559e-06,
+      "loss": 0.2722,
+      "step": 9613
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.582533989490366e-06,
+      "loss": 0.2519,
+      "step": 9614
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.580865793644174e-06,
+      "loss": 0.2923,
+      "step": 9615
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.579197597797982e-06,
+      "loss": 0.2678,
+      "step": 9616
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.57752940195179e-06,
+      "loss": 0.2953,
+      "step": 9617
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.575861206105598e-06,
+      "loss": 0.2617,
+      "step": 9618
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.574193010259405e-06,
+      "loss": 0.2828,
+      "step": 9619
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.572524814413213e-06,
+      "loss": 0.287,
+      "step": 9620
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.57085661856702e-06,
+      "loss": 0.2747,
+      "step": 9621
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5691884227208275e-06,
+      "loss": 0.2681,
+      "step": 9622
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.567520226874635e-06,
+      "loss": 0.2552,
+      "step": 9623
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.565852031028443e-06,
+      "loss": 0.2737,
+      "step": 9624
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.564183835182251e-06,
+      "loss": 0.2705,
+      "step": 9625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.562515639336059e-06,
+      "loss": 0.2829,
+      "step": 9626
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.560847443489866e-06,
+      "loss": 0.2588,
+      "step": 9627
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.559179247643674e-06,
+      "loss": 0.2686,
+      "step": 9628
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.557511051797481e-06,
+      "loss": 0.2602,
+      "step": 9629
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.555842855951289e-06,
+      "loss": 0.2728,
+      "step": 9630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.554174660105097e-06,
+      "loss": 0.2607,
+      "step": 9631
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.552506464258905e-06,
+      "loss": 0.2822,
+      "step": 9632
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.550838268412713e-06,
+      "loss": 0.2864,
+      "step": 9633
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.54917007256652e-06,
+      "loss": 0.2765,
+      "step": 9634
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5475018767203275e-06,
+      "loss": 0.2719,
+      "step": 9635
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5458336808741346e-06,
+      "loss": 0.2935,
+      "step": 9636
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5441654850279424e-06,
+      "loss": 0.2471,
+      "step": 9637
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.54249728918175e-06,
+      "loss": 0.2804,
+      "step": 9638
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.540829093335558e-06,
+      "loss": 0.2754,
+      "step": 9639
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.539160897489366e-06,
+      "loss": 0.2928,
+      "step": 9640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.537492701643173e-06,
+      "loss": 0.2874,
+      "step": 9641
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.535824505796981e-06,
+      "loss": 0.2637,
+      "step": 9642
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.534156309950788e-06,
+      "loss": 0.2844,
+      "step": 9643
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.532488114104596e-06,
+      "loss": 0.2649,
+      "step": 9644
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.530819918258404e-06,
+      "loss": 0.2648,
+      "step": 9645
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.529151722412212e-06,
+      "loss": 0.2576,
+      "step": 9646
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.52748352656602e-06,
+      "loss": 0.2612,
+      "step": 9647
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5258153307198275e-06,
+      "loss": 0.2686,
+      "step": 9648
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5241471348736345e-06,
+      "loss": 0.2588,
+      "step": 9649
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5224789390274424e-06,
+      "loss": 0.2606,
+      "step": 9650
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5208107431812495e-06,
+      "loss": 0.2638,
+      "step": 9651
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.519142547335057e-06,
+      "loss": 0.2615,
+      "step": 9652
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.517474351488865e-06,
+      "loss": 0.2678,
+      "step": 9653
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.515806155642673e-06,
+      "loss": 0.2755,
+      "step": 9654
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.514137959796481e-06,
+      "loss": 0.2539,
+      "step": 9655
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.512469763950288e-06,
+      "loss": 0.2627,
+      "step": 9656
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.510801568104096e-06,
+      "loss": 0.2864,
+      "step": 9657
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.509133372257903e-06,
+      "loss": 0.2699,
+      "step": 9658
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.507465176411711e-06,
+      "loss": 0.2641,
+      "step": 9659
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.505796980565519e-06,
+      "loss": 0.259,
+      "step": 9660
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.504128784719327e-06,
+      "loss": 0.2613,
+      "step": 9661
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.5024605888731345e-06,
+      "loss": 0.2735,
+      "step": 9662
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.500792393026942e-06,
+      "loss": 0.2649,
+      "step": 9663
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4991241971807495e-06,
+      "loss": 0.2743,
+      "step": 9664
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.497456001334557e-06,
+      "loss": 0.257,
+      "step": 9665
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.495787805488364e-06,
+      "loss": 0.2623,
+      "step": 9666
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.494119609642172e-06,
+      "loss": 0.2639,
+      "step": 9667
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.49245141379598e-06,
+      "loss": 0.3004,
+      "step": 9668
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.490783217949788e-06,
+      "loss": 0.2579,
+      "step": 9669
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.489115022103595e-06,
+      "loss": 0.2851,
+      "step": 9670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.487446826257403e-06,
+      "loss": 0.2546,
+      "step": 9671
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.485778630411211e-06,
+      "loss": 0.2618,
+      "step": 9672
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.484110434565018e-06,
+      "loss": 0.2384,
+      "step": 9673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.482442238718826e-06,
+      "loss": 0.2598,
+      "step": 9674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.480774042872634e-06,
+      "loss": 0.27,
+      "step": 9675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4791058470264416e-06,
+      "loss": 0.2626,
+      "step": 9676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4774376511802495e-06,
+      "loss": 0.2497,
+      "step": 9677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4757694553340565e-06,
+      "loss": 0.2967,
+      "step": 9678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.474101259487864e-06,
+      "loss": 0.2759,
+      "step": 9679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4724330636416714e-06,
+      "loss": 0.2981,
+      "step": 9680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.470764867795479e-06,
+      "loss": 0.2781,
+      "step": 9681
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.469096671949287e-06,
+      "loss": 0.2748,
+      "step": 9682
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.467428476103095e-06,
+      "loss": 0.2582,
+      "step": 9683
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.465760280256903e-06,
+      "loss": 0.2881,
+      "step": 9684
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.46409208441071e-06,
+      "loss": 0.2708,
+      "step": 9685
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.462423888564518e-06,
+      "loss": 0.2651,
+      "step": 9686
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.460755692718326e-06,
+      "loss": 0.2743,
+      "step": 9687
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.459087496872133e-06,
+      "loss": 0.2924,
+      "step": 9688
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.457419301025941e-06,
+      "loss": 0.2771,
+      "step": 9689
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.455751105179749e-06,
+      "loss": 0.267,
+      "step": 9690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4540829093335565e-06,
+      "loss": 0.2867,
+      "step": 9691
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.4524147134873635e-06,
+      "loss": 0.2565,
+      "step": 9692
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.450746517641171e-06,
+      "loss": 0.2878,
+      "step": 9693
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.449078321794979e-06,
+      "loss": 0.269,
+      "step": 9694
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.447410125948786e-06,
+      "loss": 0.2634,
+      "step": 9695
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.445741930102594e-06,
+      "loss": 0.2621,
+      "step": 9696
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.444073734256402e-06,
+      "loss": 0.2491,
+      "step": 9697
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.44240553841021e-06,
+      "loss": 0.2776,
+      "step": 9698
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.440737342564017e-06,
+      "loss": 0.2893,
+      "step": 9699
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.439069146717825e-06,
+      "loss": 0.285,
+      "step": 9700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.437400950871633e-06,
+      "loss": 0.2644,
+      "step": 9701
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.435732755025441e-06,
+      "loss": 0.2522,
+      "step": 9702
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 4.434064559179248e-06,
+      "loss": 0.2794,
+      "step": 9703
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.432396363333056e-06,
+      "loss": 0.2687,
+      "step": 9704
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4307281674868635e-06,
+      "loss": 0.2852,
+      "step": 9705
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.429059971640671e-06,
+      "loss": 0.2628,
+      "step": 9706
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4273917757944784e-06,
+      "loss": 0.2682,
+      "step": 9707
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.425723579948286e-06,
+      "loss": 0.297,
+      "step": 9708
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.424055384102094e-06,
+      "loss": 0.2794,
+      "step": 9709
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.422387188255901e-06,
+      "loss": 0.269,
+      "step": 9710
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.420718992409709e-06,
+      "loss": 0.2772,
+      "step": 9711
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.419050796563517e-06,
+      "loss": 0.2676,
+      "step": 9712
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.417382600717325e-06,
+      "loss": 0.2853,
+      "step": 9713
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.415714404871132e-06,
+      "loss": 0.2699,
+      "step": 9714
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.41404620902494e-06,
+      "loss": 0.2857,
+      "step": 9715
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.412378013178748e-06,
+      "loss": 0.2719,
+      "step": 9716
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.410709817332556e-06,
+      "loss": 0.2678,
+      "step": 9717
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.409041621486363e-06,
+      "loss": 0.2689,
+      "step": 9718
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4073734256401705e-06,
+      "loss": 0.2805,
+      "step": 9719
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4057052297939784e-06,
+      "loss": 0.2765,
+      "step": 9720
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.4040370339477855e-06,
+      "loss": 0.2909,
+      "step": 9721
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.402368838101593e-06,
+      "loss": 0.2606,
+      "step": 9722
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.400700642255401e-06,
+      "loss": 0.2777,
+      "step": 9723
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.399032446409209e-06,
+      "loss": 0.2643,
+      "step": 9724
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.397364250563016e-06,
+      "loss": 0.2875,
+      "step": 9725
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.395696054716824e-06,
+      "loss": 0.268,
+      "step": 9726
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.394027858870632e-06,
+      "loss": 0.2468,
+      "step": 9727
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.392359663024439e-06,
+      "loss": 0.2625,
+      "step": 9728
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.390691467178247e-06,
+      "loss": 0.2947,
+      "step": 9729
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.389023271332055e-06,
+      "loss": 0.2769,
+      "step": 9730
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.387355075485863e-06,
+      "loss": 0.2705,
+      "step": 9731
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.38568687963967e-06,
+      "loss": 0.2688,
+      "step": 9732
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.384018683793478e-06,
+      "loss": 0.2659,
+      "step": 9733
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3823504879472855e-06,
+      "loss": 0.2931,
+      "step": 9734
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.380682292101093e-06,
+      "loss": 0.29,
+      "step": 9735
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.3790140962549e-06,
+      "loss": 0.2708,
+      "step": 9736
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.377345900408708e-06,
+      "loss": 0.2725,
+      "step": 9737
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.375677704562516e-06,
+      "loss": 0.2585,
+      "step": 9738
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.374009508716324e-06,
+      "loss": 0.2719,
+      "step": 9739
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.372341312870131e-06,
+      "loss": 0.2816,
+      "step": 9740
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.370673117023939e-06,
+      "loss": 0.2869,
+      "step": 9741
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.369004921177747e-06,
+      "loss": 0.2696,
+      "step": 9742
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.367336725331554e-06,
+      "loss": 0.2874,
+      "step": 9743
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.365668529485362e-06,
+      "loss": 0.2698,
+      "step": 9744
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.36400033363917e-06,
+      "loss": 0.2617,
+      "step": 9745
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3623321377929776e-06,
+      "loss": 0.2749,
+      "step": 9746
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.360663941946785e-06,
+      "loss": 0.2686,
+      "step": 9747
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3589957461005925e-06,
+      "loss": 0.2772,
+      "step": 9748
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3573275502544e-06,
+      "loss": 0.2741,
+      "step": 9749
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3556593544082074e-06,
+      "loss": 0.2859,
+      "step": 9750
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.353991158562015e-06,
+      "loss": 0.2673,
+      "step": 9751
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.352322962715823e-06,
+      "loss": 0.2657,
+      "step": 9752
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.350654766869631e-06,
+      "loss": 0.2693,
+      "step": 9753
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.348986571023439e-06,
+      "loss": 0.2821,
+      "step": 9754
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.347318375177246e-06,
+      "loss": 0.2918,
+      "step": 9755
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.345650179331054e-06,
+      "loss": 0.2674,
+      "step": 9756
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.343981983484861e-06,
+      "loss": 0.2636,
+      "step": 9757
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.342313787638669e-06,
+      "loss": 0.2627,
+      "step": 9758
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.340645591792477e-06,
+      "loss": 0.2724,
+      "step": 9759
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.338977395946285e-06,
+      "loss": 0.2463,
+      "step": 9760
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3373092001000925e-06,
+      "loss": 0.2649,
+      "step": 9761
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3356410042538995e-06,
+      "loss": 0.2687,
+      "step": 9762
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.333972808407707e-06,
+      "loss": 0.2814,
+      "step": 9763
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3323046125615145e-06,
+      "loss": 0.2784,
+      "step": 9764
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.330636416715322e-06,
+      "loss": 0.2609,
+      "step": 9765
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.32896822086913e-06,
+      "loss": 0.2921,
+      "step": 9766
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.327300025022938e-06,
+      "loss": 0.2603,
+      "step": 9767
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.325631829176746e-06,
+      "loss": 0.2677,
+      "step": 9768
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.323963633330553e-06,
+      "loss": 0.2771,
+      "step": 9769
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.322295437484361e-06,
+      "loss": 0.2754,
+      "step": 9770
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.320627241638169e-06,
+      "loss": 0.2684,
+      "step": 9771
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.318959045791976e-06,
+      "loss": 0.2805,
+      "step": 9772
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.317290849945784e-06,
+      "loss": 0.2807,
+      "step": 9773
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.315622654099592e-06,
+      "loss": 0.2761,
+      "step": 9774
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3139544582533995e-06,
+      "loss": 0.2842,
+      "step": 9775
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.312286262407207e-06,
+      "loss": 0.253,
+      "step": 9776
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.3106180665610144e-06,
+      "loss": 0.282,
+      "step": 9777
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.308949870714822e-06,
+      "loss": 0.2841,
+      "step": 9778
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.307281674868629e-06,
+      "loss": 0.279,
+      "step": 9779
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.305613479022437e-06,
+      "loss": 0.2668,
+      "step": 9780
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.303945283176245e-06,
+      "loss": 0.2906,
+      "step": 9781
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.302277087330053e-06,
+      "loss": 0.2753,
+      "step": 9782
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.300608891483861e-06,
+      "loss": 0.2894,
+      "step": 9783
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.298940695637668e-06,
+      "loss": 0.2883,
+      "step": 9784
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.297272499791476e-06,
+      "loss": 0.2707,
+      "step": 9785
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 4.295604303945283e-06,
+      "loss": 0.2582,
+      "step": 9786
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.293936108099091e-06,
+      "loss": 0.2554,
+      "step": 9787
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.292267912252899e-06,
+      "loss": 0.2625,
+      "step": 9788
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2905997164067066e-06,
+      "loss": 0.2773,
+      "step": 9789
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2889315205605144e-06,
+      "loss": 0.2833,
+      "step": 9790
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.287263324714322e-06,
+      "loss": 0.2754,
+      "step": 9791
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.285595128868129e-06,
+      "loss": 0.2655,
+      "step": 9792
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.283926933021937e-06,
+      "loss": 0.2987,
+      "step": 9793
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.282258737175744e-06,
+      "loss": 0.2717,
+      "step": 9794
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.280590541329552e-06,
+      "loss": 0.2812,
+      "step": 9795
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.27892234548336e-06,
+      "loss": 0.2592,
+      "step": 9796
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.277254149637168e-06,
+      "loss": 0.2785,
+      "step": 9797
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.275585953790976e-06,
+      "loss": 0.2712,
+      "step": 9798
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.273917757944783e-06,
+      "loss": 0.2647,
+      "step": 9799
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.272249562098591e-06,
+      "loss": 0.2603,
+      "step": 9800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.270581366252398e-06,
+      "loss": 0.2753,
+      "step": 9801
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.268913170406206e-06,
+      "loss": 0.2741,
+      "step": 9802
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.267244974560014e-06,
+      "loss": 0.2816,
+      "step": 9803
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2655767787138215e-06,
+      "loss": 0.2748,
+      "step": 9804
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.263908582867629e-06,
+      "loss": 0.2758,
+      "step": 9805
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.262240387021437e-06,
+      "loss": 0.2797,
+      "step": 9806
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.260572191175244e-06,
+      "loss": 0.258,
+      "step": 9807
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.258903995329051e-06,
+      "loss": 0.2738,
+      "step": 9808
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.257235799482859e-06,
+      "loss": 0.2969,
+      "step": 9809
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.255567603636667e-06,
+      "loss": 0.2913,
+      "step": 9810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.253899407790475e-06,
+      "loss": 0.2771,
+      "step": 9811
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.252231211944283e-06,
+      "loss": 0.2913,
+      "step": 9812
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.250563016098091e-06,
+      "loss": 0.2744,
+      "step": 9813
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.248894820251898e-06,
+      "loss": 0.2589,
+      "step": 9814
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.247226624405706e-06,
+      "loss": 0.3032,
+      "step": 9815
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.245558428559513e-06,
+      "loss": 0.279,
+      "step": 9816
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.243890232713321e-06,
+      "loss": 0.2712,
+      "step": 9817
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.2422220368671285e-06,
+      "loss": 0.2542,
+      "step": 9818
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.240553841020936e-06,
+      "loss": 0.2848,
+      "step": 9819
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.238885645174744e-06,
+      "loss": 0.2577,
+      "step": 9820
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.237217449328551e-06,
+      "loss": 0.2696,
+      "step": 9821
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.235549253482359e-06,
+      "loss": 0.2867,
+      "step": 9822
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.233881057636166e-06,
+      "loss": 0.2849,
+      "step": 9823
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.232212861789974e-06,
+      "loss": 0.2809,
+      "step": 9824
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.230544665943782e-06,
+      "loss": 0.2612,
+      "step": 9825
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.22887647009759e-06,
+      "loss": 0.271,
+      "step": 9826
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.227208274251398e-06,
+      "loss": 0.276,
+      "step": 9827
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.225540078405206e-06,
+      "loss": 0.2552,
+      "step": 9828
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.223871882559013e-06,
+      "loss": 0.2738,
+      "step": 9829
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.222203686712821e-06,
+      "loss": 0.2784,
+      "step": 9830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.220535490866628e-06,
+      "loss": 0.278,
+      "step": 9831
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.2188672950204355e-06,
+      "loss": 0.2689,
+      "step": 9832
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.217199099174243e-06,
+      "loss": 0.258,
+      "step": 9833
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.215530903328051e-06,
+      "loss": 0.2558,
+      "step": 9834
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.213862707481859e-06,
+      "loss": 0.2601,
+      "step": 9835
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.212194511635666e-06,
+      "loss": 0.2788,
+      "step": 9836
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 0.2773,
+      "step": 9837
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.208858119943281e-06,
+      "loss": 0.2657,
+      "step": 9838
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.207189924097089e-06,
+      "loss": 0.2564,
+      "step": 9839
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.205521728250897e-06,
+      "loss": 0.2925,
+      "step": 9840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.203853532404705e-06,
+      "loss": 0.245,
+      "step": 9841
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.202185336558513e-06,
+      "loss": 0.2612,
+      "step": 9842
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.200517140712321e-06,
+      "loss": 0.2807,
+      "step": 9843
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.198848944866128e-06,
+      "loss": 0.2768,
+      "step": 9844
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1971807490199355e-06,
+      "loss": 0.2742,
+      "step": 9845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1955125531737426e-06,
+      "loss": 0.2501,
+      "step": 9846
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1938443573275504e-06,
+      "loss": 0.2767,
+      "step": 9847
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.192176161481358e-06,
+      "loss": 0.2806,
+      "step": 9848
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.190507965635166e-06,
+      "loss": 0.2814,
+      "step": 9849
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.188839769788974e-06,
+      "loss": 0.2688,
+      "step": 9850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.187171573942781e-06,
+      "loss": 0.261,
+      "step": 9851
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.185503378096589e-06,
+      "loss": 0.2559,
+      "step": 9852
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.183835182250396e-06,
+      "loss": 0.2632,
+      "step": 9853
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.182166986404204e-06,
+      "loss": 0.2852,
+      "step": 9854
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.180498790558012e-06,
+      "loss": 0.2745,
+      "step": 9855
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.17883059471182e-06,
+      "loss": 0.2414,
+      "step": 9856
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.177162398865628e-06,
+      "loss": 0.2649,
+      "step": 9857
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1754942030194355e-06,
+      "loss": 0.2786,
+      "step": 9858
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1738260071732426e-06,
+      "loss": 0.2929,
+      "step": 9859
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.17215781132705e-06,
+      "loss": 0.2685,
+      "step": 9860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.1704896154808575e-06,
+      "loss": 0.2717,
+      "step": 9861
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.168821419634665e-06,
+      "loss": 0.2916,
+      "step": 9862
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.167153223788473e-06,
+      "loss": 0.283,
+      "step": 9863
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.165485027942281e-06,
+      "loss": 0.282,
+      "step": 9864
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.163816832096089e-06,
+      "loss": 0.2759,
+      "step": 9865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.162148636249896e-06,
+      "loss": 0.2741,
+      "step": 9866
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.160480440403704e-06,
+      "loss": 0.2898,
+      "step": 9867
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 4.158812244557511e-06,
+      "loss": 0.2855,
+      "step": 9868
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.157144048711319e-06,
+      "loss": 0.2559,
+      "step": 9869
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.155475852865127e-06,
+      "loss": 0.2461,
+      "step": 9870
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.153807657018935e-06,
+      "loss": 0.2851,
+      "step": 9871
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1521394611727425e-06,
+      "loss": 0.2595,
+      "step": 9872
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.15047126532655e-06,
+      "loss": 0.2588,
+      "step": 9873
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1488030694803575e-06,
+      "loss": 0.2521,
+      "step": 9874
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1471348736341645e-06,
+      "loss": 0.3089,
+      "step": 9875
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.145466677787972e-06,
+      "loss": 0.2767,
+      "step": 9876
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.14379848194178e-06,
+      "loss": 0.242,
+      "step": 9877
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.142130286095588e-06,
+      "loss": 0.2717,
+      "step": 9878
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.140462090249396e-06,
+      "loss": 0.2755,
+      "step": 9879
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.138793894403203e-06,
+      "loss": 0.273,
+      "step": 9880
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.137125698557011e-06,
+      "loss": 0.2635,
+      "step": 9881
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.135457502710819e-06,
+      "loss": 0.2628,
+      "step": 9882
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.133789306864626e-06,
+      "loss": 0.2713,
+      "step": 9883
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.132121111018434e-06,
+      "loss": 0.2643,
+      "step": 9884
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.130452915172242e-06,
+      "loss": 0.2629,
+      "step": 9885
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1287847193260496e-06,
+      "loss": 0.2781,
+      "step": 9886
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1271165234798575e-06,
+      "loss": 0.2636,
+      "step": 9887
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1254483276336645e-06,
+      "loss": 0.2781,
+      "step": 9888
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.123780131787472e-06,
+      "loss": 0.2889,
+      "step": 9889
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1221119359412794e-06,
+      "loss": 0.283,
+      "step": 9890
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.120443740095087e-06,
+      "loss": 0.2642,
+      "step": 9891
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.118775544248895e-06,
+      "loss": 0.2687,
+      "step": 9892
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.117107348402703e-06,
+      "loss": 0.2663,
+      "step": 9893
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.115439152556511e-06,
+      "loss": 0.2691,
+      "step": 9894
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.113770956710318e-06,
+      "loss": 0.2411,
+      "step": 9895
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.112102760864126e-06,
+      "loss": 0.2838,
+      "step": 9896
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.110434565017933e-06,
+      "loss": 0.2725,
+      "step": 9897
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.108766369171741e-06,
+      "loss": 0.2651,
+      "step": 9898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.107098173325549e-06,
+      "loss": 0.2488,
+      "step": 9899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.105429977479357e-06,
+      "loss": 0.2512,
+      "step": 9900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1037617816331645e-06,
+      "loss": 0.2684,
+      "step": 9901
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.1020935857869715e-06,
+      "loss": 0.2641,
+      "step": 9902
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.100425389940779e-06,
+      "loss": 0.2791,
+      "step": 9903
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.098757194094587e-06,
+      "loss": 0.2905,
+      "step": 9904
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.097088998248394e-06,
+      "loss": 0.2615,
+      "step": 9905
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.095420802402202e-06,
+      "loss": 0.2873,
+      "step": 9906
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.09375260655601e-06,
+      "loss": 0.2845,
+      "step": 9907
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.092084410709818e-06,
+      "loss": 0.2698,
+      "step": 9908
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.090416214863625e-06,
+      "loss": 0.2745,
+      "step": 9909
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.088748019017433e-06,
+      "loss": 0.2686,
+      "step": 9910
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.087079823171241e-06,
+      "loss": 0.2759,
+      "step": 9911
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.085411627325048e-06,
+      "loss": 0.2671,
+      "step": 9912
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.083743431478856e-06,
+      "loss": 0.2509,
+      "step": 9913
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.082075235632664e-06,
+      "loss": 0.2726,
+      "step": 9914
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0804070397864715e-06,
+      "loss": 0.2724,
+      "step": 9915
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.078738843940279e-06,
+      "loss": 0.253,
+      "step": 9916
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0770706480940864e-06,
+      "loss": 0.274,
+      "step": 9917
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.075402452247894e-06,
+      "loss": 0.2786,
+      "step": 9918
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.073734256401702e-06,
+      "loss": 0.268,
+      "step": 9919
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.072066060555509e-06,
+      "loss": 0.2822,
+      "step": 9920
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.070397864709317e-06,
+      "loss": 0.246,
+      "step": 9921
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.068729668863125e-06,
+      "loss": 0.2668,
+      "step": 9922
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.067061473016933e-06,
+      "loss": 0.2591,
+      "step": 9923
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.06539327717074e-06,
+      "loss": 0.2637,
+      "step": 9924
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.063725081324548e-06,
+      "loss": 0.2624,
+      "step": 9925
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.062056885478356e-06,
+      "loss": 0.2791,
+      "step": 9926
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.060388689632163e-06,
+      "loss": 0.2886,
+      "step": 9927
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.058720493785971e-06,
+      "loss": 0.2741,
+      "step": 9928
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0570522979397786e-06,
+      "loss": 0.2707,
+      "step": 9929
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0553841020935864e-06,
+      "loss": 0.2825,
+      "step": 9930
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0537159062473935e-06,
+      "loss": 0.2662,
+      "step": 9931
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.052047710401201e-06,
+      "loss": 0.2764,
+      "step": 9932
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.050379514555009e-06,
+      "loss": 0.2792,
+      "step": 9933
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.048711318708817e-06,
+      "loss": 0.276,
+      "step": 9934
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.047043122862624e-06,
+      "loss": 0.2549,
+      "step": 9935
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.045374927016432e-06,
+      "loss": 0.2717,
+      "step": 9936
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.04370673117024e-06,
+      "loss": 0.2848,
+      "step": 9937
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.042038535324047e-06,
+      "loss": 0.2586,
+      "step": 9938
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.040370339477855e-06,
+      "loss": 0.2883,
+      "step": 9939
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.038702143631663e-06,
+      "loss": 0.3005,
+      "step": 9940
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.037033947785471e-06,
+      "loss": 0.2668,
+      "step": 9941
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.035365751939278e-06,
+      "loss": 0.2584,
+      "step": 9942
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.033697556093086e-06,
+      "loss": 0.2665,
+      "step": 9943
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.0320293602468935e-06,
+      "loss": 0.2754,
+      "step": 9944
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.030361164400701e-06,
+      "loss": 0.2567,
+      "step": 9945
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.028692968554508e-06,
+      "loss": 0.2588,
+      "step": 9946
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.027024772708316e-06,
+      "loss": 0.2703,
+      "step": 9947
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.025356576862124e-06,
+      "loss": 0.2763,
+      "step": 9948
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.023688381015931e-06,
+      "loss": 0.278,
+      "step": 9949
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.022020185169739e-06,
+      "loss": 0.2684,
+      "step": 9950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.020351989323547e-06,
+      "loss": 0.2832,
+      "step": 9951
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.018683793477355e-06,
+      "loss": 0.2763,
+      "step": 9952
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.017015597631162e-06,
+      "loss": 0.2782,
+      "step": 9953
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.01534740178497e-06,
+      "loss": 0.2686,
+      "step": 9954
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.013679205938778e-06,
+      "loss": 0.2485,
+      "step": 9955
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0120110100925856e-06,
+      "loss": 0.2688,
+      "step": 9956
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.010342814246393e-06,
+      "loss": 0.2617,
+      "step": 9957
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0086746184002005e-06,
+      "loss": 0.2614,
+      "step": 9958
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.007006422554008e-06,
+      "loss": 0.2817,
+      "step": 9959
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.0053382267078154e-06,
+      "loss": 0.2838,
+      "step": 9960
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.003670030861623e-06,
+      "loss": 0.2839,
+      "step": 9961
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.002001835015431e-06,
+      "loss": 0.2571,
+      "step": 9962
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 4.000333639169239e-06,
+      "loss": 0.2508,
+      "step": 9963
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.998665443323046e-06,
+      "loss": 0.2525,
+      "step": 9964
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.996997247476854e-06,
+      "loss": 0.2636,
+      "step": 9965
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.995329051630662e-06,
+      "loss": 0.2815,
+      "step": 9966
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.993660855784469e-06,
+      "loss": 0.289,
+      "step": 9967
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.991992659938277e-06,
+      "loss": 0.272,
+      "step": 9968
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.990324464092085e-06,
+      "loss": 0.2627,
+      "step": 9969
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.988656268245893e-06,
+      "loss": 0.2783,
+      "step": 9970
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9869880723997005e-06,
+      "loss": 0.2746,
+      "step": 9971
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9853198765535075e-06,
+      "loss": 0.2622,
+      "step": 9972
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.983651680707315e-06,
+      "loss": 0.2782,
+      "step": 9973
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.981983484861123e-06,
+      "loss": 0.2774,
+      "step": 9974
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.98031528901493e-06,
+      "loss": 0.2754,
+      "step": 9975
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.978647093168738e-06,
+      "loss": 0.2611,
+      "step": 9976
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.976978897322546e-06,
+      "loss": 0.2516,
+      "step": 9977
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.975310701476354e-06,
+      "loss": 0.2474,
+      "step": 9978
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.973642505630161e-06,
+      "loss": 0.2543,
+      "step": 9979
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.971974309783969e-06,
+      "loss": 0.2625,
+      "step": 9980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.970306113937777e-06,
+      "loss": 0.2791,
+      "step": 9981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.968637918091584e-06,
+      "loss": 0.2626,
+      "step": 9982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.966969722245392e-06,
+      "loss": 0.2711,
+      "step": 9983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9653015263992e-06,
+      "loss": 0.2531,
+      "step": 9984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9636333305530075e-06,
+      "loss": 0.2761,
+      "step": 9985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.961965134706815e-06,
+      "loss": 0.2888,
+      "step": 9986
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.9602969388606224e-06,
+      "loss": 0.2709,
+      "step": 9987
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.95862874301443e-06,
+      "loss": 0.2685,
+      "step": 9988
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.956960547168237e-06,
+      "loss": 0.3068,
+      "step": 9989
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.955292351322045e-06,
+      "loss": 0.2899,
+      "step": 9990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.953624155475853e-06,
+      "loss": 0.2764,
+      "step": 9991
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.951955959629661e-06,
+      "loss": 0.2668,
+      "step": 9992
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.950287763783469e-06,
+      "loss": 0.2619,
+      "step": 9993
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.948619567937276e-06,
+      "loss": 0.2739,
+      "step": 9994
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.946951372091084e-06,
+      "loss": 0.2741,
+      "step": 9995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.945283176244891e-06,
+      "loss": 0.2783,
+      "step": 9996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.943614980398699e-06,
+      "loss": 0.2764,
+      "step": 9997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.941946784552507e-06,
+      "loss": 0.2671,
+      "step": 9998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9402785887063146e-06,
+      "loss": 0.2812,
+      "step": 9999
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9386103928601224e-06,
+      "loss": 0.2831,
+      "step": 10000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9369421970139295e-06,
+      "loss": 0.2808,
+      "step": 10001
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.935274001167737e-06,
+      "loss": 0.2664,
+      "step": 10002
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.933605805321545e-06,
+      "loss": 0.2816,
+      "step": 10003
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.931937609475352e-06,
+      "loss": 0.264,
+      "step": 10004
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.93026941362916e-06,
+      "loss": 0.2692,
+      "step": 10005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.928601217782968e-06,
+      "loss": 0.2745,
+      "step": 10006
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.926933021936776e-06,
+      "loss": 0.2787,
+      "step": 10007
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.925264826090584e-06,
+      "loss": 0.2738,
+      "step": 10008
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.923596630244391e-06,
+      "loss": 0.2657,
+      "step": 10009
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.921928434398199e-06,
+      "loss": 0.2699,
+      "step": 10010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.920260238552006e-06,
+      "loss": 0.2775,
+      "step": 10011
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.918592042705814e-06,
+      "loss": 0.2831,
+      "step": 10012
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.916923846859622e-06,
+      "loss": 0.2905,
+      "step": 10013
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.9152556510134295e-06,
+      "loss": 0.2802,
+      "step": 10014
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.913587455167237e-06,
+      "loss": 0.2649,
+      "step": 10015
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.911919259321044e-06,
+      "loss": 0.2703,
+      "step": 10016
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.910251063474852e-06,
+      "loss": 0.272,
+      "step": 10017
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.908582867628659e-06,
+      "loss": 0.2712,
+      "step": 10018
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.906914671782467e-06,
+      "loss": 0.2629,
+      "step": 10019
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.905246475936275e-06,
+      "loss": 0.2614,
+      "step": 10020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.903578280090083e-06,
+      "loss": 0.2699,
+      "step": 10021
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.901910084243891e-06,
+      "loss": 0.2625,
+      "step": 10022
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.900241888397699e-06,
+      "loss": 0.2491,
+      "step": 10023
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.898573692551506e-06,
+      "loss": 0.2587,
+      "step": 10024
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.896905496705313e-06,
+      "loss": 0.2803,
+      "step": 10025
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.895237300859121e-06,
+      "loss": 0.2929,
+      "step": 10026
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.893569105012929e-06,
+      "loss": 0.2938,
+      "step": 10027
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.8919009091667365e-06,
+      "loss": 0.2722,
+      "step": 10028
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.890232713320544e-06,
+      "loss": 0.2769,
+      "step": 10029
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.888564517474352e-06,
+      "loss": 0.2729,
+      "step": 10030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.886896321628159e-06,
+      "loss": 0.2836,
+      "step": 10031
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.885228125781967e-06,
+      "loss": 0.2553,
+      "step": 10032
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.883559929935774e-06,
+      "loss": 0.2696,
+      "step": 10033
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.881891734089582e-06,
+      "loss": 0.2686,
+      "step": 10034
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.88022353824339e-06,
+      "loss": 0.2819,
+      "step": 10035
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.878555342397198e-06,
+      "loss": 0.2753,
+      "step": 10036
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.876887146551006e-06,
+      "loss": 0.2679,
+      "step": 10037
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.875218950704813e-06,
+      "loss": 0.2764,
+      "step": 10038
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.873550754858621e-06,
+      "loss": 0.2703,
+      "step": 10039
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.871882559012428e-06,
+      "loss": 0.2761,
+      "step": 10040
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.870214363166236e-06,
+      "loss": 0.2746,
+      "step": 10041
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8685461673200435e-06,
+      "loss": 0.2726,
+      "step": 10042
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.866877971473851e-06,
+      "loss": 0.2552,
+      "step": 10043
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.865209775627659e-06,
+      "loss": 0.2516,
+      "step": 10044
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.863541579781467e-06,
+      "loss": 0.2931,
+      "step": 10045
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.861873383935274e-06,
+      "loss": 0.2556,
+      "step": 10046
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.860205188089082e-06,
+      "loss": 0.2585,
+      "step": 10047
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.858536992242889e-06,
+      "loss": 0.2754,
+      "step": 10048
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.856868796396697e-06,
+      "loss": 0.277,
+      "step": 10049
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.855200600550505e-06,
+      "loss": 0.2789,
+      "step": 10050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.853532404704313e-06,
+      "loss": 0.2732,
+      "step": 10051
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.851864208858121e-06,
+      "loss": 0.2776,
+      "step": 10052
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.850196013011928e-06,
+      "loss": 0.2664,
+      "step": 10053
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.848527817165736e-06,
+      "loss": 0.2618,
+      "step": 10054
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.846859621319543e-06,
+      "loss": 0.2657,
+      "step": 10055
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8451914254733506e-06,
+      "loss": 0.2828,
+      "step": 10056
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8435232296271584e-06,
+      "loss": 0.2823,
+      "step": 10057
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.841855033780966e-06,
+      "loss": 0.2683,
+      "step": 10058
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.840186837934774e-06,
+      "loss": 0.2787,
+      "step": 10059
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.838518642088582e-06,
+      "loss": 0.2636,
+      "step": 10060
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.836850446242389e-06,
+      "loss": 0.2701,
+      "step": 10061
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.835182250396197e-06,
+      "loss": 0.251,
+      "step": 10062
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.833514054550004e-06,
+      "loss": 0.277,
+      "step": 10063
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.831845858703812e-06,
+      "loss": 0.2602,
+      "step": 10064
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.83017766285762e-06,
+      "loss": 0.2827,
+      "step": 10065
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.828509467011428e-06,
+      "loss": 0.2621,
+      "step": 10066
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.826841271165236e-06,
+      "loss": 0.249,
+      "step": 10067
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.825173075319043e-06,
+      "loss": 0.2759,
+      "step": 10068
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8235048794728506e-06,
+      "loss": 0.2583,
+      "step": 10069
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.821836683626658e-06,
+      "loss": 0.2679,
+      "step": 10070
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.8201684877804655e-06,
+      "loss": 0.2718,
+      "step": 10071
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.818500291934273e-06,
+      "loss": 0.263,
+      "step": 10072
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.816832096088081e-06,
+      "loss": 0.2844,
+      "step": 10073
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.815163900241889e-06,
+      "loss": 0.2743,
+      "step": 10074
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8134957043956966e-06,
+      "loss": 0.2724,
+      "step": 10075
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8118275085495036e-06,
+      "loss": 0.27,
+      "step": 10076
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8101593127033115e-06,
+      "loss": 0.2886,
+      "step": 10077
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8084911168571194e-06,
+      "loss": 0.2719,
+      "step": 10078
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.806822921010927e-06,
+      "loss": 0.2709,
+      "step": 10079
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8051547251647348e-06,
+      "loss": 0.2717,
+      "step": 10080
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.8034865293185427e-06,
+      "loss": 0.2679,
+      "step": 10081
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.80181833347235e-06,
+      "loss": 0.2511,
+      "step": 10082
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.800150137626157e-06,
+      "loss": 0.2885,
+      "step": 10083
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.798481941779965e-06,
+      "loss": 0.2729,
+      "step": 10084
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.796813745933773e-06,
+      "loss": 0.2577,
+      "step": 10085
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7951455500875804e-06,
+      "loss": 0.2734,
+      "step": 10086
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7934773542413883e-06,
+      "loss": 0.2866,
+      "step": 10087
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.791809158395196e-06,
+      "loss": 0.27,
+      "step": 10088
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7901409625490036e-06,
+      "loss": 0.2694,
+      "step": 10089
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.788472766702811e-06,
+      "loss": 0.2995,
+      "step": 10090
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7868045708566186e-06,
+      "loss": 0.2482,
+      "step": 10091
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7851363750104265e-06,
+      "loss": 0.26,
+      "step": 10092
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7834681791642343e-06,
+      "loss": 0.2951,
+      "step": 10093
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.781799983318042e-06,
+      "loss": 0.2576,
+      "step": 10094
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7801317874718497e-06,
+      "loss": 0.259,
+      "step": 10095
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.778463591625657e-06,
+      "loss": 0.2615,
+      "step": 10096
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.776795395779465e-06,
+      "loss": 0.264,
+      "step": 10097
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.775127199933272e-06,
+      "loss": 0.2884,
+      "step": 10098
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77345900408708e-06,
+      "loss": 0.2827,
+      "step": 10099
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.771790808240888e-06,
+      "loss": 0.2718,
+      "step": 10100
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7701226123946953e-06,
+      "loss": 0.2568,
+      "step": 10101
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.768454416548503e-06,
+      "loss": 0.2723,
+      "step": 10102
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.766786220702311e-06,
+      "loss": 0.2682,
+      "step": 10103
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7651180248561186e-06,
+      "loss": 0.2763,
+      "step": 10104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.763449829009926e-06,
+      "loss": 0.2895,
+      "step": 10105
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7617816331637335e-06,
+      "loss": 0.2828,
+      "step": 10106
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7601134373175414e-06,
+      "loss": 0.2607,
+      "step": 10107
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.758445241471349e-06,
+      "loss": 0.2795,
+      "step": 10108
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7567770456251567e-06,
+      "loss": 0.2569,
+      "step": 10109
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7551088497789646e-06,
+      "loss": 0.2915,
+      "step": 10110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.753440653932772e-06,
+      "loss": 0.2868,
+      "step": 10111
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7517724580865795e-06,
+      "loss": 0.2892,
+      "step": 10112
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.750104262240387e-06,
+      "loss": 0.2675,
+      "step": 10113
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.748436066394195e-06,
+      "loss": 0.2681,
+      "step": 10114
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.7467678705480028e-06,
+      "loss": 0.2614,
+      "step": 10115
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7450996747018102e-06,
+      "loss": 0.2678,
+      "step": 10116
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.743431478855618e-06,
+      "loss": 0.257,
+      "step": 10117
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.741763283009426e-06,
+      "loss": 0.262,
+      "step": 10118
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.740095087163233e-06,
+      "loss": 0.2732,
+      "step": 10119
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.738426891317041e-06,
+      "loss": 0.2727,
+      "step": 10120
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7367586954708484e-06,
+      "loss": 0.2646,
+      "step": 10121
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7350904996246563e-06,
+      "loss": 0.2569,
+      "step": 10122
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7334223037784637e-06,
+      "loss": 0.2799,
+      "step": 10123
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7317541079322716e-06,
+      "loss": 0.2533,
+      "step": 10124
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7300859120860795e-06,
+      "loss": 0.2717,
+      "step": 10125
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.728417716239887e-06,
+      "loss": 0.2941,
+      "step": 10126
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7267495203936945e-06,
+      "loss": 0.2679,
+      "step": 10127
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.725081324547502e-06,
+      "loss": 0.2815,
+      "step": 10128
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.72341312870131e-06,
+      "loss": 0.2641,
+      "step": 10129
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7217449328551177e-06,
+      "loss": 0.2677,
+      "step": 10130
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.720076737008925e-06,
+      "loss": 0.2743,
+      "step": 10131
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.718408541162733e-06,
+      "loss": 0.2843,
+      "step": 10132
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7167403453165405e-06,
+      "loss": 0.2687,
+      "step": 10133
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.715072149470348e-06,
+      "loss": 0.2822,
+      "step": 10134
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7134039536241554e-06,
+      "loss": 0.2785,
+      "step": 10135
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7117357577779633e-06,
+      "loss": 0.2735,
+      "step": 10136
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.710067561931771e-06,
+      "loss": 0.2825,
+      "step": 10137
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7083993660855787e-06,
+      "loss": 0.2825,
+      "step": 10138
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7067311702393866e-06,
+      "loss": 0.2771,
+      "step": 10139
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7050629743931944e-06,
+      "loss": 0.2693,
+      "step": 10140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7033947785470015e-06,
+      "loss": 0.2962,
+      "step": 10141
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.7017265827008094e-06,
+      "loss": 0.2839,
+      "step": 10142
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.700058386854617e-06,
+      "loss": 0.2777,
+      "step": 10143
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6983901910084247e-06,
+      "loss": 0.2872,
+      "step": 10144
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6967219951622326e-06,
+      "loss": 0.2669,
+      "step": 10145
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.69505379931604e-06,
+      "loss": 0.2878,
+      "step": 10146
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.693385603469848e-06,
+      "loss": 0.2523,
+      "step": 10147
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.691717407623655e-06,
+      "loss": 0.2828,
+      "step": 10148
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.690049211777463e-06,
+      "loss": 0.2868,
+      "step": 10149
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6883810159312703e-06,
+      "loss": 0.2987,
+      "step": 10150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6867128200850782e-06,
+      "loss": 0.26,
+      "step": 10151
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.685044624238886e-06,
+      "loss": 0.297,
+      "step": 10152
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6833764283926936e-06,
+      "loss": 0.2945,
+      "step": 10153
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6817082325465015e-06,
+      "loss": 0.2834,
+      "step": 10154
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6800400367003094e-06,
+      "loss": 0.2863,
+      "step": 10155
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6783718408541164e-06,
+      "loss": 0.2693,
+      "step": 10156
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.6767036450079243e-06,
+      "loss": 0.2754,
+      "step": 10157
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6750354491617317e-06,
+      "loss": 0.2961,
+      "step": 10158
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6733672533155396e-06,
+      "loss": 0.2605,
+      "step": 10159
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.671699057469347e-06,
+      "loss": 0.2501,
+      "step": 10160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.670030861623155e-06,
+      "loss": 0.2572,
+      "step": 10161
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.668362665776963e-06,
+      "loss": 0.2691,
+      "step": 10162
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.66669446993077e-06,
+      "loss": 0.282,
+      "step": 10163
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.665026274084578e-06,
+      "loss": 0.2598,
+      "step": 10164
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6633580782383853e-06,
+      "loss": 0.2667,
+      "step": 10165
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.661689882392193e-06,
+      "loss": 0.2546,
+      "step": 10166
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.660021686546001e-06,
+      "loss": 0.2674,
+      "step": 10167
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6583534906998085e-06,
+      "loss": 0.2581,
+      "step": 10168
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6566852948536164e-06,
+      "loss": 0.2632,
+      "step": 10169
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6550170990074234e-06,
+      "loss": 0.2723,
+      "step": 10170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6533489031612313e-06,
+      "loss": 0.2748,
+      "step": 10171
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6516807073150388e-06,
+      "loss": 0.2966,
+      "step": 10172
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6500125114688467e-06,
+      "loss": 0.2782,
+      "step": 10173
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6483443156226546e-06,
+      "loss": 0.2546,
+      "step": 10174
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.646676119776462e-06,
+      "loss": 0.2688,
+      "step": 10175
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.64500792393027e-06,
+      "loss": 0.2788,
+      "step": 10176
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.643339728084077e-06,
+      "loss": 0.2673,
+      "step": 10177
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.641671532237885e-06,
+      "loss": 0.2606,
+      "step": 10178
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6400033363916927e-06,
+      "loss": 0.2597,
+      "step": 10179
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6383351405455e-06,
+      "loss": 0.267,
+      "step": 10180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.636666944699308e-06,
+      "loss": 0.2738,
+      "step": 10181
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.634998748853116e-06,
+      "loss": 0.269,
+      "step": 10182
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6333305530069234e-06,
+      "loss": 0.2778,
+      "step": 10183
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6316623571607313e-06,
+      "loss": 0.2885,
+      "step": 10184
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6299941613145383e-06,
+      "loss": 0.2839,
+      "step": 10185
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6283259654683462e-06,
+      "loss": 0.2551,
+      "step": 10186
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6266577696221537e-06,
+      "loss": 0.2646,
+      "step": 10187
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6249895737759616e-06,
+      "loss": 0.2703,
+      "step": 10188
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6233213779297695e-06,
+      "loss": 0.2769,
+      "step": 10189
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.621653182083577e-06,
+      "loss": 0.2667,
+      "step": 10190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.619984986237385e-06,
+      "loss": 0.2746,
+      "step": 10191
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.618316790391192e-06,
+      "loss": 0.2528,
+      "step": 10192
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6166485945449998e-06,
+      "loss": 0.2765,
+      "step": 10193
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6149803986988076e-06,
+      "loss": 0.2857,
+      "step": 10194
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.613312202852615e-06,
+      "loss": 0.265,
+      "step": 10195
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.611644007006423e-06,
+      "loss": 0.2667,
+      "step": 10196
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6099758111602305e-06,
+      "loss": 0.2857,
+      "step": 10197
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.6083076153140383e-06,
+      "loss": 0.2671,
+      "step": 10198
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6066394194678454e-06,
+      "loss": 0.2798,
+      "step": 10199
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6049712236216533e-06,
+      "loss": 0.2634,
+      "step": 10200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.603303027775461e-06,
+      "loss": 0.2746,
+      "step": 10201
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.6016348319292686e-06,
+      "loss": 0.2782,
+      "step": 10202
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5999666360830765e-06,
+      "loss": 0.2729,
+      "step": 10203
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5982984402368844e-06,
+      "loss": 0.2602,
+      "step": 10204
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.596630244390692e-06,
+      "loss": 0.2695,
+      "step": 10205
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5949620485444993e-06,
+      "loss": 0.2903,
+      "step": 10206
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5932938526983068e-06,
+      "loss": 0.2753,
+      "step": 10207
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5916256568521147e-06,
+      "loss": 0.2848,
+      "step": 10208
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5899574610059226e-06,
+      "loss": 0.2691,
+      "step": 10209
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.58828926515973e-06,
+      "loss": 0.2496,
+      "step": 10210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.586621069313538e-06,
+      "loss": 0.2443,
+      "step": 10211
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5849528734673454e-06,
+      "loss": 0.2656,
+      "step": 10212
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5832846776211533e-06,
+      "loss": 0.2672,
+      "step": 10213
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5816164817749603e-06,
+      "loss": 0.2773,
+      "step": 10214
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.579948285928768e-06,
+      "loss": 0.2734,
+      "step": 10215
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.578280090082576e-06,
+      "loss": 0.2667,
+      "step": 10216
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5766118942363835e-06,
+      "loss": 0.2575,
+      "step": 10217
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5749436983901914e-06,
+      "loss": 0.2845,
+      "step": 10218
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5732755025439993e-06,
+      "loss": 0.2396,
+      "step": 10219
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5716073066978068e-06,
+      "loss": 0.2693,
+      "step": 10220
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5699391108516142e-06,
+      "loss": 0.2599,
+      "step": 10221
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5682709150054217e-06,
+      "loss": 0.2593,
+      "step": 10222
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5666027191592296e-06,
+      "loss": 0.2705,
+      "step": 10223
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.564934523313037e-06,
+      "loss": 0.2723,
+      "step": 10224
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.563266327466845e-06,
+      "loss": 0.2835,
+      "step": 10225
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.561598131620653e-06,
+      "loss": 0.2667,
+      "step": 10226
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5599299357744603e-06,
+      "loss": 0.2783,
+      "step": 10227
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5582617399282678e-06,
+      "loss": 0.2612,
+      "step": 10228
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5565935440820752e-06,
+      "loss": 0.2543,
+      "step": 10229
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.554925348235883e-06,
+      "loss": 0.2585,
+      "step": 10230
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.553257152389691e-06,
+      "loss": 0.287,
+      "step": 10231
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5515889565434985e-06,
+      "loss": 0.274,
+      "step": 10232
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5499207606973063e-06,
+      "loss": 0.2826,
+      "step": 10233
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5482525648511142e-06,
+      "loss": 0.269,
+      "step": 10234
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5465843690049213e-06,
+      "loss": 0.2447,
+      "step": 10235
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5449161731587287e-06,
+      "loss": 0.2726,
+      "step": 10236
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5432479773125366e-06,
+      "loss": 0.261,
+      "step": 10237
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.5415797814663445e-06,
+      "loss": 0.2618,
+      "step": 10238
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.539911585620152e-06,
+      "loss": 0.27,
+      "step": 10239
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.53824338977396e-06,
+      "loss": 0.2741,
+      "step": 10240
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5365751939277677e-06,
+      "loss": 0.2783,
+      "step": 10241
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.534906998081575e-06,
+      "loss": 0.2623,
+      "step": 10242
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5332388022353827e-06,
+      "loss": 0.2737,
+      "step": 10243
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.53157060638919e-06,
+      "loss": 0.2637,
+      "step": 10244
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.529902410542998e-06,
+      "loss": 0.2709,
+      "step": 10245
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.528234214696806e-06,
+      "loss": 0.2679,
+      "step": 10246
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5265660188506134e-06,
+      "loss": 0.2713,
+      "step": 10247
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5248978230044213e-06,
+      "loss": 0.2849,
+      "step": 10248
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5232296271582287e-06,
+      "loss": 0.2811,
+      "step": 10249
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.521561431312036e-06,
+      "loss": 0.2523,
+      "step": 10250
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5198932354658436e-06,
+      "loss": 0.2801,
+      "step": 10251
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5182250396196515e-06,
+      "loss": 0.2588,
+      "step": 10252
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5165568437734594e-06,
+      "loss": 0.2719,
+      "step": 10253
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.514888647927267e-06,
+      "loss": 0.267,
+      "step": 10254
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5132204520810748e-06,
+      "loss": 0.2766,
+      "step": 10255
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5115522562348827e-06,
+      "loss": 0.265,
+      "step": 10256
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5098840603886897e-06,
+      "loss": 0.2639,
+      "step": 10257
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5082158645424976e-06,
+      "loss": 0.2564,
+      "step": 10258
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.506547668696305e-06,
+      "loss": 0.2615,
+      "step": 10259
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.504879472850113e-06,
+      "loss": 0.2691,
+      "step": 10260
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5032112770039204e-06,
+      "loss": 0.2826,
+      "step": 10261
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5015430811577283e-06,
+      "loss": 0.2594,
+      "step": 10262
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.499874885311536e-06,
+      "loss": 0.2974,
+      "step": 10263
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4982066894653432e-06,
+      "loss": 0.2608,
+      "step": 10264
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.496538493619151e-06,
+      "loss": 0.2573,
+      "step": 10265
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4948702977729586e-06,
+      "loss": 0.2565,
+      "step": 10266
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4932021019267665e-06,
+      "loss": 0.2847,
+      "step": 10267
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4915339060805743e-06,
+      "loss": 0.2697,
+      "step": 10268
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.489865710234382e-06,
+      "loss": 0.262,
+      "step": 10269
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4881975143881897e-06,
+      "loss": 0.2608,
+      "step": 10270
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4865293185419976e-06,
+      "loss": 0.2703,
+      "step": 10271
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4848611226958046e-06,
+      "loss": 0.2683,
+      "step": 10272
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4831929268496125e-06,
+      "loss": 0.2694,
+      "step": 10273
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.48152473100342e-06,
+      "loss": 0.2518,
+      "step": 10274
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.479856535157228e-06,
+      "loss": 0.2611,
+      "step": 10275
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.4781883393110353e-06,
+      "loss": 0.2561,
+      "step": 10276
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.476520143464843e-06,
+      "loss": 0.2813,
+      "step": 10277
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.474851947618651e-06,
+      "loss": 0.272,
+      "step": 10278
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.473183751772458e-06,
+      "loss": 0.2892,
+      "step": 10279
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.471515555926266e-06,
+      "loss": 0.2736,
+      "step": 10280
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4698473600800735e-06,
+      "loss": 0.251,
+      "step": 10281
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4681791642338814e-06,
+      "loss": 0.2702,
+      "step": 10282
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4665109683876893e-06,
+      "loss": 0.2748,
+      "step": 10283
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4648427725414967e-06,
+      "loss": 0.253,
+      "step": 10284
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4631745766953046e-06,
+      "loss": 0.2784,
+      "step": 10285
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4615063808491116e-06,
+      "loss": 0.2775,
+      "step": 10286
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4598381850029195e-06,
+      "loss": 0.2644,
+      "step": 10287
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.458169989156727e-06,
+      "loss": 0.2785,
+      "step": 10288
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.456501793310535e-06,
+      "loss": 0.2569,
+      "step": 10289
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4548335974643428e-06,
+      "loss": 0.2643,
+      "step": 10290
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4531654016181502e-06,
+      "loss": 0.261,
+      "step": 10291
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.451497205771958e-06,
+      "loss": 0.2687,
+      "step": 10292
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.449829009925765e-06,
+      "loss": 0.2704,
+      "step": 10293
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.448160814079573e-06,
+      "loss": 0.2789,
+      "step": 10294
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.446492618233381e-06,
+      "loss": 0.2523,
+      "step": 10295
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4448244223871884e-06,
+      "loss": 0.2757,
+      "step": 10296
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4431562265409963e-06,
+      "loss": 0.27,
+      "step": 10297
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.441488030694804e-06,
+      "loss": 0.2665,
+      "step": 10298
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4398198348486116e-06,
+      "loss": 0.2724,
+      "step": 10299
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4381516390024195e-06,
+      "loss": 0.2636,
+      "step": 10300
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4364834431562266e-06,
+      "loss": 0.2776,
+      "step": 10301
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4348152473100345e-06,
+      "loss": 0.268,
+      "step": 10302
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.433147051463842e-06,
+      "loss": 0.2593,
+      "step": 10303
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.43147885561765e-06,
+      "loss": 0.2729,
+      "step": 10304
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4298106597714577e-06,
+      "loss": 0.2598,
+      "step": 10305
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.428142463925265e-06,
+      "loss": 0.264,
+      "step": 10306
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.426474268079073e-06,
+      "loss": 0.2822,
+      "step": 10307
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.42480607223288e-06,
+      "loss": 0.2793,
+      "step": 10308
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.423137876386688e-06,
+      "loss": 0.2792,
+      "step": 10309
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.421469680540496e-06,
+      "loss": 0.257,
+      "step": 10310
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4198014846943033e-06,
+      "loss": 0.267,
+      "step": 10311
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.418133288848111e-06,
+      "loss": 0.2802,
+      "step": 10312
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4164650930019187e-06,
+      "loss": 0.2808,
+      "step": 10313
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4147968971557266e-06,
+      "loss": 0.281,
+      "step": 10314
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4131287013095336e-06,
+      "loss": 0.2591,
+      "step": 10315
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4114605054633415e-06,
+      "loss": 0.2783,
+      "step": 10316
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4097923096171494e-06,
+      "loss": 0.249,
+      "step": 10317
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.408124113770957e-06,
+      "loss": 0.2522,
+      "step": 10318
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4064559179247647e-06,
+      "loss": 0.2696,
+      "step": 10319
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.4047877220785726e-06,
+      "loss": 0.2696,
+      "step": 10320
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.40311952623238e-06,
+      "loss": 0.2728,
+      "step": 10321
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.4014513303861875e-06,
+      "loss": 0.2726,
+      "step": 10322
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.399783134539995e-06,
+      "loss": 0.2676,
+      "step": 10323
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.398114938693803e-06,
+      "loss": 0.2551,
+      "step": 10324
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3964467428476103e-06,
+      "loss": 0.2657,
+      "step": 10325
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3947785470014182e-06,
+      "loss": 0.2872,
+      "step": 10326
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.393110351155226e-06,
+      "loss": 0.2587,
+      "step": 10327
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3914421553090336e-06,
+      "loss": 0.2538,
+      "step": 10328
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3897739594628415e-06,
+      "loss": 0.2739,
+      "step": 10329
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3881057636166485e-06,
+      "loss": 0.2817,
+      "step": 10330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3864375677704564e-06,
+      "loss": 0.2729,
+      "step": 10331
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3847693719242643e-06,
+      "loss": 0.2817,
+      "step": 10332
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3831011760780718e-06,
+      "loss": 0.29,
+      "step": 10333
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3814329802318796e-06,
+      "loss": 0.277,
+      "step": 10334
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3797647843856875e-06,
+      "loss": 0.2585,
+      "step": 10335
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.378096588539495e-06,
+      "loss": 0.2491,
+      "step": 10336
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3764283926933025e-06,
+      "loss": 0.293,
+      "step": 10337
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.37476019684711e-06,
+      "loss": 0.2699,
+      "step": 10338
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.373092001000918e-06,
+      "loss": 0.2703,
+      "step": 10339
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3714238051547253e-06,
+      "loss": 0.2686,
+      "step": 10340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.369755609308533e-06,
+      "loss": 0.2685,
+      "step": 10341
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.368087413462341e-06,
+      "loss": 0.2494,
+      "step": 10342
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3664192176161485e-06,
+      "loss": 0.2892,
+      "step": 10343
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.364751021769956e-06,
+      "loss": 0.28,
+      "step": 10344
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3630828259237634e-06,
+      "loss": 0.281,
+      "step": 10345
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3614146300775713e-06,
+      "loss": 0.2588,
+      "step": 10346
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.359746434231379e-06,
+      "loss": 0.2659,
+      "step": 10347
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3580782383851867e-06,
+      "loss": 0.2648,
+      "step": 10348
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3564100425389946e-06,
+      "loss": 0.2655,
+      "step": 10349
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.354741846692802e-06,
+      "loss": 0.2773,
+      "step": 10350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3530736508466095e-06,
+      "loss": 0.2695,
+      "step": 10351
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.351405455000417e-06,
+      "loss": 0.2812,
+      "step": 10352
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.349737259154225e-06,
+      "loss": 0.2857,
+      "step": 10353
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3480690633080327e-06,
+      "loss": 0.2729,
+      "step": 10354
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.34640086746184e-06,
+      "loss": 0.2702,
+      "step": 10355
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.344732671615648e-06,
+      "loss": 0.2488,
+      "step": 10356
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.343064475769456e-06,
+      "loss": 0.3087,
+      "step": 10357
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3413962799232634e-06,
+      "loss": 0.2774,
+      "step": 10358
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.339728084077071e-06,
+      "loss": 0.2428,
+      "step": 10359
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3380598882308783e-06,
+      "loss": 0.2749,
+      "step": 10360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3363916923846862e-06,
+      "loss": 0.2759,
+      "step": 10361
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.334723496538494e-06,
+      "loss": 0.2794,
+      "step": 10362
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.3330553006923016e-06,
+      "loss": 0.2705,
+      "step": 10363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3313871048461095e-06,
+      "loss": 0.2686,
+      "step": 10364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.329718908999917e-06,
+      "loss": 0.2773,
+      "step": 10365
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3280507131537244e-06,
+      "loss": 0.2593,
+      "step": 10366
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.326382517307532e-06,
+      "loss": 0.2806,
+      "step": 10367
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3247143214613398e-06,
+      "loss": 0.2795,
+      "step": 10368
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3230461256151476e-06,
+      "loss": 0.2708,
+      "step": 10369
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.321377929768955e-06,
+      "loss": 0.2595,
+      "step": 10370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.319709733922763e-06,
+      "loss": 0.2885,
+      "step": 10371
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.318041538076571e-06,
+      "loss": 0.2672,
+      "step": 10372
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.316373342230378e-06,
+      "loss": 0.2886,
+      "step": 10373
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.314705146384186e-06,
+      "loss": 0.2781,
+      "step": 10374
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3130369505379933e-06,
+      "loss": 0.2651,
+      "step": 10375
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.311368754691801e-06,
+      "loss": 0.2604,
+      "step": 10376
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3097005588456086e-06,
+      "loss": 0.2693,
+      "step": 10377
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3080323629994165e-06,
+      "loss": 0.2642,
+      "step": 10378
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3063641671532244e-06,
+      "loss": 0.2758,
+      "step": 10379
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3046959713070314e-06,
+      "loss": 0.2792,
+      "step": 10380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3030277754608393e-06,
+      "loss": 0.2464,
+      "step": 10381
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.3013595796146468e-06,
+      "loss": 0.2863,
+      "step": 10382
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2996913837684547e-06,
+      "loss": 0.2693,
+      "step": 10383
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2980231879222626e-06,
+      "loss": 0.2571,
+      "step": 10384
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.29635499207607e-06,
+      "loss": 0.2646,
+      "step": 10385
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.294686796229878e-06,
+      "loss": 0.2697,
+      "step": 10386
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.293018600383685e-06,
+      "loss": 0.2701,
+      "step": 10387
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.291350404537493e-06,
+      "loss": 0.268,
+      "step": 10388
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2896822086913003e-06,
+      "loss": 0.2632,
+      "step": 10389
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.288014012845108e-06,
+      "loss": 0.2769,
+      "step": 10390
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.286345816998916e-06,
+      "loss": 0.2707,
+      "step": 10391
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2846776211527235e-06,
+      "loss": 0.2663,
+      "step": 10392
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2830094253065314e-06,
+      "loss": 0.2614,
+      "step": 10393
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2813412294603393e-06,
+      "loss": 0.2475,
+      "step": 10394
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2796730336141463e-06,
+      "loss": 0.2663,
+      "step": 10395
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2780048377679542e-06,
+      "loss": 0.2728,
+      "step": 10396
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2763366419217617e-06,
+      "loss": 0.2791,
+      "step": 10397
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2746684460755696e-06,
+      "loss": 0.2698,
+      "step": 10398
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2730002502293775e-06,
+      "loss": 0.2754,
+      "step": 10399
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.271332054383185e-06,
+      "loss": 0.2771,
+      "step": 10400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.269663858536993e-06,
+      "loss": 0.2789,
+      "step": 10401
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2679956626908e-06,
+      "loss": 0.2683,
+      "step": 10402
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2663274668446078e-06,
+      "loss": 0.2741,
+      "step": 10403
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.2646592709984152e-06,
+      "loss": 0.2868,
+      "step": 10404
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.262991075152223e-06,
+      "loss": 0.269,
+      "step": 10405
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.261322879306031e-06,
+      "loss": 0.2745,
+      "step": 10406
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2596546834598385e-06,
+      "loss": 0.2828,
+      "step": 10407
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2579864876136463e-06,
+      "loss": 0.2779,
+      "step": 10408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2563182917674534e-06,
+      "loss": 0.2817,
+      "step": 10409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2546500959212613e-06,
+      "loss": 0.2684,
+      "step": 10410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.252981900075069e-06,
+      "loss": 0.2732,
+      "step": 10411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2513137042288766e-06,
+      "loss": 0.2794,
+      "step": 10412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2496455083826845e-06,
+      "loss": 0.2559,
+      "step": 10413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.247977312536492e-06,
+      "loss": 0.2699,
+      "step": 10414
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2463091166903e-06,
+      "loss": 0.2732,
+      "step": 10415
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.244640920844107e-06,
+      "loss": 0.2813,
+      "step": 10416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2429727249979148e-06,
+      "loss": 0.2637,
+      "step": 10417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2413045291517227e-06,
+      "loss": 0.2622,
+      "step": 10418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.23963633330553e-06,
+      "loss": 0.2774,
+      "step": 10419
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.237968137459338e-06,
+      "loss": 0.2735,
+      "step": 10420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.236299941613146e-06,
+      "loss": 0.2719,
+      "step": 10421
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2346317457669534e-06,
+      "loss": 0.2649,
+      "step": 10422
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2329635499207613e-06,
+      "loss": 0.2846,
+      "step": 10423
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2312953540745683e-06,
+      "loss": 0.2742,
+      "step": 10424
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.229627158228376e-06,
+      "loss": 0.2743,
+      "step": 10425
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.227958962382184e-06,
+      "loss": 0.2742,
+      "step": 10426
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2262907665359915e-06,
+      "loss": 0.2481,
+      "step": 10427
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2246225706897994e-06,
+      "loss": 0.2847,
+      "step": 10428
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.222954374843607e-06,
+      "loss": 0.2551,
+      "step": 10429
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2212861789974148e-06,
+      "loss": 0.2625,
+      "step": 10430
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.219617983151222e-06,
+      "loss": 0.2688,
+      "step": 10431
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2179497873050297e-06,
+      "loss": 0.2808,
+      "step": 10432
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2162815914588376e-06,
+      "loss": 0.2732,
+      "step": 10433
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.214613395612645e-06,
+      "loss": 0.2568,
+      "step": 10434
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.212945199766453e-06,
+      "loss": 0.2757,
+      "step": 10435
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.211277003920261e-06,
+      "loss": 0.2746,
+      "step": 10436
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2096088080740683e-06,
+      "loss": 0.2562,
+      "step": 10437
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2079406122278758e-06,
+      "loss": 0.2711,
+      "step": 10438
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2062724163816832e-06,
+      "loss": 0.2696,
+      "step": 10439
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.204604220535491e-06,
+      "loss": 0.2592,
+      "step": 10440
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2029360246892986e-06,
+      "loss": 0.2717,
+      "step": 10441
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.2012678288431065e-06,
+      "loss": 0.2744,
+      "step": 10442
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1995996329969143e-06,
+      "loss": 0.2498,
+      "step": 10443
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.197931437150722e-06,
+      "loss": 0.2629,
+      "step": 10444
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.1962632413045293e-06,
+      "loss": 0.254,
+      "step": 10445
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1945950454583367e-06,
+      "loss": 0.282,
+      "step": 10446
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1929268496121446e-06,
+      "loss": 0.2714,
+      "step": 10447
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1912586537659525e-06,
+      "loss": 0.2757,
+      "step": 10448
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.18959045791976e-06,
+      "loss": 0.2613,
+      "step": 10449
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.187922262073568e-06,
+      "loss": 0.2577,
+      "step": 10450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1862540662273757e-06,
+      "loss": 0.2453,
+      "step": 10451
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.184585870381183e-06,
+      "loss": 0.2687,
+      "step": 10452
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1829176745349902e-06,
+      "loss": 0.2557,
+      "step": 10453
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.181249478688798e-06,
+      "loss": 0.2648,
+      "step": 10454
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.179581282842606e-06,
+      "loss": 0.2748,
+      "step": 10455
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1779130869964135e-06,
+      "loss": 0.263,
+      "step": 10456
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1762448911502214e-06,
+      "loss": 0.2917,
+      "step": 10457
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1745766953040293e-06,
+      "loss": 0.2854,
+      "step": 10458
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1729084994578367e-06,
+      "loss": 0.2658,
+      "step": 10459
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.171240303611644e-06,
+      "loss": 0.2596,
+      "step": 10460
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1695721077654516e-06,
+      "loss": 0.2785,
+      "step": 10461
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1679039119192595e-06,
+      "loss": 0.2604,
+      "step": 10462
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1662357160730674e-06,
+      "loss": 0.2906,
+      "step": 10463
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.164567520226875e-06,
+      "loss": 0.2789,
+      "step": 10464
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1628993243806828e-06,
+      "loss": 0.2564,
+      "step": 10465
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1612311285344902e-06,
+      "loss": 0.2701,
+      "step": 10466
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1595629326882977e-06,
+      "loss": 0.2701,
+      "step": 10467
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 0.261,
+      "step": 10468
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.156226540995913e-06,
+      "loss": 0.2611,
+      "step": 10469
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.154558345149721e-06,
+      "loss": 0.2805,
+      "step": 10470
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1528901493035284e-06,
+      "loss": 0.2492,
+      "step": 10471
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1512219534573363e-06,
+      "loss": 0.2606,
+      "step": 10472
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.149553757611144e-06,
+      "loss": 0.271,
+      "step": 10473
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1478855617649512e-06,
+      "loss": 0.2722,
+      "step": 10474
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.146217365918759e-06,
+      "loss": 0.2708,
+      "step": 10475
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1445491700725666e-06,
+      "loss": 0.2691,
+      "step": 10476
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1428809742263745e-06,
+      "loss": 0.2597,
+      "step": 10477
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.141212778380182e-06,
+      "loss": 0.2642,
+      "step": 10478
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.13954458253399e-06,
+      "loss": 0.2733,
+      "step": 10479
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1378763866877977e-06,
+      "loss": 0.2755,
+      "step": 10480
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.136208190841605e-06,
+      "loss": 0.268,
+      "step": 10481
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1345399949954126e-06,
+      "loss": 0.2815,
+      "step": 10482
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.13287179914922e-06,
+      "loss": 0.2612,
+      "step": 10483
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.131203603303028e-06,
+      "loss": 0.2659,
+      "step": 10484
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.129535407456836e-06,
+      "loss": 0.2766,
+      "step": 10485
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.1278672116106433e-06,
+      "loss": 0.2624,
+      "step": 10486
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.126199015764451e-06,
+      "loss": 0.2834,
+      "step": 10487
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.124530819918259e-06,
+      "loss": 0.2848,
+      "step": 10488
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.122862624072066e-06,
+      "loss": 0.2574,
+      "step": 10489
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.121194428225874e-06,
+      "loss": 0.2737,
+      "step": 10490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1195262323796815e-06,
+      "loss": 0.2732,
+      "step": 10491
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1178580365334894e-06,
+      "loss": 0.2725,
+      "step": 10492
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.116189840687297e-06,
+      "loss": 0.2688,
+      "step": 10493
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1145216448411047e-06,
+      "loss": 0.2671,
+      "step": 10494
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1128534489949126e-06,
+      "loss": 0.2713,
+      "step": 10495
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1111852531487196e-06,
+      "loss": 0.264,
+      "step": 10496
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1095170573025275e-06,
+      "loss": 0.2636,
+      "step": 10497
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.107848861456335e-06,
+      "loss": 0.2782,
+      "step": 10498
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.106180665610143e-06,
+      "loss": 0.2649,
+      "step": 10499
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1045124697639508e-06,
+      "loss": 0.2719,
+      "step": 10500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.1028442739177582e-06,
+      "loss": 0.2662,
+      "step": 10501
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.101176078071566e-06,
+      "loss": 0.2705,
+      "step": 10502
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.099507882225373e-06,
+      "loss": 0.2804,
+      "step": 10503
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.097839686379181e-06,
+      "loss": 0.2761,
+      "step": 10504
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0961714905329885e-06,
+      "loss": 0.2699,
+      "step": 10505
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0945032946867964e-06,
+      "loss": 0.2555,
+      "step": 10506
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0928350988406043e-06,
+      "loss": 0.279,
+      "step": 10507
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0911669029944118e-06,
+      "loss": 0.2706,
+      "step": 10508
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0894987071482196e-06,
+      "loss": 0.2684,
+      "step": 10509
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0878305113020275e-06,
+      "loss": 0.2951,
+      "step": 10510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0861623154558346e-06,
+      "loss": 0.2763,
+      "step": 10511
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0844941196096425e-06,
+      "loss": 0.2837,
+      "step": 10512
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.08282592376345e-06,
+      "loss": 0.2745,
+      "step": 10513
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.081157727917258e-06,
+      "loss": 0.2929,
+      "step": 10514
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0794895320710657e-06,
+      "loss": 0.2762,
+      "step": 10515
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.077821336224873e-06,
+      "loss": 0.2605,
+      "step": 10516
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.076153140378681e-06,
+      "loss": 0.2763,
+      "step": 10517
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.074484944532488e-06,
+      "loss": 0.267,
+      "step": 10518
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.072816748686296e-06,
+      "loss": 0.287,
+      "step": 10519
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0711485528401034e-06,
+      "loss": 0.2912,
+      "step": 10520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0694803569939113e-06,
+      "loss": 0.2824,
+      "step": 10521
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.067812161147719e-06,
+      "loss": 0.2563,
+      "step": 10522
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0661439653015267e-06,
+      "loss": 0.2709,
+      "step": 10523
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0644757694553346e-06,
+      "loss": 0.2813,
+      "step": 10524
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0628075736091416e-06,
+      "loss": 0.258,
+      "step": 10525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0611393777629495e-06,
+      "loss": 0.2812,
+      "step": 10526
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.0594711819167574e-06,
+      "loss": 0.264,
+      "step": 10527
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.057802986070565e-06,
+      "loss": 0.2712,
+      "step": 10528
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0561347902243727e-06,
+      "loss": 0.2763,
+      "step": 10529
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.05446659437818e-06,
+      "loss": 0.2747,
+      "step": 10530
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.052798398531988e-06,
+      "loss": 0.2816,
+      "step": 10531
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.051130202685795e-06,
+      "loss": 0.2694,
+      "step": 10532
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.049462006839603e-06,
+      "loss": 0.2601,
+      "step": 10533
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.047793810993411e-06,
+      "loss": 0.2811,
+      "step": 10534
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0461256151472184e-06,
+      "loss": 0.2686,
+      "step": 10535
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0444574193010262e-06,
+      "loss": 0.2783,
+      "step": 10536
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.042789223454834e-06,
+      "loss": 0.2642,
+      "step": 10537
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0411210276086416e-06,
+      "loss": 0.2463,
+      "step": 10538
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0394528317624495e-06,
+      "loss": 0.2629,
+      "step": 10539
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0377846359162565e-06,
+      "loss": 0.273,
+      "step": 10540
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0361164400700644e-06,
+      "loss": 0.2764,
+      "step": 10541
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.034448244223872e-06,
+      "loss": 0.2628,
+      "step": 10542
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0327800483776798e-06,
+      "loss": 0.2649,
+      "step": 10543
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0311118525314876e-06,
+      "loss": 0.2742,
+      "step": 10544
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.029443656685295e-06,
+      "loss": 0.2816,
+      "step": 10545
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.027775460839103e-06,
+      "loss": 0.2628,
+      "step": 10546
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.02610726499291e-06,
+      "loss": 0.2565,
+      "step": 10547
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.024439069146718e-06,
+      "loss": 0.273,
+      "step": 10548
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.022770873300526e-06,
+      "loss": 0.2923,
+      "step": 10549
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0211026774543333e-06,
+      "loss": 0.2394,
+      "step": 10550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.019434481608141e-06,
+      "loss": 0.2697,
+      "step": 10551
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.017766285761949e-06,
+      "loss": 0.2755,
+      "step": 10552
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0160980899157565e-06,
+      "loss": 0.2644,
+      "step": 10553
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.014429894069564e-06,
+      "loss": 0.2705,
+      "step": 10554
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0127616982233714e-06,
+      "loss": 0.2671,
+      "step": 10555
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0110935023771793e-06,
+      "loss": 0.257,
+      "step": 10556
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0094253065309868e-06,
+      "loss": 0.2889,
+      "step": 10557
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0077571106847947e-06,
+      "loss": 0.2952,
+      "step": 10558
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0060889148386026e-06,
+      "loss": 0.2574,
+      "step": 10559
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.00442071899241e-06,
+      "loss": 0.2548,
+      "step": 10560
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.0027525231462175e-06,
+      "loss": 0.2807,
+      "step": 10561
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.001084327300025e-06,
+      "loss": 0.2613,
+      "step": 10562
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.999416131453833e-06,
+      "loss": 0.2672,
+      "step": 10563
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9977479356076407e-06,
+      "loss": 0.2585,
+      "step": 10564
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.996079739761448e-06,
+      "loss": 0.2849,
+      "step": 10565
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.994411543915256e-06,
+      "loss": 0.2701,
+      "step": 10566
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9927433480690635e-06,
+      "loss": 0.2806,
+      "step": 10567
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9910751522228714e-06,
+      "loss": 0.2806,
+      "step": 10568
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.9894069563766785e-06,
+      "loss": 0.2809,
+      "step": 10569
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9877387605304864e-06,
+      "loss": 0.2665,
+      "step": 10570
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9860705646842942e-06,
+      "loss": 0.275,
+      "step": 10571
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9844023688381017e-06,
+      "loss": 0.2838,
+      "step": 10572
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9827341729919096e-06,
+      "loss": 0.2489,
+      "step": 10573
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9810659771457175e-06,
+      "loss": 0.2469,
+      "step": 10574
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.979397781299525e-06,
+      "loss": 0.2646,
+      "step": 10575
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9777295854533324e-06,
+      "loss": 0.2634,
+      "step": 10576
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.97606138960714e-06,
+      "loss": 0.2703,
+      "step": 10577
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9743931937609478e-06,
+      "loss": 0.2756,
+      "step": 10578
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9727249979147556e-06,
+      "loss": 0.265,
+      "step": 10579
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.971056802068563e-06,
+      "loss": 0.2569,
+      "step": 10580
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.969388606222371e-06,
+      "loss": 0.263,
+      "step": 10581
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9677204103761785e-06,
+      "loss": 0.2697,
+      "step": 10582
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.966052214529986e-06,
+      "loss": 0.2671,
+      "step": 10583
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9643840186837934e-06,
+      "loss": 0.256,
+      "step": 10584
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9627158228376013e-06,
+      "loss": 0.2647,
+      "step": 10585
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.961047626991409e-06,
+      "loss": 0.2915,
+      "step": 10586
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9593794311452166e-06,
+      "loss": 0.2681,
+      "step": 10587
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9577112352990245e-06,
+      "loss": 0.2558,
+      "step": 10588
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9560430394528324e-06,
+      "loss": 0.2803,
+      "step": 10589
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9543748436066394e-06,
+      "loss": 0.2613,
+      "step": 10590
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9527066477604473e-06,
+      "loss": 0.2858,
+      "step": 10591
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9510384519142548e-06,
+      "loss": 0.265,
+      "step": 10592
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9493702560680627e-06,
+      "loss": 0.2562,
+      "step": 10593
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.94770206022187e-06,
+      "loss": 0.2515,
+      "step": 10594
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.946033864375678e-06,
+      "loss": 0.2495,
+      "step": 10595
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.944365668529486e-06,
+      "loss": 0.2744,
+      "step": 10596
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9426974726832934e-06,
+      "loss": 0.2688,
+      "step": 10597
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.941029276837101e-06,
+      "loss": 0.2704,
+      "step": 10598
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9393610809909083e-06,
+      "loss": 0.2625,
+      "step": 10599
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.937692885144716e-06,
+      "loss": 0.2727,
+      "step": 10600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.936024689298524e-06,
+      "loss": 0.2646,
+      "step": 10601
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9343564934523315e-06,
+      "loss": 0.2665,
+      "step": 10602
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9326882976061394e-06,
+      "loss": 0.2798,
+      "step": 10603
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9310201017599473e-06,
+      "loss": 0.267,
+      "step": 10604
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9293519059137544e-06,
+      "loss": 0.2613,
+      "step": 10605
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.927683710067562e-06,
+      "loss": 0.2802,
+      "step": 10606
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9260155142213697e-06,
+      "loss": 0.278,
+      "step": 10607
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.9243473183751776e-06,
+      "loss": 0.2963,
+      "step": 10608
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.922679122528985e-06,
+      "loss": 0.2619,
+      "step": 10609
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.921010926682793e-06,
+      "loss": 0.2673,
+      "step": 10610
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.919342730836601e-06,
+      "loss": 0.2627,
+      "step": 10611
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.917674534990408e-06,
+      "loss": 0.2746,
+      "step": 10612
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9160063391442158e-06,
+      "loss": 0.2637,
+      "step": 10613
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9143381432980232e-06,
+      "loss": 0.2619,
+      "step": 10614
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.912669947451831e-06,
+      "loss": 0.2611,
+      "step": 10615
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.911001751605639e-06,
+      "loss": 0.2725,
+      "step": 10616
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9093335557594465e-06,
+      "loss": 0.2529,
+      "step": 10617
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9076653599132543e-06,
+      "loss": 0.2862,
+      "step": 10618
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9059971640670614e-06,
+      "loss": 0.2749,
+      "step": 10619
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9043289682208693e-06,
+      "loss": 0.259,
+      "step": 10620
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9026607723746767e-06,
+      "loss": 0.2939,
+      "step": 10621
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.9009925765284846e-06,
+      "loss": 0.2745,
+      "step": 10622
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8993243806822925e-06,
+      "loss": 0.2863,
+      "step": 10623
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8976561848361e-06,
+      "loss": 0.2521,
+      "step": 10624
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.895987988989908e-06,
+      "loss": 0.2493,
+      "step": 10625
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8943197931437157e-06,
+      "loss": 0.2855,
+      "step": 10626
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8926515972975228e-06,
+      "loss": 0.283,
+      "step": 10627
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8909834014513307e-06,
+      "loss": 0.2832,
+      "step": 10628
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.889315205605138e-06,
+      "loss": 0.2844,
+      "step": 10629
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.887647009758946e-06,
+      "loss": 0.2648,
+      "step": 10630
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8859788139127535e-06,
+      "loss": 0.293,
+      "step": 10631
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8843106180665614e-06,
+      "loss": 0.262,
+      "step": 10632
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8826424222203693e-06,
+      "loss": 0.2603,
+      "step": 10633
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8809742263741763e-06,
+      "loss": 0.2564,
+      "step": 10634
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.879306030527984e-06,
+      "loss": 0.2699,
+      "step": 10635
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8776378346817917e-06,
+      "loss": 0.2849,
+      "step": 10636
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8759696388355995e-06,
+      "loss": 0.2573,
+      "step": 10637
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8743014429894074e-06,
+      "loss": 0.27,
+      "step": 10638
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.872633247143215e-06,
+      "loss": 0.2548,
+      "step": 10639
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8709650512970228e-06,
+      "loss": 0.2626,
+      "step": 10640
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.86929685545083e-06,
+      "loss": 0.2555,
+      "step": 10641
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8676286596046377e-06,
+      "loss": 0.2583,
+      "step": 10642
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8659604637584456e-06,
+      "loss": 0.2891,
+      "step": 10643
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.864292267912253e-06,
+      "loss": 0.2739,
+      "step": 10644
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.862624072066061e-06,
+      "loss": 0.2922,
+      "step": 10645
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8609558762198684e-06,
+      "loss": 0.2622,
+      "step": 10646
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8592876803736763e-06,
+      "loss": 0.2847,
+      "step": 10647
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8576194845274833e-06,
+      "loss": 0.2489,
+      "step": 10648
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8559512886812912e-06,
+      "loss": 0.2903,
+      "step": 10649
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.854283092835099e-06,
+      "loss": 0.2789,
+      "step": 10650
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.8526148969889066e-06,
+      "loss": 0.2748,
+      "step": 10651
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8509467011427145e-06,
+      "loss": 0.2553,
+      "step": 10652
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8492785052965223e-06,
+      "loss": 0.2646,
+      "step": 10653
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.84761030945033e-06,
+      "loss": 0.2559,
+      "step": 10654
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8459421136041377e-06,
+      "loss": 0.2489,
+      "step": 10655
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8442739177579447e-06,
+      "loss": 0.2598,
+      "step": 10656
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8426057219117526e-06,
+      "loss": 0.2502,
+      "step": 10657
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.84093752606556e-06,
+      "loss": 0.2863,
+      "step": 10658
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.839269330219368e-06,
+      "loss": 0.2741,
+      "step": 10659
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.837601134373176e-06,
+      "loss": 0.2775,
+      "step": 10660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8359329385269833e-06,
+      "loss": 0.2772,
+      "step": 10661
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.834264742680791e-06,
+      "loss": 0.2653,
+      "step": 10662
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8325965468345982e-06,
+      "loss": 0.2854,
+      "step": 10663
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.830928350988406e-06,
+      "loss": 0.281,
+      "step": 10664
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.829260155142214e-06,
+      "loss": 0.2687,
+      "step": 10665
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8275919592960215e-06,
+      "loss": 0.2565,
+      "step": 10666
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8259237634498294e-06,
+      "loss": 0.2769,
+      "step": 10667
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8242555676036373e-06,
+      "loss": 0.2585,
+      "step": 10668
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8225873717574447e-06,
+      "loss": 0.2696,
+      "step": 10669
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8209191759112518e-06,
+      "loss": 0.286,
+      "step": 10670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8192509800650597e-06,
+      "loss": 0.2813,
+      "step": 10671
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8175827842188675e-06,
+      "loss": 0.2687,
+      "step": 10672
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.815914588372675e-06,
+      "loss": 0.2683,
+      "step": 10673
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.814246392526483e-06,
+      "loss": 0.2783,
+      "step": 10674
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8125781966802908e-06,
+      "loss": 0.2697,
+      "step": 10675
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8109100008340982e-06,
+      "loss": 0.2688,
+      "step": 10676
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8092418049879057e-06,
+      "loss": 0.2588,
+      "step": 10677
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.807573609141713e-06,
+      "loss": 0.2741,
+      "step": 10678
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.805905413295521e-06,
+      "loss": 0.2738,
+      "step": 10679
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.804237217449329e-06,
+      "loss": 0.2614,
+      "step": 10680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8025690216031364e-06,
+      "loss": 0.2876,
+      "step": 10681
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.8009008257569443e-06,
+      "loss": 0.2693,
+      "step": 10682
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7992326299107518e-06,
+      "loss": 0.2629,
+      "step": 10683
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7975644340645592e-06,
+      "loss": 0.2616,
+      "step": 10684
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7958962382183667e-06,
+      "loss": 0.2621,
+      "step": 10685
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7942280423721746e-06,
+      "loss": 0.2671,
+      "step": 10686
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7925598465259825e-06,
+      "loss": 0.2632,
+      "step": 10687
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.79089165067979e-06,
+      "loss": 0.2511,
+      "step": 10688
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.789223454833598e-06,
+      "loss": 0.28,
+      "step": 10689
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7875552589874057e-06,
+      "loss": 0.2446,
+      "step": 10690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.785887063141213e-06,
+      "loss": 0.2616,
+      "step": 10691
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.7842188672950206e-06,
+      "loss": 0.278,
+      "step": 10692
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.782550671448828e-06,
+      "loss": 0.2747,
+      "step": 10693
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.780882475602636e-06,
+      "loss": 0.2623,
+      "step": 10694
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7792142797564434e-06,
+      "loss": 0.278,
+      "step": 10695
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7775460839102513e-06,
+      "loss": 0.2815,
+      "step": 10696
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.775877888064059e-06,
+      "loss": 0.2631,
+      "step": 10697
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7742096922178667e-06,
+      "loss": 0.2751,
+      "step": 10698
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.772541496371674e-06,
+      "loss": 0.2776,
+      "step": 10699
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7708733005254816e-06,
+      "loss": 0.2747,
+      "step": 10700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7692051046792895e-06,
+      "loss": 0.2664,
+      "step": 10701
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7675369088330974e-06,
+      "loss": 0.2836,
+      "step": 10702
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.765868712986905e-06,
+      "loss": 0.2528,
+      "step": 10703
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7642005171407127e-06,
+      "loss": 0.2667,
+      "step": 10704
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7625323212945206e-06,
+      "loss": 0.2826,
+      "step": 10705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7608641254483277e-06,
+      "loss": 0.2601,
+      "step": 10706
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7591959296021355e-06,
+      "loss": 0.2678,
+      "step": 10707
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.757527733755943e-06,
+      "loss": 0.2551,
+      "step": 10708
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.755859537909751e-06,
+      "loss": 0.2796,
+      "step": 10709
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7541913420635584e-06,
+      "loss": 0.2393,
+      "step": 10710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7525231462173662e-06,
+      "loss": 0.2686,
+      "step": 10711
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.750854950371174e-06,
+      "loss": 0.3004,
+      "step": 10712
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.749186754524981e-06,
+      "loss": 0.2677,
+      "step": 10713
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.747518558678789e-06,
+      "loss": 0.257,
+      "step": 10714
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7458503628325965e-06,
+      "loss": 0.2614,
+      "step": 10715
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7441821669864044e-06,
+      "loss": 0.2651,
+      "step": 10716
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7425139711402123e-06,
+      "loss": 0.2492,
+      "step": 10717
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7408457752940198e-06,
+      "loss": 0.2689,
+      "step": 10718
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7391775794478276e-06,
+      "loss": 0.2681,
+      "step": 10719
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.737509383601635e-06,
+      "loss": 0.2607,
+      "step": 10720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7358411877554426e-06,
+      "loss": 0.2928,
+      "step": 10721
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.73417299190925e-06,
+      "loss": 0.2559,
+      "step": 10722
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.732504796063058e-06,
+      "loss": 0.2756,
+      "step": 10723
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.730836600216866e-06,
+      "loss": 0.2654,
+      "step": 10724
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7291684043706733e-06,
+      "loss": 0.2572,
+      "step": 10725
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.727500208524481e-06,
+      "loss": 0.2759,
+      "step": 10726
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.725832012678289e-06,
+      "loss": 0.2634,
+      "step": 10727
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.724163816832096e-06,
+      "loss": 0.2753,
+      "step": 10728
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.722495620985904e-06,
+      "loss": 0.2668,
+      "step": 10729
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7208274251397114e-06,
+      "loss": 0.2534,
+      "step": 10730
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7191592292935193e-06,
+      "loss": 0.2895,
+      "step": 10731
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.717491033447327e-06,
+      "loss": 0.259,
+      "step": 10732
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.7158228376011347e-06,
+      "loss": 0.2385,
+      "step": 10733
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7141546417549426e-06,
+      "loss": 0.2644,
+      "step": 10734
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7124864459087496e-06,
+      "loss": 0.2865,
+      "step": 10735
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7108182500625575e-06,
+      "loss": 0.264,
+      "step": 10736
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.709150054216365e-06,
+      "loss": 0.26,
+      "step": 10737
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.707481858370173e-06,
+      "loss": 0.263,
+      "step": 10738
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.7058136625239807e-06,
+      "loss": 0.2733,
+      "step": 10739
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.704145466677788e-06,
+      "loss": 0.2634,
+      "step": 10740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.702477270831596e-06,
+      "loss": 0.2747,
+      "step": 10741
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.700809074985403e-06,
+      "loss": 0.272,
+      "step": 10742
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.699140879139211e-06,
+      "loss": 0.269,
+      "step": 10743
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.697472683293019e-06,
+      "loss": 0.269,
+      "step": 10744
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6958044874468264e-06,
+      "loss": 0.2712,
+      "step": 10745
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6941362916006342e-06,
+      "loss": 0.2791,
+      "step": 10746
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6924680957544417e-06,
+      "loss": 0.2582,
+      "step": 10747
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6907998999082496e-06,
+      "loss": 0.2626,
+      "step": 10748
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6891317040620575e-06,
+      "loss": 0.2837,
+      "step": 10749
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6874635082158645e-06,
+      "loss": 0.2705,
+      "step": 10750
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6857953123696724e-06,
+      "loss": 0.2679,
+      "step": 10751
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.68412711652348e-06,
+      "loss": 0.2601,
+      "step": 10752
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6824589206772878e-06,
+      "loss": 0.2628,
+      "step": 10753
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6807907248310956e-06,
+      "loss": 0.2537,
+      "step": 10754
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.679122528984903e-06,
+      "loss": 0.2749,
+      "step": 10755
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.677454333138711e-06,
+      "loss": 0.2833,
+      "step": 10756
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.675786137292518e-06,
+      "loss": 0.2631,
+      "step": 10757
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.674117941446326e-06,
+      "loss": 0.269,
+      "step": 10758
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6724497456001334e-06,
+      "loss": 0.2495,
+      "step": 10759
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6707815497539413e-06,
+      "loss": 0.263,
+      "step": 10760
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.669113353907749e-06,
+      "loss": 0.2957,
+      "step": 10761
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6674451580615566e-06,
+      "loss": 0.2816,
+      "step": 10762
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6657769622153645e-06,
+      "loss": 0.2832,
+      "step": 10763
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6641087663691715e-06,
+      "loss": 0.2807,
+      "step": 10764
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6624405705229794e-06,
+      "loss": 0.2577,
+      "step": 10765
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6607723746767873e-06,
+      "loss": 0.2715,
+      "step": 10766
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6591041788305948e-06,
+      "loss": 0.2737,
+      "step": 10767
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6574359829844027e-06,
+      "loss": 0.2638,
+      "step": 10768
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6557677871382106e-06,
+      "loss": 0.2817,
+      "step": 10769
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.654099591292018e-06,
+      "loss": 0.2783,
+      "step": 10770
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6524313954458255e-06,
+      "loss": 0.2707,
+      "step": 10771
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.650763199599633e-06,
+      "loss": 0.2757,
+      "step": 10772
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.649095003753441e-06,
+      "loss": 0.2698,
+      "step": 10773
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.6474268079072483e-06,
+      "loss": 0.2855,
+      "step": 10774
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.645758612061056e-06,
+      "loss": 0.2718,
+      "step": 10775
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.644090416214864e-06,
+      "loss": 0.2722,
+      "step": 10776
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6424222203686715e-06,
+      "loss": 0.2562,
+      "step": 10777
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6407540245224794e-06,
+      "loss": 0.2795,
+      "step": 10778
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6390858286762865e-06,
+      "loss": 0.2648,
+      "step": 10779
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6374176328300944e-06,
+      "loss": 0.2773,
+      "step": 10780
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6357494369839022e-06,
+      "loss": 0.2552,
+      "step": 10781
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6340812411377097e-06,
+      "loss": 0.2733,
+      "step": 10782
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6324130452915176e-06,
+      "loss": 0.2704,
+      "step": 10783
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.630744849445325e-06,
+      "loss": 0.2906,
+      "step": 10784
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.629076653599133e-06,
+      "loss": 0.2693,
+      "step": 10785
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.62740845775294e-06,
+      "loss": 0.2696,
+      "step": 10786
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.625740261906748e-06,
+      "loss": 0.2518,
+      "step": 10787
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6240720660605558e-06,
+      "loss": 0.2801,
+      "step": 10788
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6224038702143632e-06,
+      "loss": 0.2852,
+      "step": 10789
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.620735674368171e-06,
+      "loss": 0.2662,
+      "step": 10790
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.619067478521979e-06,
+      "loss": 0.2508,
+      "step": 10791
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6173992826757865e-06,
+      "loss": 0.2692,
+      "step": 10792
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.615731086829594e-06,
+      "loss": 0.2618,
+      "step": 10793
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6140628909834014e-06,
+      "loss": 0.2801,
+      "step": 10794
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6123946951372093e-06,
+      "loss": 0.2564,
+      "step": 10795
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.610726499291017e-06,
+      "loss": 0.2885,
+      "step": 10796
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6090583034448246e-06,
+      "loss": 0.2647,
+      "step": 10797
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6073901075986325e-06,
+      "loss": 0.2591,
+      "step": 10798
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.60572191175244e-06,
+      "loss": 0.2622,
+      "step": 10799
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6040537159062474e-06,
+      "loss": 0.2715,
+      "step": 10800
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.602385520060055e-06,
+      "loss": 0.2535,
+      "step": 10801
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.6007173242138628e-06,
+      "loss": 0.2686,
+      "step": 10802
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5990491283676707e-06,
+      "loss": 0.2448,
+      "step": 10803
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.597380932521478e-06,
+      "loss": 0.2685,
+      "step": 10804
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.595712736675286e-06,
+      "loss": 0.2604,
+      "step": 10805
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.594044540829094e-06,
+      "loss": 0.2891,
+      "step": 10806
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5923763449829014e-06,
+      "loss": 0.2634,
+      "step": 10807
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.590708149136709e-06,
+      "loss": 0.289,
+      "step": 10808
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5890399532905163e-06,
+      "loss": 0.2599,
+      "step": 10809
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.587371757444324e-06,
+      "loss": 0.2597,
+      "step": 10810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5857035615981317e-06,
+      "loss": 0.251,
+      "step": 10811
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5840353657519395e-06,
+      "loss": 0.2457,
+      "step": 10812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5823671699057474e-06,
+      "loss": 0.2674,
+      "step": 10813
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.580698974059555e-06,
+      "loss": 0.261,
+      "step": 10814
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.5790307782133624e-06,
+      "loss": 0.2845,
+      "step": 10815
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.57736258236717e-06,
+      "loss": 0.279,
+      "step": 10816
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5756943865209777e-06,
+      "loss": 0.2823,
+      "step": 10817
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5740261906747856e-06,
+      "loss": 0.259,
+      "step": 10818
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.572357994828593e-06,
+      "loss": 0.254,
+      "step": 10819
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.570689798982401e-06,
+      "loss": 0.2596,
+      "step": 10820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.569021603136209e-06,
+      "loss": 0.2604,
+      "step": 10821
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.567353407290016e-06,
+      "loss": 0.2558,
+      "step": 10822
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5656852114438233e-06,
+      "loss": 0.2677,
+      "step": 10823
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5640170155976312e-06,
+      "loss": 0.2628,
+      "step": 10824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.562348819751439e-06,
+      "loss": 0.2721,
+      "step": 10825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5606806239052466e-06,
+      "loss": 0.2575,
+      "step": 10826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5590124280590545e-06,
+      "loss": 0.2662,
+      "step": 10827
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5573442322128623e-06,
+      "loss": 0.2686,
+      "step": 10828
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5556760363666694e-06,
+      "loss": 0.2534,
+      "step": 10829
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5540078405204773e-06,
+      "loss": 0.2551,
+      "step": 10830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5523396446742847e-06,
+      "loss": 0.2758,
+      "step": 10831
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5506714488280926e-06,
+      "loss": 0.2695,
+      "step": 10832
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5490032529819005e-06,
+      "loss": 0.2541,
+      "step": 10833
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.547335057135708e-06,
+      "loss": 0.2602,
+      "step": 10834
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.545666861289516e-06,
+      "loss": 0.2756,
+      "step": 10835
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5439986654433233e-06,
+      "loss": 0.2711,
+      "step": 10836
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5423304695971308e-06,
+      "loss": 0.2787,
+      "step": 10837
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5406622737509383e-06,
+      "loss": 0.2767,
+      "step": 10838
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.538994077904746e-06,
+      "loss": 0.2754,
+      "step": 10839
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.537325882058554e-06,
+      "loss": 0.2442,
+      "step": 10840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5356576862123615e-06,
+      "loss": 0.2815,
+      "step": 10841
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5339894903661694e-06,
+      "loss": 0.2685,
+      "step": 10842
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5323212945199773e-06,
+      "loss": 0.2782,
+      "step": 10843
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5306530986737843e-06,
+      "loss": 0.2647,
+      "step": 10844
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.528984902827592e-06,
+      "loss": 0.2919,
+      "step": 10845
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5273167069813997e-06,
+      "loss": 0.2754,
+      "step": 10846
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5256485111352075e-06,
+      "loss": 0.2887,
+      "step": 10847
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.523980315289015e-06,
+      "loss": 0.265,
+      "step": 10848
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.522312119442823e-06,
+      "loss": 0.2661,
+      "step": 10849
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5206439235966308e-06,
+      "loss": 0.2715,
+      "step": 10850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.518975727750438e-06,
+      "loss": 0.2739,
+      "step": 10851
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5173075319042457e-06,
+      "loss": 0.2782,
+      "step": 10852
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.515639336058053e-06,
+      "loss": 0.2706,
+      "step": 10853
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.513971140211861e-06,
+      "loss": 0.278,
+      "step": 10854
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.512302944365669e-06,
+      "loss": 0.2804,
+      "step": 10855
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5106347485194764e-06,
+      "loss": 0.2722,
+      "step": 10856
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.5089665526732843e-06,
+      "loss": 0.2629,
+      "step": 10857
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5072983568270913e-06,
+      "loss": 0.2538,
+      "step": 10858
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5056301609808992e-06,
+      "loss": 0.2558,
+      "step": 10859
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.503961965134707e-06,
+      "loss": 0.2622,
+      "step": 10860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5022937692885146e-06,
+      "loss": 0.2806,
+      "step": 10861
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.5006255734423225e-06,
+      "loss": 0.269,
+      "step": 10862
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.49895737759613e-06,
+      "loss": 0.2788,
+      "step": 10863
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4972891817499374e-06,
+      "loss": 0.2681,
+      "step": 10864
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4956209859037453e-06,
+      "loss": 0.2729,
+      "step": 10865
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.493952790057553e-06,
+      "loss": 0.2757,
+      "step": 10866
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4922845942113606e-06,
+      "loss": 0.2771,
+      "step": 10867
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.490616398365168e-06,
+      "loss": 0.2728,
+      "step": 10868
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.488948202518976e-06,
+      "loss": 0.277,
+      "step": 10869
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.487280006672784e-06,
+      "loss": 0.2562,
+      "step": 10870
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4856118108265913e-06,
+      "loss": 0.2927,
+      "step": 10871
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4839436149803988e-06,
+      "loss": 0.277,
+      "step": 10872
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4822754191342067e-06,
+      "loss": 0.27,
+      "step": 10873
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.480607223288014e-06,
+      "loss": 0.2863,
+      "step": 10874
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4789390274418216e-06,
+      "loss": 0.2764,
+      "step": 10875
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4772708315956295e-06,
+      "loss": 0.269,
+      "step": 10876
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4756026357494374e-06,
+      "loss": 0.2815,
+      "step": 10877
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.473934439903245e-06,
+      "loss": 0.2689,
+      "step": 10878
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4722662440570523e-06,
+      "loss": 0.2793,
+      "step": 10879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.47059804821086e-06,
+      "loss": 0.2696,
+      "step": 10880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.468929852364668e-06,
+      "loss": 0.2733,
+      "step": 10881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4672616565184755e-06,
+      "loss": 0.2837,
+      "step": 10882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.465593460672283e-06,
+      "loss": 0.2782,
+      "step": 10883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.463925264826091e-06,
+      "loss": 0.26,
+      "step": 10884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4622570689798988e-06,
+      "loss": 0.2697,
+      "step": 10885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4605888731337062e-06,
+      "loss": 0.2814,
+      "step": 10886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4589206772875137e-06,
+      "loss": 0.2782,
+      "step": 10887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4572524814413216e-06,
+      "loss": 0.28,
+      "step": 10888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.455584285595129e-06,
+      "loss": 0.2802,
+      "step": 10889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4539160897489365e-06,
+      "loss": 0.2838,
+      "step": 10890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4522478939027444e-06,
+      "loss": 0.2616,
+      "step": 10891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4505796980565523e-06,
+      "loss": 0.2511,
+      "step": 10892
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4489115022103598e-06,
+      "loss": 0.27,
+      "step": 10893
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4472433063641672e-06,
+      "loss": 0.2705,
+      "step": 10894
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.445575110517975e-06,
+      "loss": 0.2652,
+      "step": 10895
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4439069146717826e-06,
+      "loss": 0.2603,
+      "step": 10896
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4422387188255905e-06,
+      "loss": 0.266,
+      "step": 10897
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.440570522979398e-06,
+      "loss": 0.2633,
+      "step": 10898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.438902327133206e-06,
+      "loss": 0.2944,
+      "step": 10899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4372341312870133e-06,
+      "loss": 0.2608,
+      "step": 10900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4355659354408207e-06,
+      "loss": 0.2674,
+      "step": 10901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4338977395946286e-06,
+      "loss": 0.2687,
+      "step": 10902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4322295437484365e-06,
+      "loss": 0.2839,
+      "step": 10903
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.430561347902244e-06,
+      "loss": 0.2521,
+      "step": 10904
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4288931520560514e-06,
+      "loss": 0.2814,
+      "step": 10905
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4272249562098593e-06,
+      "loss": 0.2465,
+      "step": 10906
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.425556760363667e-06,
+      "loss": 0.2704,
+      "step": 10907
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4238885645174747e-06,
+      "loss": 0.2614,
+      "step": 10908
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.422220368671282e-06,
+      "loss": 0.26,
+      "step": 10909
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.42055217282509e-06,
+      "loss": 0.2867,
+      "step": 10910
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4188839769788975e-06,
+      "loss": 0.2725,
+      "step": 10911
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.417215781132705e-06,
+      "loss": 0.2654,
+      "step": 10912
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.415547585286513e-06,
+      "loss": 0.2906,
+      "step": 10913
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4138793894403203e-06,
+      "loss": 0.273,
+      "step": 10914
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.412211193594128e-06,
+      "loss": 0.2599,
+      "step": 10915
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4105429977479357e-06,
+      "loss": 0.2739,
+      "step": 10916
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4088748019017435e-06,
+      "loss": 0.2682,
+      "step": 10917
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.407206606055551e-06,
+      "loss": 0.2744,
+      "step": 10918
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.405538410209359e-06,
+      "loss": 0.2702,
+      "step": 10919
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4038702143631664e-06,
+      "loss": 0.2708,
+      "step": 10920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4022020185169742e-06,
+      "loss": 0.2613,
+      "step": 10921
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.4005338226707817e-06,
+      "loss": 0.2605,
+      "step": 10922
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3988656268245896e-06,
+      "loss": 0.262,
+      "step": 10923
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.397197430978397e-06,
+      "loss": 0.2781,
+      "step": 10924
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3955292351322045e-06,
+      "loss": 0.2775,
+      "step": 10925
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3938610392860124e-06,
+      "loss": 0.2828,
+      "step": 10926
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.39219284343982e-06,
+      "loss": 0.3013,
+      "step": 10927
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3905246475936278e-06,
+      "loss": 0.2467,
+      "step": 10928
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3888564517474352e-06,
+      "loss": 0.257,
+      "step": 10929
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.387188255901243e-06,
+      "loss": 0.2679,
+      "step": 10930
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3855200600550506e-06,
+      "loss": 0.272,
+      "step": 10931
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3838518642088585e-06,
+      "loss": 0.262,
+      "step": 10932
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.382183668362666e-06,
+      "loss": 0.2585,
+      "step": 10933
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380515472516474e-06,
+      "loss": 0.261,
+      "step": 10934
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3788472766702813e-06,
+      "loss": 0.2659,
+      "step": 10935
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3771790808240887e-06,
+      "loss": 0.2636,
+      "step": 10936
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3755108849778966e-06,
+      "loss": 0.2458,
+      "step": 10937
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.373842689131704e-06,
+      "loss": 0.2834,
+      "step": 10938
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.372174493285512e-06,
+      "loss": 0.2795,
+      "step": 10939
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3705062974393194e-06,
+      "loss": 0.277,
+      "step": 10940
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3688381015931273e-06,
+      "loss": 0.2515,
+      "step": 10941
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.367169905746935e-06,
+      "loss": 0.2518,
+      "step": 10942
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3655017099007423e-06,
+      "loss": 0.2738,
+      "step": 10943
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.36383351405455e-06,
+      "loss": 0.2496,
+      "step": 10944
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.362165318208358e-06,
+      "loss": 0.2542,
+      "step": 10945
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3604971223621655e-06,
+      "loss": 0.272,
+      "step": 10946
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.358828926515973e-06,
+      "loss": 0.2714,
+      "step": 10947
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.357160730669781e-06,
+      "loss": 0.2555,
+      "step": 10948
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3554925348235887e-06,
+      "loss": 0.2625,
+      "step": 10949
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.353824338977396e-06,
+      "loss": 0.2745,
+      "step": 10950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3521561431312037e-06,
+      "loss": 0.2648,
+      "step": 10951
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3504879472850115e-06,
+      "loss": 0.2818,
+      "step": 10952
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.348819751438819e-06,
+      "loss": 0.2709,
+      "step": 10953
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3471515555926265e-06,
+      "loss": 0.2543,
+      "step": 10954
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3454833597464344e-06,
+      "loss": 0.2571,
+      "step": 10955
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3438151639002422e-06,
+      "loss": 0.2804,
+      "step": 10956
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3421469680540497e-06,
+      "loss": 0.2584,
+      "step": 10957
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.340478772207857e-06,
+      "loss": 0.2768,
+      "step": 10958
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.338810576361665e-06,
+      "loss": 0.2575,
+      "step": 10959
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.337142380515473e-06,
+      "loss": 0.2582,
+      "step": 10960
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3354741846692804e-06,
+      "loss": 0.2648,
+      "step": 10961
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.333805988823088e-06,
+      "loss": 0.2629,
+      "step": 10962
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3321377929768958e-06,
+      "loss": 0.2697,
+      "step": 10963
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3304695971307032e-06,
+      "loss": 0.2665,
+      "step": 10964
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3288014012845107e-06,
+      "loss": 0.2665,
+      "step": 10965
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3271332054383186e-06,
+      "loss": 0.2544,
+      "step": 10966
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3254650095921265e-06,
+      "loss": 0.26,
+      "step": 10967
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.323796813745934e-06,
+      "loss": 0.2567,
+      "step": 10968
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3221286178997414e-06,
+      "loss": 0.2555,
+      "step": 10969
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3204604220535493e-06,
+      "loss": 0.2593,
+      "step": 10970
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.318792226207357e-06,
+      "loss": 0.2638,
+      "step": 10971
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3171240303611646e-06,
+      "loss": 0.25,
+      "step": 10972
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.315455834514972e-06,
+      "loss": 0.271,
+      "step": 10973
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.31378763866878e-06,
+      "loss": 0.2578,
+      "step": 10974
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.312119442822588e-06,
+      "loss": 0.2751,
+      "step": 10975
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.310451246976395e-06,
+      "loss": 0.2745,
+      "step": 10976
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.308783051130203e-06,
+      "loss": 0.2758,
+      "step": 10977
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3071148552840107e-06,
+      "loss": 0.2735,
+      "step": 10978
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.305446659437818e-06,
+      "loss": 0.2587,
+      "step": 10979
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3037784635916256e-06,
+      "loss": 0.2688,
+      "step": 10980
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3021102677454335e-06,
+      "loss": 0.2625,
+      "step": 10981
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3004420718992414e-06,
+      "loss": 0.2715,
+      "step": 10982
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.298773876053049e-06,
+      "loss": 0.2731,
+      "step": 10983
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2971056802068563e-06,
+      "loss": 0.2726,
+      "step": 10984
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.295437484360664e-06,
+      "loss": 0.2543,
+      "step": 10985
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.293769288514472e-06,
+      "loss": 0.2815,
+      "step": 10986
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2921010926682795e-06,
+      "loss": 0.2654,
+      "step": 10987
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.290432896822087e-06,
+      "loss": 0.2952,
+      "step": 10988
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.288764700975895e-06,
+      "loss": 0.2796,
+      "step": 10989
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2870965051297024e-06,
+      "loss": 0.2449,
+      "step": 10990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.28542830928351e-06,
+      "loss": 0.2615,
+      "step": 10991
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2837601134373177e-06,
+      "loss": 0.2782,
+      "step": 10992
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2820919175911256e-06,
+      "loss": 0.2683,
+      "step": 10993
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.280423721744933e-06,
+      "loss": 0.2651,
+      "step": 10994
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2787555258987405e-06,
+      "loss": 0.2829,
+      "step": 10995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2770873300525484e-06,
+      "loss": 0.2534,
+      "step": 10996
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2754191342063563e-06,
+      "loss": 0.2784,
+      "step": 10997
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2737509383601638e-06,
+      "loss": 0.2637,
+      "step": 10998
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2720827425139712e-06,
+      "loss": 0.2729,
+      "step": 10999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.270414546667779e-06,
+      "loss": 0.2824,
+      "step": 11000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2687463508215866e-06,
+      "loss": 0.2561,
+      "step": 11001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.267078154975394e-06,
+      "loss": 0.2927,
+      "step": 11002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.265409959129202e-06,
+      "loss": 0.2647,
+      "step": 11003
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.26374176328301e-06,
+      "loss": 0.2601,
+      "step": 11004
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2620735674368173e-06,
+      "loss": 0.2709,
+      "step": 11005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2604053715906247e-06,
+      "loss": 0.2755,
+      "step": 11006
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2587371757444326e-06,
+      "loss": 0.2705,
+      "step": 11007
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2570689798982405e-06,
+      "loss": 0.3015,
+      "step": 11008
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.255400784052048e-06,
+      "loss": 0.2845,
+      "step": 11009
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2537325882058554e-06,
+      "loss": 0.2774,
+      "step": 11010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2520643923596633e-06,
+      "loss": 0.2661,
+      "step": 11011
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.250396196513471e-06,
+      "loss": 0.2703,
+      "step": 11012
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2487280006672787e-06,
+      "loss": 0.2753,
+      "step": 11013
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.247059804821086e-06,
+      "loss": 0.2545,
+      "step": 11014
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.245391608974894e-06,
+      "loss": 0.2813,
+      "step": 11015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2437234131287015e-06,
+      "loss": 0.2714,
+      "step": 11016
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.242055217282509e-06,
+      "loss": 0.2706,
+      "step": 11017
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.240387021436317e-06,
+      "loss": 0.2748,
+      "step": 11018
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2387188255901247e-06,
+      "loss": 0.2728,
+      "step": 11019
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.237050629743932e-06,
+      "loss": 0.2628,
+      "step": 11020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2353824338977397e-06,
+      "loss": 0.2669,
+      "step": 11021
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.2337142380515475e-06,
+      "loss": 0.276,
+      "step": 11022
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.232046042205355e-06,
+      "loss": 0.283,
+      "step": 11023
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.230377846359163e-06,
+      "loss": 0.2709,
+      "step": 11024
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2287096505129704e-06,
+      "loss": 0.2666,
+      "step": 11025
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2270414546667782e-06,
+      "loss": 0.2652,
+      "step": 11026
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2253732588205857e-06,
+      "loss": 0.276,
+      "step": 11027
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.223705062974393e-06,
+      "loss": 0.2735,
+      "step": 11028
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.222036867128201e-06,
+      "loss": 0.2853,
+      "step": 11029
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2203686712820085e-06,
+      "loss": 0.2533,
+      "step": 11030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2187004754358164e-06,
+      "loss": 0.2788,
+      "step": 11031
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.217032279589624e-06,
+      "loss": 0.2736,
+      "step": 11032
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2153640837434318e-06,
+      "loss": 0.2738,
+      "step": 11033
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2136958878972392e-06,
+      "loss": 0.283,
+      "step": 11034
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.212027692051047e-06,
+      "loss": 0.2794,
+      "step": 11035
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2103594962048546e-06,
+      "loss": 0.281,
+      "step": 11036
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2086913003586625e-06,
+      "loss": 0.2564,
+      "step": 11037
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.20702310451247e-06,
+      "loss": 0.2428,
+      "step": 11038
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.205354908666278e-06,
+      "loss": 0.2665,
+      "step": 11039
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2036867128200853e-06,
+      "loss": 0.2656,
+      "step": 11040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2020185169738927e-06,
+      "loss": 0.2441,
+      "step": 11041
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.2003503211277006e-06,
+      "loss": 0.2574,
+      "step": 11042
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.198682125281508e-06,
+      "loss": 0.2746,
+      "step": 11043
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.197013929435316e-06,
+      "loss": 0.2372,
+      "step": 11044
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1953457335891234e-06,
+      "loss": 0.2754,
+      "step": 11045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1936775377429313e-06,
+      "loss": 0.2829,
+      "step": 11046
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.192009341896739e-06,
+      "loss": 0.2585,
+      "step": 11047
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1903411460505467e-06,
+      "loss": 0.2724,
+      "step": 11048
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.188672950204354e-06,
+      "loss": 0.2624,
+      "step": 11049
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.187004754358162e-06,
+      "loss": 0.2607,
+      "step": 11050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1853365585119695e-06,
+      "loss": 0.274,
+      "step": 11051
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.183668362665777e-06,
+      "loss": 0.243,
+      "step": 11052
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.182000166819585e-06,
+      "loss": 0.2722,
+      "step": 11053
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1803319709733923e-06,
+      "loss": 0.2502,
+      "step": 11054
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1786637751272e-06,
+      "loss": 0.2678,
+      "step": 11055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1769955792810077e-06,
+      "loss": 0.2667,
+      "step": 11056
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1753273834348155e-06,
+      "loss": 0.2955,
+      "step": 11057
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.173659187588623e-06,
+      "loss": 0.2762,
+      "step": 11058
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1719909917424305e-06,
+      "loss": 0.2674,
+      "step": 11059
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1703227958962384e-06,
+      "loss": 0.2719,
+      "step": 11060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1686546000500462e-06,
+      "loss": 0.2756,
+      "step": 11061
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.1669864042038537e-06,
+      "loss": 0.2602,
+      "step": 11062
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.165318208357661e-06,
+      "loss": 0.2678,
+      "step": 11063
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.163650012511469e-06,
+      "loss": 0.2619,
+      "step": 11064
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1619818166652765e-06,
+      "loss": 0.2568,
+      "step": 11065
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1603136208190844e-06,
+      "loss": 0.2915,
+      "step": 11066
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.158645424972892e-06,
+      "loss": 0.2758,
+      "step": 11067
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1569772291266998e-06,
+      "loss": 0.2602,
+      "step": 11068
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1553090332805072e-06,
+      "loss": 0.2551,
+      "step": 11069
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1536408374343147e-06,
+      "loss": 0.274,
+      "step": 11070
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1519726415881226e-06,
+      "loss": 0.2785,
+      "step": 11071
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1503044457419305e-06,
+      "loss": 0.2631,
+      "step": 11072
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.148636249895738e-06,
+      "loss": 0.2564,
+      "step": 11073
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1469680540495454e-06,
+      "loss": 0.2676,
+      "step": 11074
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1452998582033533e-06,
+      "loss": 0.2611,
+      "step": 11075
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.143631662357161e-06,
+      "loss": 0.2894,
+      "step": 11076
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1419634665109686e-06,
+      "loss": 0.2713,
+      "step": 11077
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.140295270664776e-06,
+      "loss": 0.2905,
+      "step": 11078
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.138627074818584e-06,
+      "loss": 0.2693,
+      "step": 11079
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1369588789723914e-06,
+      "loss": 0.271,
+      "step": 11080
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.135290683126199e-06,
+      "loss": 0.2992,
+      "step": 11081
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.133622487280007e-06,
+      "loss": 0.2896,
+      "step": 11082
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1319542914338147e-06,
+      "loss": 0.269,
+      "step": 11083
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.130286095587622e-06,
+      "loss": 0.2789,
+      "step": 11084
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1286178997414296e-06,
+      "loss": 0.2651,
+      "step": 11085
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1269497038952375e-06,
+      "loss": 0.2835,
+      "step": 11086
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1252815080490454e-06,
+      "loss": 0.2684,
+      "step": 11087
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.123613312202853e-06,
+      "loss": 0.2721,
+      "step": 11088
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1219451163566603e-06,
+      "loss": 0.2598,
+      "step": 11089
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.120276920510468e-06,
+      "loss": 0.2684,
+      "step": 11090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1186087246642757e-06,
+      "loss": 0.2703,
+      "step": 11091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.116940528818083e-06,
+      "loss": 0.2487,
+      "step": 11092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.115272332971891e-06,
+      "loss": 0.2681,
+      "step": 11093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.113604137125699e-06,
+      "loss": 0.2781,
+      "step": 11094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1119359412795064e-06,
+      "loss": 0.2608,
+      "step": 11095
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.110267745433314e-06,
+      "loss": 0.2528,
+      "step": 11096
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1085995495871217e-06,
+      "loss": 0.2625,
+      "step": 11097
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1069313537409296e-06,
+      "loss": 0.264,
+      "step": 11098
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 0.2705,
+      "step": 11099
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1035949620485445e-06,
+      "loss": 0.2702,
+      "step": 11100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1019267662023524e-06,
+      "loss": 0.2681,
+      "step": 11101
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.1002585703561603e-06,
+      "loss": 0.2665,
+      "step": 11102
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0985903745099678e-06,
+      "loss": 0.2809,
+      "step": 11103
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.0969221786637752e-06,
+      "loss": 0.2582,
+      "step": 11104
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.095253982817583e-06,
+      "loss": 0.273,
+      "step": 11105
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0935857869713906e-06,
+      "loss": 0.2718,
+      "step": 11106
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.091917591125198e-06,
+      "loss": 0.2778,
+      "step": 11107
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.090249395279006e-06,
+      "loss": 0.2665,
+      "step": 11108
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.088581199432814e-06,
+      "loss": 0.2787,
+      "step": 11109
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0869130035866213e-06,
+      "loss": 0.2562,
+      "step": 11110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0852448077404287e-06,
+      "loss": 0.2725,
+      "step": 11111
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0835766118942366e-06,
+      "loss": 0.2796,
+      "step": 11112
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0819084160480445e-06,
+      "loss": 0.2637,
+      "step": 11113
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.080240220201852e-06,
+      "loss": 0.256,
+      "step": 11114
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0785720243556594e-06,
+      "loss": 0.2907,
+      "step": 11115
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0769038285094673e-06,
+      "loss": 0.2646,
+      "step": 11116
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.075235632663275e-06,
+      "loss": 0.269,
+      "step": 11117
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0735674368170823e-06,
+      "loss": 0.2539,
+      "step": 11118
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.07189924097089e-06,
+      "loss": 0.2724,
+      "step": 11119
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.070231045124698e-06,
+      "loss": 0.2647,
+      "step": 11120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0685628492785055e-06,
+      "loss": 0.2652,
+      "step": 11121
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.066894653432313e-06,
+      "loss": 0.2797,
+      "step": 11122
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.065226457586121e-06,
+      "loss": 0.2643,
+      "step": 11123
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0635582617399287e-06,
+      "loss": 0.269,
+      "step": 11124
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.061890065893736e-06,
+      "loss": 0.2825,
+      "step": 11125
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0602218700475437e-06,
+      "loss": 0.2733,
+      "step": 11126
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0585536742013515e-06,
+      "loss": 0.2585,
+      "step": 11127
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.056885478355159e-06,
+      "loss": 0.2547,
+      "step": 11128
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0552172825089665e-06,
+      "loss": 0.2725,
+      "step": 11129
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0535490866627744e-06,
+      "loss": 0.2592,
+      "step": 11130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0518808908165822e-06,
+      "loss": 0.2555,
+      "step": 11131
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0502126949703897e-06,
+      "loss": 0.2764,
+      "step": 11132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.048544499124197e-06,
+      "loss": 0.2705,
+      "step": 11133
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.046876303278005e-06,
+      "loss": 0.2583,
+      "step": 11134
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0452081074318125e-06,
+      "loss": 0.2709,
+      "step": 11135
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0435399115856204e-06,
+      "loss": 0.2554,
+      "step": 11136
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.041871715739428e-06,
+      "loss": 0.2591,
+      "step": 11137
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0402035198932358e-06,
+      "loss": 0.2533,
+      "step": 11138
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0385353240470432e-06,
+      "loss": 0.2765,
+      "step": 11139
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.036867128200851e-06,
+      "loss": 0.2676,
+      "step": 11140
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0351989323546586e-06,
+      "loss": 0.2781,
+      "step": 11141
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0335307365084665e-06,
+      "loss": 0.2339,
+      "step": 11142
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.031862540662274e-06,
+      "loss": 0.2783,
+      "step": 11143
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0301943448160814e-06,
+      "loss": 0.2666,
+      "step": 11144
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.0285261489698893e-06,
+      "loss": 0.2605,
+      "step": 11145
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0268579531236967e-06,
+      "loss": 0.2961,
+      "step": 11146
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0251897572775046e-06,
+      "loss": 0.2778,
+      "step": 11147
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.023521561431312e-06,
+      "loss": 0.2967,
+      "step": 11148
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.02185336558512e-06,
+      "loss": 0.2687,
+      "step": 11149
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0201851697389274e-06,
+      "loss": 0.2659,
+      "step": 11150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0185169738927353e-06,
+      "loss": 0.2883,
+      "step": 11151
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.016848778046543e-06,
+      "loss": 0.2534,
+      "step": 11152
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0151805822003507e-06,
+      "loss": 0.2851,
+      "step": 11153
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.013512386354158e-06,
+      "loss": 0.2817,
+      "step": 11154
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0118441905079656e-06,
+      "loss": 0.2781,
+      "step": 11155
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0101759946617735e-06,
+      "loss": 0.2563,
+      "step": 11156
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.008507798815581e-06,
+      "loss": 0.2775,
+      "step": 11157
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.006839602969389e-06,
+      "loss": 0.2595,
+      "step": 11158
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0051714071231963e-06,
+      "loss": 0.2629,
+      "step": 11159
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.003503211277004e-06,
+      "loss": 0.2584,
+      "step": 11160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0018350154308117e-06,
+      "loss": 0.2541,
+      "step": 11161
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.0001668195846195e-06,
+      "loss": 0.2977,
+      "step": 11162
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.998498623738427e-06,
+      "loss": 0.2558,
+      "step": 11163
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9968304278922345e-06,
+      "loss": 0.2817,
+      "step": 11164
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9951622320460424e-06,
+      "loss": 0.3068,
+      "step": 11165
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9934940361998502e-06,
+      "loss": 0.2756,
+      "step": 11166
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9918258403536577e-06,
+      "loss": 0.2586,
+      "step": 11167
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.990157644507465e-06,
+      "loss": 0.2593,
+      "step": 11168
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.988489448661273e-06,
+      "loss": 0.2667,
+      "step": 11169
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9868212528150805e-06,
+      "loss": 0.2697,
+      "step": 11170
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9851530569688884e-06,
+      "loss": 0.2569,
+      "step": 11171
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.983484861122696e-06,
+      "loss": 0.2668,
+      "step": 11172
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9818166652765038e-06,
+      "loss": 0.263,
+      "step": 11173
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9801484694303112e-06,
+      "loss": 0.2743,
+      "step": 11174
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9784802735841187e-06,
+      "loss": 0.2819,
+      "step": 11175
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9768120777379266e-06,
+      "loss": 0.2714,
+      "step": 11176
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9751438818917345e-06,
+      "loss": 0.281,
+      "step": 11177
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.973475686045542e-06,
+      "loss": 0.2709,
+      "step": 11178
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9718074901993494e-06,
+      "loss": 0.2474,
+      "step": 11179
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9701392943531573e-06,
+      "loss": 0.2771,
+      "step": 11180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9684710985069647e-06,
+      "loss": 0.2527,
+      "step": 11181
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9668029026607726e-06,
+      "loss": 0.2603,
+      "step": 11182
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.96513470681458e-06,
+      "loss": 0.2571,
+      "step": 11183
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.963466510968388e-06,
+      "loss": 0.271,
+      "step": 11184
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.9617983151221954e-06,
+      "loss": 0.2817,
+      "step": 11185
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.960130119276003e-06,
+      "loss": 0.2706,
+      "step": 11186
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.958461923429811e-06,
+      "loss": 0.264,
+      "step": 11187
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9567937275836187e-06,
+      "loss": 0.2717,
+      "step": 11188
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.955125531737426e-06,
+      "loss": 0.2504,
+      "step": 11189
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9534573358912336e-06,
+      "loss": 0.2597,
+      "step": 11190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9517891400450415e-06,
+      "loss": 0.2989,
+      "step": 11191
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9501209441988494e-06,
+      "loss": 0.2539,
+      "step": 11192
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9484527483526564e-06,
+      "loss": 0.2567,
+      "step": 11193
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9467845525064643e-06,
+      "loss": 0.2618,
+      "step": 11194
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.945116356660272e-06,
+      "loss": 0.2822,
+      "step": 11195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9434481608140797e-06,
+      "loss": 0.2523,
+      "step": 11196
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.941779964967887e-06,
+      "loss": 0.2645,
+      "step": 11197
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.940111769121695e-06,
+      "loss": 0.2678,
+      "step": 11198
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.938443573275503e-06,
+      "loss": 0.2606,
+      "step": 11199
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9367753774293104e-06,
+      "loss": 0.2573,
+      "step": 11200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.935107181583118e-06,
+      "loss": 0.2712,
+      "step": 11201
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9334389857369257e-06,
+      "loss": 0.2878,
+      "step": 11202
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9317707898907336e-06,
+      "loss": 0.2749,
+      "step": 11203
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.930102594044541e-06,
+      "loss": 0.2889,
+      "step": 11204
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9284343981983485e-06,
+      "loss": 0.262,
+      "step": 11205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9267662023521564e-06,
+      "loss": 0.269,
+      "step": 11206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.925098006505964e-06,
+      "loss": 0.2737,
+      "step": 11207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9234298106597713e-06,
+      "loss": 0.2726,
+      "step": 11208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9217616148135792e-06,
+      "loss": 0.2524,
+      "step": 11209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.920093418967387e-06,
+      "loss": 0.2771,
+      "step": 11210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9184252231211946e-06,
+      "loss": 0.2704,
+      "step": 11211
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.916757027275002e-06,
+      "loss": 0.2693,
+      "step": 11212
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.91508883142881e-06,
+      "loss": 0.2581,
+      "step": 11213
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.913420635582618e-06,
+      "loss": 0.2697,
+      "step": 11214
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9117524397364253e-06,
+      "loss": 0.273,
+      "step": 11215
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9100842438902327e-06,
+      "loss": 0.2695,
+      "step": 11216
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9084160480440406e-06,
+      "loss": 0.2684,
+      "step": 11217
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9067478521978483e-06,
+      "loss": 0.2693,
+      "step": 11218
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9050796563516558e-06,
+      "loss": 0.2704,
+      "step": 11219
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9034114605054634e-06,
+      "loss": 0.2518,
+      "step": 11220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9017432646592713e-06,
+      "loss": 0.2595,
+      "step": 11221
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9000750688130786e-06,
+      "loss": 0.2738,
+      "step": 11222
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8984068729668865e-06,
+      "loss": 0.2512,
+      "step": 11223
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8967386771206941e-06,
+      "loss": 0.2628,
+      "step": 11224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8950704812745018e-06,
+      "loss": 0.2743,
+      "step": 11225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8934022854283093e-06,
+      "loss": 0.2697,
+      "step": 11226
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8917340895821172e-06,
+      "loss": 0.2698,
+      "step": 11227
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.8900658937359248e-06,
+      "loss": 0.2647,
+      "step": 11228
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8883976978897325e-06,
+      "loss": 0.2762,
+      "step": 11229
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.88672950204354e-06,
+      "loss": 0.2625,
+      "step": 11230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8850613061973477e-06,
+      "loss": 0.2763,
+      "step": 11231
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8833931103511555e-06,
+      "loss": 0.2603,
+      "step": 11232
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.881724914504963e-06,
+      "loss": 0.2552,
+      "step": 11233
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8800567186587707e-06,
+      "loss": 0.2676,
+      "step": 11234
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8783885228125784e-06,
+      "loss": 0.2702,
+      "step": 11235
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.876720326966386e-06,
+      "loss": 0.2677,
+      "step": 11236
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8750521311201935e-06,
+      "loss": 0.2772,
+      "step": 11237
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8733839352740014e-06,
+      "loss": 0.272,
+      "step": 11238
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.871715739427809e-06,
+      "loss": 0.2679,
+      "step": 11239
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8700475435816165e-06,
+      "loss": 0.2934,
+      "step": 11240
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8683793477354242e-06,
+      "loss": 0.2633,
+      "step": 11241
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8667111518892319e-06,
+      "loss": 0.2566,
+      "step": 11242
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8650429560430398e-06,
+      "loss": 0.2794,
+      "step": 11243
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8633747601968472e-06,
+      "loss": 0.2532,
+      "step": 11244
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.861706564350655e-06,
+      "loss": 0.2782,
+      "step": 11245
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8600383685044626e-06,
+      "loss": 0.2628,
+      "step": 11246
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8583701726582703e-06,
+      "loss": 0.2706,
+      "step": 11247
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8567019768120777e-06,
+      "loss": 0.27,
+      "step": 11248
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8550337809658856e-06,
+      "loss": 0.2833,
+      "step": 11249
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8533655851196933e-06,
+      "loss": 0.2603,
+      "step": 11250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8516973892735007e-06,
+      "loss": 0.279,
+      "step": 11251
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8500291934273084e-06,
+      "loss": 0.277,
+      "step": 11252
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8483609975811163e-06,
+      "loss": 0.2969,
+      "step": 11253
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.846692801734924e-06,
+      "loss": 0.2688,
+      "step": 11254
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8450246058887314e-06,
+      "loss": 0.2455,
+      "step": 11255
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8433564100425391e-06,
+      "loss": 0.2741,
+      "step": 11256
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8416882141963468e-06,
+      "loss": 0.2756,
+      "step": 11257
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8400200183501547e-06,
+      "loss": 0.2573,
+      "step": 11258
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8383518225039621e-06,
+      "loss": 0.2572,
+      "step": 11259
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8366836266577698e-06,
+      "loss": 0.2629,
+      "step": 11260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8350154308115775e-06,
+      "loss": 0.2703,
+      "step": 11261
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.833347234965385e-06,
+      "loss": 0.272,
+      "step": 11262
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8316790391191926e-06,
+      "loss": 0.2708,
+      "step": 11263
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8300108432730005e-06,
+      "loss": 0.2683,
+      "step": 11264
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8283426474268082e-06,
+      "loss": 0.2725,
+      "step": 11265
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8266744515806157e-06,
+      "loss": 0.2782,
+      "step": 11266
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8250062557344233e-06,
+      "loss": 0.276,
+      "step": 11267
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.823338059888231e-06,
+      "loss": 0.2713,
+      "step": 11268
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.8216698640420385e-06,
+      "loss": 0.2511,
+      "step": 11269
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8200016681958464e-06,
+      "loss": 0.2745,
+      "step": 11270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.818333472349654e-06,
+      "loss": 0.2741,
+      "step": 11271
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8166652765034617e-06,
+      "loss": 0.2715,
+      "step": 11272
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8149970806572692e-06,
+      "loss": 0.275,
+      "step": 11273
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8133288848110768e-06,
+      "loss": 0.2673,
+      "step": 11274
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8116606889648847e-06,
+      "loss": 0.2732,
+      "step": 11275
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8099924931186924e-06,
+      "loss": 0.2712,
+      "step": 11276
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8083242972724999e-06,
+      "loss": 0.2781,
+      "step": 11277
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8066561014263076e-06,
+      "loss": 0.2689,
+      "step": 11278
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8049879055801152e-06,
+      "loss": 0.2575,
+      "step": 11279
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8033197097339227e-06,
+      "loss": 0.2471,
+      "step": 11280
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.8016515138877306e-06,
+      "loss": 0.2738,
+      "step": 11281
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7999833180415383e-06,
+      "loss": 0.27,
+      "step": 11282
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.798315122195346e-06,
+      "loss": 0.267,
+      "step": 11283
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7966469263491534e-06,
+      "loss": 0.2735,
+      "step": 11284
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7949787305029613e-06,
+      "loss": 0.2482,
+      "step": 11285
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.793310534656769e-06,
+      "loss": 0.254,
+      "step": 11286
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7916423388105766e-06,
+      "loss": 0.3077,
+      "step": 11287
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.789974142964384e-06,
+      "loss": 0.2464,
+      "step": 11288
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7883059471181918e-06,
+      "loss": 0.2743,
+      "step": 11289
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7866377512719997e-06,
+      "loss": 0.2722,
+      "step": 11290
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7849695554258071e-06,
+      "loss": 0.2527,
+      "step": 11291
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7833013595796148e-06,
+      "loss": 0.2702,
+      "step": 11292
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7816331637334225e-06,
+      "loss": 0.2924,
+      "step": 11293
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7799649678872301e-06,
+      "loss": 0.2581,
+      "step": 11294
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7782967720410376e-06,
+      "loss": 0.2833,
+      "step": 11295
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7766285761948455e-06,
+      "loss": 0.2753,
+      "step": 11296
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7749603803486532e-06,
+      "loss": 0.2747,
+      "step": 11297
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7732921845024606e-06,
+      "loss": 0.2511,
+      "step": 11298
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7716239886562683e-06,
+      "loss": 0.2493,
+      "step": 11299
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.769955792810076e-06,
+      "loss": 0.2744,
+      "step": 11300
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7682875969638839e-06,
+      "loss": 0.2534,
+      "step": 11301
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7666194011176913e-06,
+      "loss": 0.2671,
+      "step": 11302
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.764951205271499e-06,
+      "loss": 0.2918,
+      "step": 11303
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7632830094253067e-06,
+      "loss": 0.2764,
+      "step": 11304
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7616148135791144e-06,
+      "loss": 0.2571,
+      "step": 11305
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7599466177329218e-06,
+      "loss": 0.273,
+      "step": 11306
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7582784218867297e-06,
+      "loss": 0.2818,
+      "step": 11307
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7566102260405374e-06,
+      "loss": 0.2487,
+      "step": 11308
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7549420301943448e-06,
+      "loss": 0.2533,
+      "step": 11309
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7532738343481525e-06,
+      "loss": 0.2806,
+      "step": 11310
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7516056385019602e-06,
+      "loss": 0.2783,
+      "step": 11311
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.749937442655768e-06,
+      "loss": 0.2731,
+      "step": 11312
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7482692468095756e-06,
+      "loss": 0.2752,
+      "step": 11313
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7466010509633832e-06,
+      "loss": 0.2652,
+      "step": 11314
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.744932855117191e-06,
+      "loss": 0.2508,
+      "step": 11315
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7432646592709988e-06,
+      "loss": 0.2626,
+      "step": 11316
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7415964634248063e-06,
+      "loss": 0.2683,
+      "step": 11317
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.739928267578614e-06,
+      "loss": 0.2667,
+      "step": 11318
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7382600717324216e-06,
+      "loss": 0.2447,
+      "step": 11319
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.736591875886229e-06,
+      "loss": 0.2814,
+      "step": 11320
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7349236800400367e-06,
+      "loss": 0.2824,
+      "step": 11321
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7332554841938446e-06,
+      "loss": 0.2827,
+      "step": 11322
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7315872883476523e-06,
+      "loss": 0.2607,
+      "step": 11323
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7299190925014598e-06,
+      "loss": 0.2632,
+      "step": 11324
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7282508966552674e-06,
+      "loss": 0.2527,
+      "step": 11325
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7265827008090751e-06,
+      "loss": 0.2765,
+      "step": 11326
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7249145049628826e-06,
+      "loss": 0.2437,
+      "step": 11327
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7232463091166905e-06,
+      "loss": 0.2556,
+      "step": 11328
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7215781132704981e-06,
+      "loss": 0.2664,
+      "step": 11329
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7199099174243058e-06,
+      "loss": 0.2752,
+      "step": 11330
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7182417215781133e-06,
+      "loss": 0.2581,
+      "step": 11331
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.716573525731921e-06,
+      "loss": 0.2868,
+      "step": 11332
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7149053298857288e-06,
+      "loss": 0.2727,
+      "step": 11333
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7132371340395365e-06,
+      "loss": 0.2487,
+      "step": 11334
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.711568938193344e-06,
+      "loss": 0.2638,
+      "step": 11335
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7099007423471517e-06,
+      "loss": 0.2663,
+      "step": 11336
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7082325465009593e-06,
+      "loss": 0.2875,
+      "step": 11337
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7065643506547668e-06,
+      "loss": 0.2496,
+      "step": 11338
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7048961548085747e-06,
+      "loss": 0.2705,
+      "step": 11339
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7032279589623824e-06,
+      "loss": 0.2727,
+      "step": 11340
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.70155976311619e-06,
+      "loss": 0.2452,
+      "step": 11341
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6998915672699975e-06,
+      "loss": 0.2735,
+      "step": 11342
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6982233714238052e-06,
+      "loss": 0.2773,
+      "step": 11343
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.696555175577613e-06,
+      "loss": 0.2921,
+      "step": 11344
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6948869797314207e-06,
+      "loss": 0.2565,
+      "step": 11345
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6932187838852282e-06,
+      "loss": 0.2632,
+      "step": 11346
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6915505880390359e-06,
+      "loss": 0.2841,
+      "step": 11347
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6898823921928438e-06,
+      "loss": 0.2609,
+      "step": 11348
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6882141963466512e-06,
+      "loss": 0.2714,
+      "step": 11349
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.686546000500459e-06,
+      "loss": 0.2607,
+      "step": 11350
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6848778046542666e-06,
+      "loss": 0.2517,
+      "step": 11351
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6832096088080743e-06,
+      "loss": 0.2525,
+      "step": 11352
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6815414129618817e-06,
+      "loss": 0.2656,
+      "step": 11353
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6798732171156896e-06,
+      "loss": 0.2719,
+      "step": 11354
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6782050212694973e-06,
+      "loss": 0.2745,
+      "step": 11355
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6765368254233047e-06,
+      "loss": 0.2684,
+      "step": 11356
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6748686295771124e-06,
+      "loss": 0.2547,
+      "step": 11357
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.67320043373092e-06,
+      "loss": 0.2619,
+      "step": 11358
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.671532237884728e-06,
+      "loss": 0.2522,
+      "step": 11359
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6698640420385354e-06,
+      "loss": 0.264,
+      "step": 11360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6681958461923431e-06,
+      "loss": 0.28,
+      "step": 11361
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6665276503461508e-06,
+      "loss": 0.2845,
+      "step": 11362
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6648594544999585e-06,
+      "loss": 0.2738,
+      "step": 11363
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.663191258653766e-06,
+      "loss": 0.2891,
+      "step": 11364
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6615230628075738e-06,
+      "loss": 0.2639,
+      "step": 11365
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6598548669613815e-06,
+      "loss": 0.2718,
+      "step": 11366
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.658186671115189e-06,
+      "loss": 0.2556,
+      "step": 11367
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6565184752689966e-06,
+      "loss": 0.2681,
+      "step": 11368
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6548502794228043e-06,
+      "loss": 0.2651,
+      "step": 11369
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6531820835766122e-06,
+      "loss": 0.2532,
+      "step": 11370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6515138877304197e-06,
+      "loss": 0.2685,
+      "step": 11371
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6498456918842273e-06,
+      "loss": 0.2743,
+      "step": 11372
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.648177496038035e-06,
+      "loss": 0.2686,
+      "step": 11373
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6465093001918425e-06,
+      "loss": 0.2734,
+      "step": 11374
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6448411043456501e-06,
+      "loss": 0.2732,
+      "step": 11375
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.643172908499458e-06,
+      "loss": 0.2968,
+      "step": 11376
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6415047126532657e-06,
+      "loss": 0.26,
+      "step": 11377
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6398365168070732e-06,
+      "loss": 0.2806,
+      "step": 11378
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6381683209608809e-06,
+      "loss": 0.2474,
+      "step": 11379
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6365001251146887e-06,
+      "loss": 0.2796,
+      "step": 11380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6348319292684964e-06,
+      "loss": 0.262,
+      "step": 11381
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6331637334223039e-06,
+      "loss": 0.2805,
+      "step": 11382
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6314955375761116e-06,
+      "loss": 0.2767,
+      "step": 11383
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6298273417299192e-06,
+      "loss": 0.2836,
+      "step": 11384
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6281591458837267e-06,
+      "loss": 0.2665,
+      "step": 11385
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6264909500375346e-06,
+      "loss": 0.2543,
+      "step": 11386
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6248227541913423e-06,
+      "loss": 0.2699,
+      "step": 11387
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.62315455834515e-06,
+      "loss": 0.271,
+      "step": 11388
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6214863624989574e-06,
+      "loss": 0.2459,
+      "step": 11389
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.619818166652765e-06,
+      "loss": 0.2555,
+      "step": 11390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.618149970806573e-06,
+      "loss": 0.2774,
+      "step": 11391
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6164817749603806e-06,
+      "loss": 0.2635,
+      "step": 11392
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.614813579114188e-06,
+      "loss": 0.2737,
+      "step": 11393
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6131453832679958e-06,
+      "loss": 0.2683,
+      "step": 11394
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6114771874218034e-06,
+      "loss": 0.2653,
+      "step": 11395
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.609808991575611e-06,
+      "loss": 0.2754,
+      "step": 11396
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6081407957294188e-06,
+      "loss": 0.2746,
+      "step": 11397
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6064725998832265e-06,
+      "loss": 0.2687,
+      "step": 11398
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6048044040370341e-06,
+      "loss": 0.2772,
+      "step": 11399
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6031362081908416e-06,
+      "loss": 0.265,
+      "step": 11400
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6014680123446493e-06,
+      "loss": 0.2817,
+      "step": 11401
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5997998164984572e-06,
+      "loss": 0.2727,
+      "step": 11402
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5981316206522646e-06,
+      "loss": 0.3012,
+      "step": 11403
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5964634248060723e-06,
+      "loss": 0.2614,
+      "step": 11404
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.59479522895988e-06,
+      "loss": 0.2837,
+      "step": 11405
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5931270331136879e-06,
+      "loss": 0.2913,
+      "step": 11406
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5914588372674951e-06,
+      "loss": 0.2686,
+      "step": 11407
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.589790641421303e-06,
+      "loss": 0.2847,
+      "step": 11408
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5881224455751107e-06,
+      "loss": 0.2926,
+      "step": 11409
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5864542497289184e-06,
+      "loss": 0.2797,
+      "step": 11410
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5847860538827258e-06,
+      "loss": 0.2836,
+      "step": 11411
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5831178580365337e-06,
+      "loss": 0.2606,
+      "step": 11412
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5814496621903414e-06,
+      "loss": 0.2534,
+      "step": 11413
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5797814663441489e-06,
+      "loss": 0.2528,
+      "step": 11414
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5781132704979565e-06,
+      "loss": 0.2799,
+      "step": 11415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5764450746517642e-06,
+      "loss": 0.2599,
+      "step": 11416
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.574776878805572e-06,
+      "loss": 0.2627,
+      "step": 11417
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5731086829593796e-06,
+      "loss": 0.2696,
+      "step": 11418
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5714404871131872e-06,
+      "loss": 0.2702,
+      "step": 11419
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.569772291266995e-06,
+      "loss": 0.2665,
+      "step": 11420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5681040954208026e-06,
+      "loss": 0.2752,
+      "step": 11421
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.56643589957461e-06,
+      "loss": 0.2607,
+      "step": 11422
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.564767703728418e-06,
+      "loss": 0.2821,
+      "step": 11423
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5630995078822256e-06,
+      "loss": 0.2562,
+      "step": 11424
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.561431312036033e-06,
+      "loss": 0.2622,
+      "step": 11425
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5597631161898407e-06,
+      "loss": 0.2685,
+      "step": 11426
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5580949203436484e-06,
+      "loss": 0.2959,
+      "step": 11427
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5564267244974563e-06,
+      "loss": 0.2617,
+      "step": 11428
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5547585286512638e-06,
+      "loss": 0.2782,
+      "step": 11429
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5530903328050714e-06,
+      "loss": 0.2739,
+      "step": 11430
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5514221369588791e-06,
+      "loss": 0.2681,
+      "step": 11431
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5497539411126866e-06,
+      "loss": 0.2375,
+      "step": 11432
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5480857452664943e-06,
+      "loss": 0.2725,
+      "step": 11433
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.5464175494203021e-06,
+      "loss": 0.2753,
+      "step": 11434
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5447493535741098e-06,
+      "loss": 0.2723,
+      "step": 11435
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5430811577279173e-06,
+      "loss": 0.2702,
+      "step": 11436
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.541412961881725e-06,
+      "loss": 0.2484,
+      "step": 11437
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5397447660355328e-06,
+      "loss": 0.2624,
+      "step": 11438
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5380765701893405e-06,
+      "loss": 0.2707,
+      "step": 11439
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.536408374343148e-06,
+      "loss": 0.2728,
+      "step": 11440
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5347401784969557e-06,
+      "loss": 0.2741,
+      "step": 11441
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5330719826507633e-06,
+      "loss": 0.2472,
+      "step": 11442
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5314037868045708e-06,
+      "loss": 0.2538,
+      "step": 11443
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5297355909583787e-06,
+      "loss": 0.2707,
+      "step": 11444
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5280673951121864e-06,
+      "loss": 0.2712,
+      "step": 11445
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.526399199265994e-06,
+      "loss": 0.2525,
+      "step": 11446
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5247310034198015e-06,
+      "loss": 0.2801,
+      "step": 11447
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5230628075736092e-06,
+      "loss": 0.2664,
+      "step": 11448
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.521394611727417e-06,
+      "loss": 0.2698,
+      "step": 11449
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5197264158812247e-06,
+      "loss": 0.2645,
+      "step": 11450
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5180582200350322e-06,
+      "loss": 0.2579,
+      "step": 11451
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5163900241888399e-06,
+      "loss": 0.2675,
+      "step": 11452
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5147218283426476e-06,
+      "loss": 0.2719,
+      "step": 11453
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.513053632496455e-06,
+      "loss": 0.2637,
+      "step": 11454
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.511385436650263e-06,
+      "loss": 0.2732,
+      "step": 11455
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5097172408040706e-06,
+      "loss": 0.2777,
+      "step": 11456
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5080490449578783e-06,
+      "loss": 0.2584,
+      "step": 11457
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5063808491116857e-06,
+      "loss": 0.2904,
+      "step": 11458
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5047126532654934e-06,
+      "loss": 0.277,
+      "step": 11459
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5030444574193013e-06,
+      "loss": 0.2776,
+      "step": 11460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.5013762615731087e-06,
+      "loss": 0.2743,
+      "step": 11461
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4997080657269164e-06,
+      "loss": 0.2942,
+      "step": 11462
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.498039869880724e-06,
+      "loss": 0.2781,
+      "step": 11463
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4963716740345318e-06,
+      "loss": 0.2814,
+      "step": 11464
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4947034781883392e-06,
+      "loss": 0.2589,
+      "step": 11465
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4930352823421471e-06,
+      "loss": 0.2617,
+      "step": 11466
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4913670864959548e-06,
+      "loss": 0.2623,
+      "step": 11467
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4896988906497625e-06,
+      "loss": 0.2742,
+      "step": 11468
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.48803069480357e-06,
+      "loss": 0.2501,
+      "step": 11469
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4863624989573778e-06,
+      "loss": 0.2856,
+      "step": 11470
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4846943031111855e-06,
+      "loss": 0.2804,
+      "step": 11471
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.483026107264993e-06,
+      "loss": 0.2712,
+      "step": 11472
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4813579114188006e-06,
+      "loss": 0.2754,
+      "step": 11473
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4796897155726083e-06,
+      "loss": 0.2798,
+      "step": 11474
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.4780215197264162e-06,
+      "loss": 0.2784,
+      "step": 11475
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4763533238802237e-06,
+      "loss": 0.2701,
+      "step": 11476
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4746851280340313e-06,
+      "loss": 0.2744,
+      "step": 11477
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.473016932187839e-06,
+      "loss": 0.2709,
+      "step": 11478
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4713487363416467e-06,
+      "loss": 0.2649,
+      "step": 11479
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4696805404954542e-06,
+      "loss": 0.2803,
+      "step": 11480
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.468012344649262e-06,
+      "loss": 0.2694,
+      "step": 11481
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4663441488030697e-06,
+      "loss": 0.263,
+      "step": 11482
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4646759529568772e-06,
+      "loss": 0.2484,
+      "step": 11483
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4630077571106849e-06,
+      "loss": 0.2574,
+      "step": 11484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4613395612644925e-06,
+      "loss": 0.2569,
+      "step": 11485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4596713654183004e-06,
+      "loss": 0.2641,
+      "step": 11486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4580031695721079e-06,
+      "loss": 0.26,
+      "step": 11487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4563349737259156e-06,
+      "loss": 0.2739,
+      "step": 11488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4546667778797232e-06,
+      "loss": 0.2658,
+      "step": 11489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4529985820335307e-06,
+      "loss": 0.2636,
+      "step": 11490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4513303861873384e-06,
+      "loss": 0.2856,
+      "step": 11491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4496621903411463e-06,
+      "loss": 0.2752,
+      "step": 11492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.447993994494954e-06,
+      "loss": 0.2619,
+      "step": 11493
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4463257986487614e-06,
+      "loss": 0.2705,
+      "step": 11494
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.444657602802569e-06,
+      "loss": 0.2628,
+      "step": 11495
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4429894069563767e-06,
+      "loss": 0.2611,
+      "step": 11496
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4413212111101846e-06,
+      "loss": 0.2762,
+      "step": 11497
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.439653015263992e-06,
+      "loss": 0.2706,
+      "step": 11498
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4379848194177998e-06,
+      "loss": 0.2549,
+      "step": 11499
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4363166235716074e-06,
+      "loss": 0.2653,
+      "step": 11500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.434648427725415e-06,
+      "loss": 0.2666,
+      "step": 11501
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4329802318792228e-06,
+      "loss": 0.2331,
+      "step": 11502
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4313120360330305e-06,
+      "loss": 0.2775,
+      "step": 11503
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4296438401868381e-06,
+      "loss": 0.2725,
+      "step": 11504
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4279756443406456e-06,
+      "loss": 0.2805,
+      "step": 11505
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4263074484944533e-06,
+      "loss": 0.2583,
+      "step": 11506
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4246392526482612e-06,
+      "loss": 0.2699,
+      "step": 11507
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4229710568020688e-06,
+      "loss": 0.2695,
+      "step": 11508
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4213028609558763e-06,
+      "loss": 0.278,
+      "step": 11509
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.419634665109684e-06,
+      "loss": 0.2565,
+      "step": 11510
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4179664692634917e-06,
+      "loss": 0.2623,
+      "step": 11511
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4162982734172991e-06,
+      "loss": 0.2662,
+      "step": 11512
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.414630077571107e-06,
+      "loss": 0.2709,
+      "step": 11513
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4129618817249147e-06,
+      "loss": 0.2633,
+      "step": 11514
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4112936858787224e-06,
+      "loss": 0.2698,
+      "step": 11515
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.4096254900325298e-06,
+      "loss": 0.2663,
+      "step": 11516
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4079572941863375e-06,
+      "loss": 0.2708,
+      "step": 11517
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4062890983401454e-06,
+      "loss": 0.2458,
+      "step": 11518
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4046209024939529e-06,
+      "loss": 0.2642,
+      "step": 11519
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4029527066477605e-06,
+      "loss": 0.2647,
+      "step": 11520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.4012845108015682e-06,
+      "loss": 0.2604,
+      "step": 11521
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3996163149553759e-06,
+      "loss": 0.2771,
+      "step": 11522
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3979481191091833e-06,
+      "loss": 0.2704,
+      "step": 11523
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3962799232629912e-06,
+      "loss": 0.2521,
+      "step": 11524
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.394611727416799e-06,
+      "loss": 0.265,
+      "step": 11525
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3929435315706066e-06,
+      "loss": 0.2685,
+      "step": 11526
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.391275335724414e-06,
+      "loss": 0.2676,
+      "step": 11527
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3896071398782217e-06,
+      "loss": 0.274,
+      "step": 11528
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3879389440320296e-06,
+      "loss": 0.2561,
+      "step": 11529
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.386270748185837e-06,
+      "loss": 0.2855,
+      "step": 11530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3846025523396447e-06,
+      "loss": 0.2427,
+      "step": 11531
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3829343564934524e-06,
+      "loss": 0.2751,
+      "step": 11532
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3812661606472603e-06,
+      "loss": 0.282,
+      "step": 11533
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3795979648010678e-06,
+      "loss": 0.2506,
+      "step": 11534
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3779297689548754e-06,
+      "loss": 0.265,
+      "step": 11535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3762615731086831e-06,
+      "loss": 0.2609,
+      "step": 11536
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3745933772624906e-06,
+      "loss": 0.2684,
+      "step": 11537
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3729251814162983e-06,
+      "loss": 0.2799,
+      "step": 11538
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3712569855701061e-06,
+      "loss": 0.2649,
+      "step": 11539
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3695887897239138e-06,
+      "loss": 0.2624,
+      "step": 11540
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3679205938777213e-06,
+      "loss": 0.2626,
+      "step": 11541
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.366252398031529e-06,
+      "loss": 0.2589,
+      "step": 11542
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3645842021853366e-06,
+      "loss": 0.2675,
+      "step": 11543
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3629160063391445e-06,
+      "loss": 0.2756,
+      "step": 11544
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.361247810492952e-06,
+      "loss": 0.2646,
+      "step": 11545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3595796146467597e-06,
+      "loss": 0.2554,
+      "step": 11546
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3579114188005673e-06,
+      "loss": 0.2605,
+      "step": 11547
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3562432229543748e-06,
+      "loss": 0.2611,
+      "step": 11548
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3545750271081825e-06,
+      "loss": 0.2636,
+      "step": 11549
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3529068312619904e-06,
+      "loss": 0.26,
+      "step": 11550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.351238635415798e-06,
+      "loss": 0.2551,
+      "step": 11551
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3495704395696055e-06,
+      "loss": 0.2758,
+      "step": 11552
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3479022437234132e-06,
+      "loss": 0.2615,
+      "step": 11553
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3462340478772209e-06,
+      "loss": 0.2676,
+      "step": 11554
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3445658520310287e-06,
+      "loss": 0.2649,
+      "step": 11555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3428976561848362e-06,
+      "loss": 0.2532,
+      "step": 11556
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3412294603386439e-06,
+      "loss": 0.2759,
+      "step": 11557
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.3395612644924516e-06,
+      "loss": 0.2735,
+      "step": 11558
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.337893068646259e-06,
+      "loss": 0.2696,
+      "step": 11559
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3362248728000667e-06,
+      "loss": 0.2542,
+      "step": 11560
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3345566769538746e-06,
+      "loss": 0.2522,
+      "step": 11561
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3328884811076823e-06,
+      "loss": 0.2541,
+      "step": 11562
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3312202852614897e-06,
+      "loss": 0.264,
+      "step": 11563
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3295520894152974e-06,
+      "loss": 0.2748,
+      "step": 11564
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3278838935691053e-06,
+      "loss": 0.2505,
+      "step": 11565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3262156977229127e-06,
+      "loss": 0.2719,
+      "step": 11566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3245475018767204e-06,
+      "loss": 0.2723,
+      "step": 11567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.322879306030528e-06,
+      "loss": 0.2873,
+      "step": 11568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3212111101843358e-06,
+      "loss": 0.2766,
+      "step": 11569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3195429143381432e-06,
+      "loss": 0.2494,
+      "step": 11570
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3178747184919511e-06,
+      "loss": 0.2643,
+      "step": 11571
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3162065226457588e-06,
+      "loss": 0.2359,
+      "step": 11572
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3145383267995665e-06,
+      "loss": 0.265,
+      "step": 11573
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.312870130953374e-06,
+      "loss": 0.2665,
+      "step": 11574
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3112019351071816e-06,
+      "loss": 0.2844,
+      "step": 11575
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3095337392609895e-06,
+      "loss": 0.2576,
+      "step": 11576
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.307865543414797e-06,
+      "loss": 0.2759,
+      "step": 11577
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3061973475686046e-06,
+      "loss": 0.2568,
+      "step": 11578
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3045291517224123e-06,
+      "loss": 0.2818,
+      "step": 11579
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.30286095587622e-06,
+      "loss": 0.2586,
+      "step": 11580
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.3011927600300275e-06,
+      "loss": 0.2774,
+      "step": 11581
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2995245641838353e-06,
+      "loss": 0.2649,
+      "step": 11582
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.297856368337643e-06,
+      "loss": 0.2439,
+      "step": 11583
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2961881724914507e-06,
+      "loss": 0.2627,
+      "step": 11584
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2945199766452582e-06,
+      "loss": 0.2808,
+      "step": 11585
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2928517807990658e-06,
+      "loss": 0.2723,
+      "step": 11586
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2911835849528737e-06,
+      "loss": 0.275,
+      "step": 11587
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2895153891066812e-06,
+      "loss": 0.2426,
+      "step": 11588
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2878471932604889e-06,
+      "loss": 0.2941,
+      "step": 11589
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2861789974142965e-06,
+      "loss": 0.26,
+      "step": 11590
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2845108015681044e-06,
+      "loss": 0.2601,
+      "step": 11591
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2828426057219117e-06,
+      "loss": 0.2682,
+      "step": 11592
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2811744098757196e-06,
+      "loss": 0.2736,
+      "step": 11593
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2795062140295272e-06,
+      "loss": 0.2655,
+      "step": 11594
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2778380181833347e-06,
+      "loss": 0.2607,
+      "step": 11595
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2761698223371424e-06,
+      "loss": 0.2665,
+      "step": 11596
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2745016264909503e-06,
+      "loss": 0.2544,
+      "step": 11597
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.272833430644758e-06,
+      "loss": 0.2778,
+      "step": 11598
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.2711652347985654e-06,
+      "loss": 0.2638,
+      "step": 11599
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.269497038952373e-06,
+      "loss": 0.2692,
+      "step": 11600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2678288431061807e-06,
+      "loss": 0.2614,
+      "step": 11601
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2661606472599886e-06,
+      "loss": 0.2543,
+      "step": 11602
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.264492451413796e-06,
+      "loss": 0.2472,
+      "step": 11603
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2628242555676038e-06,
+      "loss": 0.2519,
+      "step": 11604
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2611560597214114e-06,
+      "loss": 0.2707,
+      "step": 11605
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.259487863875219e-06,
+      "loss": 0.2444,
+      "step": 11606
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2578196680290266e-06,
+      "loss": 0.2702,
+      "step": 11607
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2561514721828345e-06,
+      "loss": 0.293,
+      "step": 11608
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2544832763366421e-06,
+      "loss": 0.2779,
+      "step": 11609
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2528150804904496e-06,
+      "loss": 0.2517,
+      "step": 11610
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2511468846442573e-06,
+      "loss": 0.2603,
+      "step": 11611
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.249478688798065e-06,
+      "loss": 0.2603,
+      "step": 11612
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2478104929518726e-06,
+      "loss": 0.283,
+      "step": 11613
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2461422971056803e-06,
+      "loss": 0.286,
+      "step": 11614
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.244474101259488e-06,
+      "loss": 0.2686,
+      "step": 11615
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2428059054132957e-06,
+      "loss": 0.2644,
+      "step": 11616
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2411377095671033e-06,
+      "loss": 0.2761,
+      "step": 11617
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2394695137209108e-06,
+      "loss": 0.2559,
+      "step": 11618
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2378013178747187e-06,
+      "loss": 0.2893,
+      "step": 11619
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2361331220285262e-06,
+      "loss": 0.2505,
+      "step": 11620
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.234464926182334e-06,
+      "loss": 0.2755,
+      "step": 11621
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2327967303361415e-06,
+      "loss": 0.2601,
+      "step": 11622
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2311285344899494e-06,
+      "loss": 0.2806,
+      "step": 11623
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2294603386437569e-06,
+      "loss": 0.267,
+      "step": 11624
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2277921427975645e-06,
+      "loss": 0.2679,
+      "step": 11625
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2261239469513722e-06,
+      "loss": 0.2582,
+      "step": 11626
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2244557511051799e-06,
+      "loss": 0.2555,
+      "step": 11627
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2227875552589876e-06,
+      "loss": 0.289,
+      "step": 11628
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2211193594127952e-06,
+      "loss": 0.293,
+      "step": 11629
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.219451163566603e-06,
+      "loss": 0.2539,
+      "step": 11630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2177829677204104e-06,
+      "loss": 0.2525,
+      "step": 11631
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2161147718742183e-06,
+      "loss": 0.2609,
+      "step": 11632
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2144465760280257e-06,
+      "loss": 0.2435,
+      "step": 11633
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2127783801818334e-06,
+      "loss": 0.2772,
+      "step": 11634
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.211110184335641e-06,
+      "loss": 0.2585,
+      "step": 11635
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2094419884894487e-06,
+      "loss": 0.2588,
+      "step": 11636
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2077737926432564e-06,
+      "loss": 0.2737,
+      "step": 11637
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.206105596797064e-06,
+      "loss": 0.2652,
+      "step": 11638
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2044374009508718e-06,
+      "loss": 0.2577,
+      "step": 11639
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.2027692051046794e-06,
+      "loss": 0.258,
+      "step": 11640
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.2011010092584871e-06,
+      "loss": 0.2885,
+      "step": 11641
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1994328134122948e-06,
+      "loss": 0.2516,
+      "step": 11642
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1977646175661023e-06,
+      "loss": 0.2667,
+      "step": 11643
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.19609642171991e-06,
+      "loss": 0.2782,
+      "step": 11644
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1944282258737176e-06,
+      "loss": 0.2558,
+      "step": 11645
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1927600300275253e-06,
+      "loss": 0.2882,
+      "step": 11646
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.191091834181333e-06,
+      "loss": 0.255,
+      "step": 11647
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1894236383351406e-06,
+      "loss": 0.2796,
+      "step": 11648
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1877554424889483e-06,
+      "loss": 0.2784,
+      "step": 11649
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.186087246642756e-06,
+      "loss": 0.292,
+      "step": 11650
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1844190507965637e-06,
+      "loss": 0.2624,
+      "step": 11651
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1827508549503711e-06,
+      "loss": 0.2855,
+      "step": 11652
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.181082659104179e-06,
+      "loss": 0.2456,
+      "step": 11653
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1794144632579865e-06,
+      "loss": 0.2587,
+      "step": 11654
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1777462674117944e-06,
+      "loss": 0.2587,
+      "step": 11655
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1760780715656018e-06,
+      "loss": 0.2834,
+      "step": 11656
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1744098757194095e-06,
+      "loss": 0.255,
+      "step": 11657
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1727416798732172e-06,
+      "loss": 0.2636,
+      "step": 11658
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1710734840270249e-06,
+      "loss": 0.2713,
+      "step": 11659
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1694052881808325e-06,
+      "loss": 0.2577,
+      "step": 11660
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1677370923346402e-06,
+      "loss": 0.2749,
+      "step": 11661
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1660688964884479e-06,
+      "loss": 0.2851,
+      "step": 11662
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1644007006422553e-06,
+      "loss": 0.2806,
+      "step": 11663
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1627325047960632e-06,
+      "loss": 0.2501,
+      "step": 11664
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1610643089498707e-06,
+      "loss": 0.2642,
+      "step": 11665
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1593961131036786e-06,
+      "loss": 0.266,
+      "step": 11666
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.157727917257486e-06,
+      "loss": 0.2565,
+      "step": 11667
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.156059721411294e-06,
+      "loss": 0.2587,
+      "step": 11668
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1543915255651014e-06,
+      "loss": 0.2653,
+      "step": 11669
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.152723329718909e-06,
+      "loss": 0.2688,
+      "step": 11670
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1510551338727167e-06,
+      "loss": 0.2932,
+      "step": 11671
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1493869380265244e-06,
+      "loss": 0.2737,
+      "step": 11672
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.147718742180332e-06,
+      "loss": 0.2731,
+      "step": 11673
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1460505463341398e-06,
+      "loss": 0.2615,
+      "step": 11674
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1443823504879474e-06,
+      "loss": 0.255,
+      "step": 11675
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.142714154641755e-06,
+      "loss": 0.2762,
+      "step": 11676
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1410459587955628e-06,
+      "loss": 0.26,
+      "step": 11677
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1393777629493703e-06,
+      "loss": 0.2903,
+      "step": 11678
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1377095671031781e-06,
+      "loss": 0.2611,
+      "step": 11679
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1360413712569856e-06,
+      "loss": 0.2619,
+      "step": 11680
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.1343731754107933e-06,
+      "loss": 0.2659,
+      "step": 11681
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.132704979564601e-06,
+      "loss": 0.2705,
+      "step": 11682
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1310367837184086e-06,
+      "loss": 0.2581,
+      "step": 11683
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1293685878722163e-06,
+      "loss": 0.2509,
+      "step": 11684
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.127700392026024e-06,
+      "loss": 0.2918,
+      "step": 11685
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1260321961798317e-06,
+      "loss": 0.2484,
+      "step": 11686
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1243640003336393e-06,
+      "loss": 0.2673,
+      "step": 11687
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.122695804487447e-06,
+      "loss": 0.252,
+      "step": 11688
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1210276086412545e-06,
+      "loss": 0.2461,
+      "step": 11689
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1193594127950624e-06,
+      "loss": 0.2542,
+      "step": 11690
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1176912169488698e-06,
+      "loss": 0.2477,
+      "step": 11691
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1160230211026775e-06,
+      "loss": 0.2599,
+      "step": 11692
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1143548252564852e-06,
+      "loss": 0.2608,
+      "step": 11693
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1126866294102929e-06,
+      "loss": 0.2948,
+      "step": 11694
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1110184335641005e-06,
+      "loss": 0.2502,
+      "step": 11695
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1093502377179082e-06,
+      "loss": 0.2837,
+      "step": 11696
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1076820418717159e-06,
+      "loss": 0.2795,
+      "step": 11697
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1060138460255236e-06,
+      "loss": 0.2675,
+      "step": 11698
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1043456501793312e-06,
+      "loss": 0.2674,
+      "step": 11699
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.102677454333139e-06,
+      "loss": 0.2661,
+      "step": 11700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.1010092584869464e-06,
+      "loss": 0.2691,
+      "step": 11701
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.099341062640754e-06,
+      "loss": 0.2481,
+      "step": 11702
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0976728667945617e-06,
+      "loss": 0.2433,
+      "step": 11703
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0960046709483694e-06,
+      "loss": 0.2867,
+      "step": 11704
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.094336475102177e-06,
+      "loss": 0.2579,
+      "step": 11705
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0926682792559847e-06,
+      "loss": 0.2617,
+      "step": 11706
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0910000834097924e-06,
+      "loss": 0.2707,
+      "step": 11707
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0893318875636e-06,
+      "loss": 0.2655,
+      "step": 11708
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0876636917174078e-06,
+      "loss": 0.2647,
+      "step": 11709
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0859954958712152e-06,
+      "loss": 0.2558,
+      "step": 11710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0843273000250231e-06,
+      "loss": 0.262,
+      "step": 11711
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0826591041788306e-06,
+      "loss": 0.2694,
+      "step": 11712
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0809909083326383e-06,
+      "loss": 0.2603,
+      "step": 11713
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.079322712486446e-06,
+      "loss": 0.2615,
+      "step": 11714
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0776545166402536e-06,
+      "loss": 0.2496,
+      "step": 11715
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0759863207940613e-06,
+      "loss": 0.2319,
+      "step": 11716
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.074318124947869e-06,
+      "loss": 0.2657,
+      "step": 11717
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0726499291016766e-06,
+      "loss": 0.2718,
+      "step": 11718
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0709817332554843e-06,
+      "loss": 0.2803,
+      "step": 11719
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.069313537409292e-06,
+      "loss": 0.258,
+      "step": 11720
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0676453415630995e-06,
+      "loss": 0.2758,
+      "step": 11721
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.0659771457169073e-06,
+      "loss": 0.2556,
+      "step": 11722
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0643089498707148e-06,
+      "loss": 0.2844,
+      "step": 11723
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0626407540245227e-06,
+      "loss": 0.2415,
+      "step": 11724
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0609725581783302e-06,
+      "loss": 0.2557,
+      "step": 11725
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0593043623321378e-06,
+      "loss": 0.2695,
+      "step": 11726
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0576361664859455e-06,
+      "loss": 0.2778,
+      "step": 11727
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0559679706397532e-06,
+      "loss": 0.2776,
+      "step": 11728
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0542997747935609e-06,
+      "loss": 0.2683,
+      "step": 11729
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 0.2541,
+      "step": 11730
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0509633831011762e-06,
+      "loss": 0.2476,
+      "step": 11731
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0492951872549839e-06,
+      "loss": 0.2556,
+      "step": 11732
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0476269914087916e-06,
+      "loss": 0.2429,
+      "step": 11733
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.045958795562599e-06,
+      "loss": 0.2734,
+      "step": 11734
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.044290599716407e-06,
+      "loss": 0.2802,
+      "step": 11735
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0426224038702144e-06,
+      "loss": 0.2529,
+      "step": 11736
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0409542080240223e-06,
+      "loss": 0.2678,
+      "step": 11737
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0392860121778297e-06,
+      "loss": 0.2712,
+      "step": 11738
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0376178163316374e-06,
+      "loss": 0.276,
+      "step": 11739
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.035949620485445e-06,
+      "loss": 0.2588,
+      "step": 11740
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0342814246392527e-06,
+      "loss": 0.2873,
+      "step": 11741
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0326132287930604e-06,
+      "loss": 0.2692,
+      "step": 11742
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.030945032946868e-06,
+      "loss": 0.2649,
+      "step": 11743
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0292768371006758e-06,
+      "loss": 0.2875,
+      "step": 11744
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0276086412544832e-06,
+      "loss": 0.2724,
+      "step": 11745
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0259404454082911e-06,
+      "loss": 0.2641,
+      "step": 11746
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0242722495620986e-06,
+      "loss": 0.2806,
+      "step": 11747
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0226040537159063e-06,
+      "loss": 0.2599,
+      "step": 11748
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.020935857869714e-06,
+      "loss": 0.2651,
+      "step": 11749
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0192676620235216e-06,
+      "loss": 0.2718,
+      "step": 11750
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0175994661773293e-06,
+      "loss": 0.2725,
+      "step": 11751
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.015931270331137e-06,
+      "loss": 0.2582,
+      "step": 11752
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0142630744849446e-06,
+      "loss": 0.2633,
+      "step": 11753
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0125948786387523e-06,
+      "loss": 0.2693,
+      "step": 11754
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.01092668279256e-06,
+      "loss": 0.2769,
+      "step": 11755
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0092584869463677e-06,
+      "loss": 0.2788,
+      "step": 11756
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0075902911001753e-06,
+      "loss": 0.2824,
+      "step": 11757
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0059220952539828e-06,
+      "loss": 0.2673,
+      "step": 11758
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0042538994077905e-06,
+      "loss": 0.2781,
+      "step": 11759
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0025857035615982e-06,
+      "loss": 0.2647,
+      "step": 11760
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.0009175077154058e-06,
+      "loss": 0.2673,
+      "step": 11761
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.992493118692135e-07,
+      "loss": 0.2682,
+      "step": 11762
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.975811160230212e-07,
+      "loss": 0.2779,
+      "step": 11763
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.959129201768289e-07,
+      "loss": 0.2589,
+      "step": 11764
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.942447243306365e-07,
+      "loss": 0.2508,
+      "step": 11765
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.925765284844442e-07,
+      "loss": 0.2649,
+      "step": 11766
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.909083326382519e-07,
+      "loss": 0.2694,
+      "step": 11767
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.892401367920593e-07,
+      "loss": 0.2518,
+      "step": 11768
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.875719409458672e-07,
+      "loss": 0.2661,
+      "step": 11769
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.859037450996747e-07,
+      "loss": 0.2784,
+      "step": 11770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.842355492534824e-07,
+      "loss": 0.2582,
+      "step": 11771
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.8256735340729e-07,
+      "loss": 0.263,
+      "step": 11772
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.808991575610977e-07,
+      "loss": 0.2651,
+      "step": 11773
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.792309617149054e-07,
+      "loss": 0.2663,
+      "step": 11774
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.77562765868713e-07,
+      "loss": 0.2556,
+      "step": 11775
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.758945700225207e-07,
+      "loss": 0.2736,
+      "step": 11776
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.742263741763282e-07,
+      "loss": 0.2588,
+      "step": 11777
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.72558178330136e-07,
+      "loss": 0.2613,
+      "step": 11778
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.708899824839436e-07,
+      "loss": 0.2649,
+      "step": 11779
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.692217866377514e-07,
+      "loss": 0.2385,
+      "step": 11780
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.67553590791559e-07,
+      "loss": 0.2523,
+      "step": 11781
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.658853949453668e-07,
+      "loss": 0.2453,
+      "step": 11782
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.642171990991743e-07,
+      "loss": 0.2695,
+      "step": 11783
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.62549003252982e-07,
+      "loss": 0.2924,
+      "step": 11784
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.608808074067896e-07,
+      "loss": 0.2725,
+      "step": 11785
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.592126115605973e-07,
+      "loss": 0.2966,
+      "step": 11786
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.57544415714405e-07,
+      "loss": 0.2596,
+      "step": 11787
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.558762198682126e-07,
+      "loss": 0.2677,
+      "step": 11788
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.542080240220203e-07,
+      "loss": 0.2615,
+      "step": 11789
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.525398281758279e-07,
+      "loss": 0.2691,
+      "step": 11790
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.508716323296357e-07,
+      "loss": 0.2623,
+      "step": 11791
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.492034364834432e-07,
+      "loss": 0.2714,
+      "step": 11792
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.475352406372509e-07,
+      "loss": 0.271,
+      "step": 11793
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.458670447910586e-07,
+      "loss": 0.2898,
+      "step": 11794
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.441988489448663e-07,
+      "loss": 0.2897,
+      "step": 11795
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.425306530986738e-07,
+      "loss": 0.2585,
+      "step": 11796
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.408624572524815e-07,
+      "loss": 0.2678,
+      "step": 11797
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.391942614062892e-07,
+      "loss": 0.261,
+      "step": 11798
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.375260655600967e-07,
+      "loss": 0.273,
+      "step": 11799
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.358578697139045e-07,
+      "loss": 0.2644,
+      "step": 11800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.341896738677121e-07,
+      "loss": 0.2665,
+      "step": 11801
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.325214780215199e-07,
+      "loss": 0.251,
+      "step": 11802
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.308532821753275e-07,
+      "loss": 0.2487,
+      "step": 11803
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.291850863291351e-07,
+      "loss": 0.2785,
+      "step": 11804
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.275168904829428e-07,
+      "loss": 0.2669,
+      "step": 11805
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.258486946367504e-07,
+      "loss": 0.2961,
+      "step": 11806
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.241804987905582e-07,
+      "loss": 0.2728,
+      "step": 11807
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.225123029443657e-07,
+      "loss": 0.2775,
+      "step": 11808
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.208441070981734e-07,
+      "loss": 0.294,
+      "step": 11809
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.191759112519811e-07,
+      "loss": 0.2755,
+      "step": 11810
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.175077154057887e-07,
+      "loss": 0.2709,
+      "step": 11811
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.158395195595963e-07,
+      "loss": 0.2802,
+      "step": 11812
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.141713237134041e-07,
+      "loss": 0.2534,
+      "step": 11813
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.125031278672117e-07,
+      "loss": 0.2718,
+      "step": 11814
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.108349320210192e-07,
+      "loss": 0.2603,
+      "step": 11815
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.09166736174827e-07,
+      "loss": 0.2792,
+      "step": 11816
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.074985403286346e-07,
+      "loss": 0.2581,
+      "step": 11817
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.058303444824424e-07,
+      "loss": 0.2733,
+      "step": 11818
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.041621486362499e-07,
+      "loss": 0.284,
+      "step": 11819
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.024939527900576e-07,
+      "loss": 0.2497,
+      "step": 11820
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.008257569438653e-07,
+      "loss": 0.257,
+      "step": 11821
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.99157561097673e-07,
+      "loss": 0.2691,
+      "step": 11822
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.974893652514806e-07,
+      "loss": 0.2838,
+      "step": 11823
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.958211694052883e-07,
+      "loss": 0.2897,
+      "step": 11824
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.941529735590959e-07,
+      "loss": 0.2702,
+      "step": 11825
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.924847777129036e-07,
+      "loss": 0.2786,
+      "step": 11826
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.908165818667112e-07,
+      "loss": 0.2719,
+      "step": 11827
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.891483860205188e-07,
+      "loss": 0.2383,
+      "step": 11828
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.874801901743266e-07,
+      "loss": 0.2787,
+      "step": 11829
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.858119943281342e-07,
+      "loss": 0.2993,
+      "step": 11830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.841437984819419e-07,
+      "loss": 0.2852,
+      "step": 11831
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.824756026357495e-07,
+      "loss": 0.2718,
+      "step": 11832
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.808074067895572e-07,
+      "loss": 0.2673,
+      "step": 11833
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.791392109433649e-07,
+      "loss": 0.267,
+      "step": 11834
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.774710150971724e-07,
+      "loss": 0.2566,
+      "step": 11835
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.758028192509801e-07,
+      "loss": 0.256,
+      "step": 11836
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.741346234047878e-07,
+      "loss": 0.2739,
+      "step": 11837
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.724664275585955e-07,
+      "loss": 0.2545,
+      "step": 11838
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.707982317124031e-07,
+      "loss": 0.2678,
+      "step": 11839
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.691300358662108e-07,
+      "loss": 0.2733,
+      "step": 11840
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.674618400200184e-07,
+      "loss": 0.28,
+      "step": 11841
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.657936441738262e-07,
+      "loss": 0.2602,
+      "step": 11842
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.641254483276337e-07,
+      "loss": 0.2605,
+      "step": 11843
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.624572524814413e-07,
+      "loss": 0.2616,
+      "step": 11844
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.607890566352491e-07,
+      "loss": 0.2691,
+      "step": 11845
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.591208607890566e-07,
+      "loss": 0.2608,
+      "step": 11846
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.574526649428644e-07,
+      "loss": 0.2618,
+      "step": 11847
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.55784469096672e-07,
+      "loss": 0.2591,
+      "step": 11848
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.541162732504797e-07,
+      "loss": 0.2676,
+      "step": 11849
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.524480774042873e-07,
+      "loss": 0.2723,
+      "step": 11850
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.50779881558095e-07,
+      "loss": 0.2515,
+      "step": 11851
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.491116857119026e-07,
+      "loss": 0.2596,
+      "step": 11852
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.474434898657104e-07,
+      "loss": 0.252,
+      "step": 11853
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.457752940195179e-07,
+      "loss": 0.2917,
+      "step": 11854
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.441070981733256e-07,
+      "loss": 0.2785,
+      "step": 11855
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.424389023271333e-07,
+      "loss": 0.2643,
+      "step": 11856
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.407707064809409e-07,
+      "loss": 0.2671,
+      "step": 11857
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.391025106347486e-07,
+      "loss": 0.2603,
+      "step": 11858
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.374343147885562e-07,
+      "loss": 0.2757,
+      "step": 11859
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.35766118942364e-07,
+      "loss": 0.2736,
+      "step": 11860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.340979230961716e-07,
+      "loss": 0.253,
+      "step": 11861
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.324297272499792e-07,
+      "loss": 0.2602,
+      "step": 11862
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.307615314037869e-07,
+      "loss": 0.2563,
+      "step": 11863
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.290933355575945e-07,
+      "loss": 0.2805,
+      "step": 11864
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.274251397114022e-07,
+      "loss": 0.2746,
+      "step": 11865
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.257569438652098e-07,
+      "loss": 0.2839,
+      "step": 11866
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.240887480190175e-07,
+      "loss": 0.276,
+      "step": 11867
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.224205521728251e-07,
+      "loss": 0.2535,
+      "step": 11868
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.207523563266329e-07,
+      "loss": 0.2679,
+      "step": 11869
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.190841604804404e-07,
+      "loss": 0.2734,
+      "step": 11870
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.174159646342482e-07,
+      "loss": 0.2841,
+      "step": 11871
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.157477687880558e-07,
+      "loss": 0.2813,
+      "step": 11872
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.140795729418633e-07,
+      "loss": 0.2817,
+      "step": 11873
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.124113770956711e-07,
+      "loss": 0.2628,
+      "step": 11874
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.107431812494787e-07,
+      "loss": 0.2835,
+      "step": 11875
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.090749854032865e-07,
+      "loss": 0.261,
+      "step": 11876
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.07406789557094e-07,
+      "loss": 0.2556,
+      "step": 11877
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.057385937109017e-07,
+      "loss": 0.2736,
+      "step": 11878
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.040703978647094e-07,
+      "loss": 0.2775,
+      "step": 11879
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.024022020185171e-07,
+      "loss": 0.264,
+      "step": 11880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.007340061723246e-07,
+      "loss": 0.241,
+      "step": 11881
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.990658103261323e-07,
+      "loss": 0.2662,
+      "step": 11882
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.9739761447994e-07,
+      "loss": 0.2796,
+      "step": 11883
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.957294186337476e-07,
+      "loss": 0.2575,
+      "step": 11884
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.940612227875553e-07,
+      "loss": 0.2604,
+      "step": 11885
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.923930269413629e-07,
+      "loss": 0.2931,
+      "step": 11886
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.907248310951707e-07,
+      "loss": 0.2684,
+      "step": 11887
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.890566352489783e-07,
+      "loss": 0.2811,
+      "step": 11888
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.87388439402786e-07,
+      "loss": 0.2681,
+      "step": 11889
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.857202435565936e-07,
+      "loss": 0.2643,
+      "step": 11890
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.840520477104013e-07,
+      "loss": 0.2572,
+      "step": 11891
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.82383851864209e-07,
+      "loss": 0.2639,
+      "step": 11892
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.807156560180165e-07,
+      "loss": 0.2476,
+      "step": 11893
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.790474601718242e-07,
+      "loss": 0.2587,
+      "step": 11894
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.773792643256319e-07,
+      "loss": 0.2948,
+      "step": 11895
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.757110684794396e-07,
+      "loss": 0.2668,
+      "step": 11896
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.740428726332471e-07,
+      "loss": 0.2727,
+      "step": 11897
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.723746767870549e-07,
+      "loss": 0.2617,
+      "step": 11898
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.707064809408625e-07,
+      "loss": 0.2663,
+      "step": 11899
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.690382850946703e-07,
+      "loss": 0.2512,
+      "step": 11900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.673700892484778e-07,
+      "loss": 0.2717,
+      "step": 11901
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.657018934022854e-07,
+      "loss": 0.2582,
+      "step": 11902
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.640336975560932e-07,
+      "loss": 0.2545,
+      "step": 11903
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.623655017099008e-07,
+      "loss": 0.2659,
+      "step": 11904
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.606973058637085e-07,
+      "loss": 0.2649,
+      "step": 11905
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.590291100175161e-07,
+      "loss": 0.2662,
+      "step": 11906
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.573609141713238e-07,
+      "loss": 0.2516,
+      "step": 11907
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.556927183251315e-07,
+      "loss": 0.2745,
+      "step": 11908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.540245224789391e-07,
+      "loss": 0.2767,
+      "step": 11909
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.523563266327467e-07,
+      "loss": 0.2614,
+      "step": 11910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.506881307865544e-07,
+      "loss": 0.2734,
+      "step": 11911
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.49019934940362e-07,
+      "loss": 0.2449,
+      "step": 11912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.473517390941696e-07,
+      "loss": 0.2606,
+      "step": 11913
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.456835432479774e-07,
+      "loss": 0.2799,
+      "step": 11914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.44015347401785e-07,
+      "loss": 0.2705,
+      "step": 11915
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.423471515555927e-07,
+      "loss": 0.2622,
+      "step": 11916
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.406789557094003e-07,
+      "loss": 0.2838,
+      "step": 11917
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.390107598632081e-07,
+      "loss": 0.2682,
+      "step": 11918
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.373425640170157e-07,
+      "loss": 0.2594,
+      "step": 11919
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.356743681708233e-07,
+      "loss": 0.2488,
+      "step": 11920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.34006172324631e-07,
+      "loss": 0.2771,
+      "step": 11921
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.323379764784386e-07,
+      "loss": 0.2775,
+      "step": 11922
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.306697806322463e-07,
+      "loss": 0.2588,
+      "step": 11923
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.290015847860539e-07,
+      "loss": 0.2708,
+      "step": 11924
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.273333889398616e-07,
+      "loss": 0.2717,
+      "step": 11925
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.256651930936692e-07,
+      "loss": 0.2609,
+      "step": 11926
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.23996997247477e-07,
+      "loss": 0.2574,
+      "step": 11927
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.223288014012845e-07,
+      "loss": 0.2684,
+      "step": 11928
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.206606055550923e-07,
+      "loss": 0.2721,
+      "step": 11929
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.189924097088999e-07,
+      "loss": 0.267,
+      "step": 11930
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.173242138627075e-07,
+      "loss": 0.2651,
+      "step": 11931
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.156560180165152e-07,
+      "loss": 0.2708,
+      "step": 11932
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.139878221703228e-07,
+      "loss": 0.2682,
+      "step": 11933
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.123196263241306e-07,
+      "loss": 0.2558,
+      "step": 11934
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.106514304779382e-07,
+      "loss": 0.2631,
+      "step": 11935
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.089832346317458e-07,
+      "loss": 0.2563,
+      "step": 11936
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.073150387855535e-07,
+      "loss": 0.2599,
+      "step": 11937
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.056468429393612e-07,
+      "loss": 0.2699,
+      "step": 11938
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.039786470931688e-07,
+      "loss": 0.2842,
+      "step": 11939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.023104512469764e-07,
+      "loss": 0.2617,
+      "step": 11940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.006422554007841e-07,
+      "loss": 0.2852,
+      "step": 11941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.989740595545917e-07,
+      "loss": 0.2746,
+      "step": 11942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.973058637083995e-07,
+      "loss": 0.2523,
+      "step": 11943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.95637667862207e-07,
+      "loss": 0.2715,
+      "step": 11944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.939694720160148e-07,
+      "loss": 0.2573,
+      "step": 11945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.923012761698224e-07,
+      "loss": 0.2593,
+      "step": 11946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.906330803236302e-07,
+      "loss": 0.2769,
+      "step": 11947
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.889648844774377e-07,
+      "loss": 0.2551,
+      "step": 11948
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.872966886312453e-07,
+      "loss": 0.2685,
+      "step": 11949
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.856284927850531e-07,
+      "loss": 0.2852,
+      "step": 11950
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.839602969388606e-07,
+      "loss": 0.2857,
+      "step": 11951
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.822921010926683e-07,
+      "loss": 0.2753,
+      "step": 11952
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.80623905246476e-07,
+      "loss": 0.2648,
+      "step": 11953
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.789557094002837e-07,
+      "loss": 0.2628,
+      "step": 11954
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.772875135540912e-07,
+      "loss": 0.2575,
+      "step": 11955
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.75619317707899e-07,
+      "loss": 0.2578,
+      "step": 11956
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.739511218617066e-07,
+      "loss": 0.2485,
+      "step": 11957
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.722829260155144e-07,
+      "loss": 0.2802,
+      "step": 11958
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.706147301693219e-07,
+      "loss": 0.2684,
+      "step": 11959
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.689465343231295e-07,
+      "loss": 0.2721,
+      "step": 11960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.672783384769373e-07,
+      "loss": 0.2892,
+      "step": 11961
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.656101426307449e-07,
+      "loss": 0.2881,
+      "step": 11962
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.639419467845526e-07,
+      "loss": 0.2591,
+      "step": 11963
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.622737509383602e-07,
+      "loss": 0.263,
+      "step": 11964
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.606055550921679e-07,
+      "loss": 0.2769,
+      "step": 11965
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.589373592459756e-07,
+      "loss": 0.2785,
+      "step": 11966
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.572691633997832e-07,
+      "loss": 0.269,
+      "step": 11967
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.556009675535908e-07,
+      "loss": 0.271,
+      "step": 11968
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.539327717073985e-07,
+      "loss": 0.2618,
+      "step": 11969
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.522645758612062e-07,
+      "loss": 0.2713,
+      "step": 11970
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.505963800150137e-07,
+      "loss": 0.2938,
+      "step": 11971
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.489281841688215e-07,
+      "loss": 0.2659,
+      "step": 11972
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.472599883226291e-07,
+      "loss": 0.2628,
+      "step": 11973
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.455917924764369e-07,
+      "loss": 0.2401,
+      "step": 11974
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.439235966302444e-07,
+      "loss": 0.2572,
+      "step": 11975
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.422554007840522e-07,
+      "loss": 0.2506,
+      "step": 11976
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.405872049378598e-07,
+      "loss": 0.2727,
+      "step": 11977
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.389190090916673e-07,
+      "loss": 0.2694,
+      "step": 11978
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.372508132454751e-07,
+      "loss": 0.2738,
+      "step": 11979
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.355826173992827e-07,
+      "loss": 0.2704,
+      "step": 11980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.339144215530904e-07,
+      "loss": 0.2629,
+      "step": 11981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.32246225706898e-07,
+      "loss": 0.2761,
+      "step": 11982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.305780298607057e-07,
+      "loss": 0.2575,
+      "step": 11983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.289098340145133e-07,
+      "loss": 0.2635,
+      "step": 11984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.272416381683211e-07,
+      "loss": 0.2639,
+      "step": 11985
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.255734423221286e-07,
+      "loss": 0.2619,
+      "step": 11986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.239052464759363e-07,
+      "loss": 0.2618,
+      "step": 11987
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.22237050629744e-07,
+      "loss": 0.2685,
+      "step": 11988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.205688547835517e-07,
+      "loss": 0.254,
+      "step": 11989
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.189006589373593e-07,
+      "loss": 0.272,
+      "step": 11990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.17232463091167e-07,
+      "loss": 0.2745,
+      "step": 11991
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.155642672449747e-07,
+      "loss": 0.2493,
+      "step": 11992
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.138960713987823e-07,
+      "loss": 0.2497,
+      "step": 11993
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.122278755525899e-07,
+      "loss": 0.258,
+      "step": 11994
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.105596797063976e-07,
+      "loss": 0.2857,
+      "step": 11995
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.088914838602052e-07,
+      "loss": 0.2612,
+      "step": 11996
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.072232880140129e-07,
+      "loss": 0.2779,
+      "step": 11997
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.055550921678205e-07,
+      "loss": 0.2791,
+      "step": 11998
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.038868963216282e-07,
+      "loss": 0.258,
+      "step": 11999
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.022187004754359e-07,
+      "loss": 0.2684,
+      "step": 12000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.005505046292436e-07,
+      "loss": 0.2668,
+      "step": 12001
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.988823087830511e-07,
+      "loss": 0.2753,
+      "step": 12002
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.972141129368588e-07,
+      "loss": 0.2627,
+      "step": 12003
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.955459170906665e-07,
+      "loss": 0.2543,
+      "step": 12004
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.938777212444742e-07,
+      "loss": 0.2753,
+      "step": 12005
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.922095253982818e-07,
+      "loss": 0.264,
+      "step": 12006
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.905413295520895e-07,
+      "loss": 0.2605,
+      "step": 12007
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.888731337058972e-07,
+      "loss": 0.2483,
+      "step": 12008
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.872049378597048e-07,
+      "loss": 0.2624,
+      "step": 12009
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.855367420135124e-07,
+      "loss": 0.2861,
+      "step": 12010
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.838685461673201e-07,
+      "loss": 0.2705,
+      "step": 12011
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.822003503211277e-07,
+      "loss": 0.2417,
+      "step": 12012
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.805321544749353e-07,
+      "loss": 0.2608,
+      "step": 12013
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.78863958628743e-07,
+      "loss": 0.2644,
+      "step": 12014
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.771957627825507e-07,
+      "loss": 0.2542,
+      "step": 12015
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.755275669363584e-07,
+      "loss": 0.2415,
+      "step": 12016
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.73859371090166e-07,
+      "loss": 0.2818,
+      "step": 12017
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.721911752439737e-07,
+      "loss": 0.2354,
+      "step": 12018
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.705229793977814e-07,
+      "loss": 0.2529,
+      "step": 12019
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.688547835515891e-07,
+      "loss": 0.2596,
+      "step": 12020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.671865877053966e-07,
+      "loss": 0.2718,
+      "step": 12021
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.655183918592043e-07,
+      "loss": 0.2802,
+      "step": 12022
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.63850196013012e-07,
+      "loss": 0.2633,
+      "step": 12023
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.621820001668197e-07,
+      "loss": 0.2586,
+      "step": 12024
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.605138043206272e-07,
+      "loss": 0.2653,
+      "step": 12025
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.588456084744349e-07,
+      "loss": 0.2785,
+      "step": 12026
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.571774126282426e-07,
+      "loss": 0.2688,
+      "step": 12027
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.555092167820503e-07,
+      "loss": 0.2802,
+      "step": 12028
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.538410209358579e-07,
+      "loss": 0.2662,
+      "step": 12029
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.521728250896656e-07,
+      "loss": 0.2526,
+      "step": 12030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.505046292434732e-07,
+      "loss": 0.2854,
+      "step": 12031
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.488364333972809e-07,
+      "loss": 0.2579,
+      "step": 12032
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.471682375510885e-07,
+      "loss": 0.2748,
+      "step": 12033
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.455000417048962e-07,
+      "loss": 0.2795,
+      "step": 12034
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.438318458587039e-07,
+      "loss": 0.2795,
+      "step": 12035
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.421636500125116e-07,
+      "loss": 0.2685,
+      "step": 12036
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.404954541663191e-07,
+      "loss": 0.2584,
+      "step": 12037
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.388272583201268e-07,
+      "loss": 0.2585,
+      "step": 12038
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.371590624739345e-07,
+      "loss": 0.2829,
+      "step": 12039
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.354908666277422e-07,
+      "loss": 0.2866,
+      "step": 12040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.338226707815497e-07,
+      "loss": 0.2745,
+      "step": 12041
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.321544749353574e-07,
+      "loss": 0.2626,
+      "step": 12042
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.304862790891651e-07,
+      "loss": 0.2596,
+      "step": 12043
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.288180832429728e-07,
+      "loss": 0.2835,
+      "step": 12044
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.271498873967804e-07,
+      "loss": 0.2564,
+      "step": 12045
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.254816915505881e-07,
+      "loss": 0.2549,
+      "step": 12046
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.238134957043958e-07,
+      "loss": 0.2778,
+      "step": 12047
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.221452998582035e-07,
+      "loss": 0.2536,
+      "step": 12048
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.204771040120111e-07,
+      "loss": 0.2627,
+      "step": 12049
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.188089081658187e-07,
+      "loss": 0.2514,
+      "step": 12050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.171407123196264e-07,
+      "loss": 0.2536,
+      "step": 12051
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.15472516473434e-07,
+      "loss": 0.2618,
+      "step": 12052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.138043206272416e-07,
+      "loss": 0.2668,
+      "step": 12053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.121361247810493e-07,
+      "loss": 0.2626,
+      "step": 12054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.10467928934857e-07,
+      "loss": 0.251,
+      "step": 12055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.087997330886646e-07,
+      "loss": 0.2559,
+      "step": 12056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.071315372424723e-07,
+      "loss": 0.2739,
+      "step": 12057
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.0546334139628e-07,
+      "loss": 0.2836,
+      "step": 12058
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.037951455500877e-07,
+      "loss": 0.2565,
+      "step": 12059
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.021269497038952e-07,
+      "loss": 0.2653,
+      "step": 12060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.004587538577029e-07,
+      "loss": 0.2632,
+      "step": 12061
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.987905580115106e-07,
+      "loss": 0.2621,
+      "step": 12062
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.971223621653183e-07,
+      "loss": 0.2858,
+      "step": 12063
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.954541663191259e-07,
+      "loss": 0.2683,
+      "step": 12064
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.937859704729336e-07,
+      "loss": 0.2743,
+      "step": 12065
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.921177746267412e-07,
+      "loss": 0.2667,
+      "step": 12066
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.904495787805489e-07,
+      "loss": 0.2706,
+      "step": 12067
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.887813829343565e-07,
+      "loss": 0.2686,
+      "step": 12068
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.871131870881641e-07,
+      "loss": 0.276,
+      "step": 12069
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.854449912419718e-07,
+      "loss": 0.2678,
+      "step": 12070
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.837767953957795e-07,
+      "loss": 0.2893,
+      "step": 12071
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.821085995495871e-07,
+      "loss": 0.2583,
+      "step": 12072
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.804404037033948e-07,
+      "loss": 0.2576,
+      "step": 12073
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.787722078572025e-07,
+      "loss": 0.2752,
+      "step": 12074
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.771040120110102e-07,
+      "loss": 0.2688,
+      "step": 12075
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7543581616481783e-07,
+      "loss": 0.2734,
+      "step": 12076
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7376762031862545e-07,
+      "loss": 0.26,
+      "step": 12077
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7209942447243313e-07,
+      "loss": 0.2774,
+      "step": 12078
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.7043122862624075e-07,
+      "loss": 0.2606,
+      "step": 12079
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.687630327800484e-07,
+      "loss": 0.2582,
+      "step": 12080
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.6709483693385605e-07,
+      "loss": 0.2585,
+      "step": 12081
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.654266410876637e-07,
+      "loss": 0.2767,
+      "step": 12082
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.637584452414714e-07,
+      "loss": 0.2801,
+      "step": 12083
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.620902493952791e-07,
+      "loss": 0.2501,
+      "step": 12084
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.604220535490867e-07,
+      "loss": 0.2858,
+      "step": 12085
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5875385770289437e-07,
+      "loss": 0.2546,
+      "step": 12086
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5708566185670205e-07,
+      "loss": 0.2772,
+      "step": 12087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.554174660105096e-07,
+      "loss": 0.2717,
+      "step": 12088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.537492701643173e-07,
+      "loss": 0.2612,
+      "step": 12089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5208107431812497e-07,
+      "loss": 0.2571,
+      "step": 12090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.5041287847193264e-07,
+      "loss": 0.2742,
+      "step": 12091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.487446826257403e-07,
+      "loss": 0.2599,
+      "step": 12092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.4707648677954794e-07,
+      "loss": 0.2641,
+      "step": 12093
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.454082909333556e-07,
+      "loss": 0.2624,
+      "step": 12094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.437400950871633e-07,
+      "loss": 0.2704,
+      "step": 12095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.4207189924097097e-07,
+      "loss": 0.2569,
+      "step": 12096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.404037033947786e-07,
+      "loss": 0.2781,
+      "step": 12097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.387355075485862e-07,
+      "loss": 0.2506,
+      "step": 12098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.370673117023939e-07,
+      "loss": 0.2598,
+      "step": 12099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3539911585620156e-07,
+      "loss": 0.2682,
+      "step": 12100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.337309200100092e-07,
+      "loss": 0.2699,
+      "step": 12101
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3206272416381686e-07,
+      "loss": 0.2663,
+      "step": 12102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.3039452831762454e-07,
+      "loss": 0.267,
+      "step": 12103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.287263324714322e-07,
+      "loss": 0.2636,
+      "step": 12104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2705813662523983e-07,
+      "loss": 0.2679,
+      "step": 12105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.253899407790475e-07,
+      "loss": 0.2562,
+      "step": 12106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.237217449328552e-07,
+      "loss": 0.2721,
+      "step": 12107
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.220535490866628e-07,
+      "loss": 0.2626,
+      "step": 12108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.2038535324047043e-07,
+      "loss": 0.2538,
+      "step": 12109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.187171573942781e-07,
+      "loss": 0.2602,
+      "step": 12110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.170489615480858e-07,
+      "loss": 0.2788,
+      "step": 12111
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1538076570189345e-07,
+      "loss": 0.2473,
+      "step": 12112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.137125698557011e-07,
+      "loss": 0.2494,
+      "step": 12113
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1204437400950875e-07,
+      "loss": 0.2563,
+      "step": 12114
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.1037617816331643e-07,
+      "loss": 0.2505,
+      "step": 12115
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.087079823171241e-07,
+      "loss": 0.2752,
+      "step": 12116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0703978647093167e-07,
+      "loss": 0.2629,
+      "step": 12117
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.0537159062473935e-07,
+      "loss": 0.2652,
+      "step": 12118
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.03703394778547e-07,
+      "loss": 0.2529,
+      "step": 12119
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.020351989323547e-07,
+      "loss": 0.2538,
+      "step": 12120
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.003670030861623e-07,
+      "loss": 0.2491,
+      "step": 12121
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9869880723997e-07,
+      "loss": 0.2427,
+      "step": 12122
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9703061139377767e-07,
+      "loss": 0.2726,
+      "step": 12123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9536241554758535e-07,
+      "loss": 0.261,
+      "step": 12124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.93694219701393e-07,
+      "loss": 0.2756,
+      "step": 12125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9202602385520064e-07,
+      "loss": 0.2605,
+      "step": 12126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.9035782800900827e-07,
+      "loss": 0.2891,
+      "step": 12127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8868963216281594e-07,
+      "loss": 0.258,
+      "step": 12128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8702143631662356e-07,
+      "loss": 0.2527,
+      "step": 12129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8535324047043124e-07,
+      "loss": 0.2869,
+      "step": 12130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.836850446242389e-07,
+      "loss": 0.2841,
+      "step": 12131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.820168487780466e-07,
+      "loss": 0.2841,
+      "step": 12132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8034865293185427e-07,
+      "loss": 0.2542,
+      "step": 12133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.786804570856619e-07,
+      "loss": 0.2536,
+      "step": 12134
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7701226123946956e-07,
+      "loss": 0.2865,
+      "step": 12135
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.753440653932772e-07,
+      "loss": 0.2576,
+      "step": 12136
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.736758695470848e-07,
+      "loss": 0.272,
+      "step": 12137
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.720076737008925e-07,
+      "loss": 0.2693,
+      "step": 12138
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7033947785470016e-07,
+      "loss": 0.2629,
+      "step": 12139
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6867128200850783e-07,
+      "loss": 0.2763,
+      "step": 12140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.670030861623155e-07,
+      "loss": 0.2625,
+      "step": 12141
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6533489031612313e-07,
+      "loss": 0.2616,
+      "step": 12142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.636666944699308e-07,
+      "loss": 0.2533,
+      "step": 12143
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.619984986237385e-07,
+      "loss": 0.2542,
+      "step": 12144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.6033030277754616e-07,
+      "loss": 0.2861,
+      "step": 12145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5866210693135373e-07,
+      "loss": 0.2663,
+      "step": 12146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.569939110851614e-07,
+      "loss": 0.2717,
+      "step": 12147
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.553257152389691e-07,
+      "loss": 0.2541,
+      "step": 12148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5365751939277675e-07,
+      "loss": 0.2686,
+      "step": 12149
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.519893235465844e-07,
+      "loss": 0.2482,
+      "step": 12150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5032112770039205e-07,
+      "loss": 0.275,
+      "step": 12151
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.486529318541997e-07,
+      "loss": 0.2524,
+      "step": 12152
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.469847360080074e-07,
+      "loss": 0.2819,
+      "step": 12153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.453165401618151e-07,
+      "loss": 0.2588,
+      "step": 12154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4364834431562265e-07,
+      "loss": 0.2677,
+      "step": 12155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.419801484694303e-07,
+      "loss": 0.2614,
+      "step": 12156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.40311952623238e-07,
+      "loss": 0.2788,
+      "step": 12157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.386437567770456e-07,
+      "loss": 0.2451,
+      "step": 12158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.369755609308533e-07,
+      "loss": 0.2866,
+      "step": 12159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3530736508466097e-07,
+      "loss": 0.2687,
+      "step": 12160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3363916923846864e-07,
+      "loss": 0.2862,
+      "step": 12161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.319709733922763e-07,
+      "loss": 0.266,
+      "step": 12162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3030277754608394e-07,
+      "loss": 0.2758,
+      "step": 12163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.286345816998916e-07,
+      "loss": 0.2518,
+      "step": 12164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2696638585369924e-07,
+      "loss": 0.2751,
+      "step": 12165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2529819000750686e-07,
+      "loss": 0.2699,
+      "step": 12166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.2362999416131454e-07,
+      "loss": 0.2564,
+      "step": 12167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.219617983151222e-07,
+      "loss": 0.2551,
+      "step": 12168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.202936024689299e-07,
+      "loss": 0.2509,
+      "step": 12169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1862540662273756e-07,
+      "loss": 0.2729,
+      "step": 12170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.169572107765452e-07,
+      "loss": 0.2678,
+      "step": 12171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1528901493035286e-07,
+      "loss": 0.2593,
+      "step": 12172
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1362081908416054e-07,
+      "loss": 0.2664,
+      "step": 12173
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1195262323796816e-07,
+      "loss": 0.2495,
+      "step": 12174
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.1028442739177583e-07,
+      "loss": 0.2654,
+      "step": 12175
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.086162315455835e-07,
+      "loss": 0.2588,
+      "step": 12176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0694803569939113e-07,
+      "loss": 0.2571,
+      "step": 12177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.052798398531988e-07,
+      "loss": 0.2747,
+      "step": 12178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.0361164400700643e-07,
+      "loss": 0.2617,
+      "step": 12179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.019434481608141e-07,
+      "loss": 0.2504,
+      "step": 12180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.002752523146218e-07,
+      "loss": 0.2787,
+      "step": 12181
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.986070564684294e-07,
+      "loss": 0.2433,
+      "step": 12182
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.969388606222371e-07,
+      "loss": 0.2537,
+      "step": 12183
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9527066477604475e-07,
+      "loss": 0.2518,
+      "step": 12184
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.936024689298524e-07,
+      "loss": 0.2364,
+      "step": 12185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9193427308366005e-07,
+      "loss": 0.251,
+      "step": 12186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.9026607723746767e-07,
+      "loss": 0.2597,
+      "step": 12187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.8859788139127535e-07,
+      "loss": 0.2741,
+      "step": 12188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.86929685545083e-07,
+      "loss": 0.2754,
+      "step": 12189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.852614896988907e-07,
+      "loss": 0.2772,
+      "step": 12190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.835932938526983e-07,
+      "loss": 0.288,
+      "step": 12191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.81925098006506e-07,
+      "loss": 0.2666,
+      "step": 12192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.802569021603136e-07,
+      "loss": 0.2367,
+      "step": 12193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.785887063141213e-07,
+      "loss": 0.269,
+      "step": 12194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7692051046792897e-07,
+      "loss": 0.2637,
+      "step": 12195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.752523146217366e-07,
+      "loss": 0.2593,
+      "step": 12196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7358411877554427e-07,
+      "loss": 0.2774,
+      "step": 12197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7191592292935194e-07,
+      "loss": 0.2497,
+      "step": 12198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.7024772708315957e-07,
+      "loss": 0.2673,
+      "step": 12199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6857953123696724e-07,
+      "loss": 0.2839,
+      "step": 12200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6691133539077486e-07,
+      "loss": 0.2734,
+      "step": 12201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6524313954458254e-07,
+      "loss": 0.2556,
+      "step": 12202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.635749436983902e-07,
+      "loss": 0.2566,
+      "step": 12203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.619067478521979e-07,
+      "loss": 0.2834,
+      "step": 12204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.6023855200600556e-07,
+      "loss": 0.2643,
+      "step": 12205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.585703561598132e-07,
+      "loss": 0.2781,
+      "step": 12206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.569021603136208e-07,
+      "loss": 0.2662,
+      "step": 12207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.552339644674285e-07,
+      "loss": 0.2668,
+      "step": 12208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5356576862123616e-07,
+      "loss": 0.2554,
+      "step": 12209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5189757277504383e-07,
+      "loss": 0.2582,
+      "step": 12210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.5022937692885146e-07,
+      "loss": 0.2621,
+      "step": 12211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4856118108265913e-07,
+      "loss": 0.2549,
+      "step": 12212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.468929852364668e-07,
+      "loss": 0.2684,
+      "step": 12213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4522478939027443e-07,
+      "loss": 0.2698,
+      "step": 12214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4355659354408205e-07,
+      "loss": 0.2813,
+      "step": 12215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.4188839769788973e-07,
+      "loss": 0.2438,
+      "step": 12216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 2.402202018516974e-07,
+      "loss": 0.2823,
+      "step": 12217
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.385520060055051e-07,
+      "loss": 0.2726,
+      "step": 12218
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3688381015931273e-07,
+      "loss": 0.2761,
+      "step": 12219
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3521561431312038e-07,
+      "loss": 0.246,
+      "step": 12220
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3354741846692802e-07,
+      "loss": 0.2445,
+      "step": 12221
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.318792226207357e-07,
+      "loss": 0.2688,
+      "step": 12222
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.3021102677454335e-07,
+      "loss": 0.254,
+      "step": 12223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2854283092835102e-07,
+      "loss": 0.2724,
+      "step": 12224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2687463508215865e-07,
+      "loss": 0.2685,
+      "step": 12225
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2520643923596632e-07,
+      "loss": 0.2572,
+      "step": 12226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2353824338977397e-07,
+      "loss": 0.2621,
+      "step": 12227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.2187004754358165e-07,
+      "loss": 0.2529,
+      "step": 12228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.202018516973893e-07,
+      "loss": 0.2601,
+      "step": 12229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1853365585119694e-07,
+      "loss": 0.2516,
+      "step": 12230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.168654600050046e-07,
+      "loss": 0.2662,
+      "step": 12231
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1519726415881227e-07,
+      "loss": 0.2585,
+      "step": 12232
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1352906831261992e-07,
+      "loss": 0.26,
+      "step": 12233
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.118608724664276e-07,
+      "loss": 0.2615,
+      "step": 12234
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.1019267662023521e-07,
+      "loss": 0.2451,
+      "step": 12235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.085244807740429e-07,
+      "loss": 0.2763,
+      "step": 12236
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0685628492785054e-07,
+      "loss": 0.2479,
+      "step": 12237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0518808908165821e-07,
+      "loss": 0.2534,
+      "step": 12238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0351989323546584e-07,
+      "loss": 0.2764,
+      "step": 12239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.018516973892735e-07,
+      "loss": 0.2549,
+      "step": 12240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.0018350154308116e-07,
+      "loss": 0.2663,
+      "step": 12241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9851530569688884e-07,
+      "loss": 0.2455,
+      "step": 12242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.968471098506965e-07,
+      "loss": 0.2562,
+      "step": 12243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9517891400450413e-07,
+      "loss": 0.2652,
+      "step": 12244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9351071815831178e-07,
+      "loss": 0.2554,
+      "step": 12245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9184252231211946e-07,
+      "loss": 0.2564,
+      "step": 12246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9017432646592713e-07,
+      "loss": 0.2738,
+      "step": 12247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8850613061973478e-07,
+      "loss": 0.2619,
+      "step": 12248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.868379347735424e-07,
+      "loss": 0.2487,
+      "step": 12249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8516973892735008e-07,
+      "loss": 0.2753,
+      "step": 12250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8350154308115775e-07,
+      "loss": 0.2738,
+      "step": 12251
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.818333472349654e-07,
+      "loss": 0.2501,
+      "step": 12252
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8016515138877308e-07,
+      "loss": 0.2655,
+      "step": 12253
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.784969555425807e-07,
+      "loss": 0.2696,
+      "step": 12254
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7682875969638838e-07,
+      "loss": 0.2547,
+      "step": 12255
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7516056385019603e-07,
+      "loss": 0.2728,
+      "step": 12256
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.734923680040037e-07,
+      "loss": 0.2694,
+      "step": 12257
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7182417215781132e-07,
+      "loss": 0.2597,
+      "step": 12258
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.70155976311619e-07,
+      "loss": 0.2718,
+      "step": 12259
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6848778046542665e-07,
+      "loss": 0.2655,
+      "step": 12260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6681958461923432e-07,
+      "loss": 0.2737,
+      "step": 12261
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6515138877304197e-07,
+      "loss": 0.2646,
+      "step": 12262
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6348319292684962e-07,
+      "loss": 0.273,
+      "step": 12263
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6181499708065727e-07,
+      "loss": 0.2769,
+      "step": 12264
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6014680123446494e-07,
+      "loss": 0.265,
+      "step": 12265
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.584786053882726e-07,
+      "loss": 0.2475,
+      "step": 12266
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5681040954208027e-07,
+      "loss": 0.2765,
+      "step": 12267
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5514221369588792e-07,
+      "loss": 0.2467,
+      "step": 12268
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5347401784969557e-07,
+      "loss": 0.2665,
+      "step": 12269
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5180582200350321e-07,
+      "loss": 0.2774,
+      "step": 12270
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.501376261573109e-07,
+      "loss": 0.2671,
+      "step": 12271
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4846943031111854e-07,
+      "loss": 0.2708,
+      "step": 12272
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.468012344649262e-07,
+      "loss": 0.2589,
+      "step": 12273
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4513303861873384e-07,
+      "loss": 0.2774,
+      "step": 12274
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.434648427725415e-07,
+      "loss": 0.2399,
+      "step": 12275
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4179664692634916e-07,
+      "loss": 0.275,
+      "step": 12276
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.401284510801568e-07,
+      "loss": 0.2884,
+      "step": 12277
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3846025523396448e-07,
+      "loss": 0.253,
+      "step": 12278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3679205938777213e-07,
+      "loss": 0.2793,
+      "step": 12279
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3512386354157978e-07,
+      "loss": 0.2662,
+      "step": 12280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3345566769538743e-07,
+      "loss": 0.281,
+      "step": 12281
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.317874718491951e-07,
+      "loss": 0.2495,
+      "step": 12282
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3011927600300278e-07,
+      "loss": 0.2799,
+      "step": 12283
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.284510801568104e-07,
+      "loss": 0.2901,
+      "step": 12284
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2678288431061808e-07,
+      "loss": 0.2657,
+      "step": 12285
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2511468846442573e-07,
+      "loss": 0.2611,
+      "step": 12286
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.234464926182334e-07,
+      "loss": 0.2763,
+      "step": 12287
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2177829677204103e-07,
+      "loss": 0.2716,
+      "step": 12288
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.201101009258487e-07,
+      "loss": 0.3039,
+      "step": 12289
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1844190507965636e-07,
+      "loss": 0.2613,
+      "step": 12290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1677370923346401e-07,
+      "loss": 0.2653,
+      "step": 12291
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1510551338727167e-07,
+      "loss": 0.2907,
+      "step": 12292
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1343731754107932e-07,
+      "loss": 0.2748,
+      "step": 12293
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1176912169488699e-07,
+      "loss": 0.2634,
+      "step": 12294
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1010092584869465e-07,
+      "loss": 0.2785,
+      "step": 12295
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.084327300025023e-07,
+      "loss": 0.2682,
+      "step": 12296
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0676453415630996e-07,
+      "loss": 0.276,
+      "step": 12297
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0509633831011761e-07,
+      "loss": 0.2777,
+      "step": 12298
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0342814246392527e-07,
+      "loss": 0.2572,
+      "step": 12299
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0175994661773292e-07,
+      "loss": 0.266,
+      "step": 12300
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0009175077154058e-07,
+      "loss": 0.2552,
+      "step": 12301
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.842355492534826e-08,
+      "loss": 0.2704,
+      "step": 12302
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.675535907915589e-08,
+      "loss": 0.2418,
+      "step": 12303
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.508716323296357e-08,
+      "loss": 0.2448,
+      "step": 12304
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.34189673867712e-08,
+      "loss": 0.2698,
+      "step": 12305
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.175077154057888e-08,
+      "loss": 0.2634,
+      "step": 12306
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.008257569438654e-08,
+      "loss": 0.2631,
+      "step": 12307
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.841437984819419e-08,
+      "loss": 0.2513,
+      "step": 12308
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.674618400200185e-08,
+      "loss": 0.2672,
+      "step": 12309
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.50779881558095e-08,
+      "loss": 0.2683,
+      "step": 12310
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.340979230961716e-08,
+      "loss": 0.2607,
+      "step": 12311
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.174159646342481e-08,
+      "loss": 0.2707,
+      "step": 12312
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.007340061723247e-08,
+      "loss": 0.2679,
+      "step": 12313
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.840520477104013e-08,
+      "loss": 0.2544,
+      "step": 12314
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.673700892484778e-08,
+      "loss": 0.2826,
+      "step": 12315
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.506881307865545e-08,
+      "loss": 0.2582,
+      "step": 12316
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.34006172324631e-08,
+      "loss": 0.2912,
+      "step": 12317
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.173242138627076e-08,
+      "loss": 0.2719,
+      "step": 12318
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.00642255400784e-08,
+      "loss": 0.2524,
+      "step": 12319
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.839602969388607e-08,
+      "loss": 0.2685,
+      "step": 12320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.672783384769372e-08,
+      "loss": 0.2514,
+      "step": 12321
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.505963800150139e-08,
+      "loss": 0.2615,
+      "step": 12322
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.339144215530904e-08,
+      "loss": 0.2576,
+      "step": 12323
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.17232463091167e-08,
+      "loss": 0.268,
+      "step": 12324
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.005505046292435e-08,
+      "loss": 0.2763,
+      "step": 12325
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.8386854616732006e-08,
+      "loss": 0.2594,
+      "step": 12326
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.671865877053966e-08,
+      "loss": 0.2447,
+      "step": 12327
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.5050462924347324e-08,
+      "loss": 0.2562,
+      "step": 12328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.338226707815498e-08,
+      "loss": 0.2687,
+      "step": 12329
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.1714071231962635e-08,
+      "loss": 0.2891,
+      "step": 12330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.004587538577029e-08,
+      "loss": 0.2513,
+      "step": 12331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.8377679539577946e-08,
+      "loss": 0.2674,
+      "step": 12332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.67094836933856e-08,
+      "loss": 0.2685,
+      "step": 12333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.504128784719327e-08,
+      "loss": 0.2565,
+      "step": 12334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3373092001000925e-08,
+      "loss": 0.2532,
+      "step": 12335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.170489615480858e-08,
+      "loss": 0.2472,
+      "step": 12336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0036700308616236e-08,
+      "loss": 0.2675,
+      "step": 12337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.836850446242389e-08,
+      "loss": 0.2532,
+      "step": 12338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.670030861623155e-08,
+      "loss": 0.2655,
+      "step": 12339
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.50321127700392e-08,
+      "loss": 0.2537,
+      "step": 12340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.336391692384686e-08,
+      "loss": 0.2585,
+      "step": 12341
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.169572107765452e-08,
+      "loss": 0.2756,
+      "step": 12342
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0027525231462175e-08,
+      "loss": 0.3097,
+      "step": 12343
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.835932938526983e-08,
+      "loss": 0.2556,
+      "step": 12344
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.669113353907749e-08,
+      "loss": 0.2446,
+      "step": 12345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.5022937692885145e-08,
+      "loss": 0.2598,
+      "step": 12346
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.33547418466928e-08,
+      "loss": 0.2737,
+      "step": 12347
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.1686546000500463e-08,
+      "loss": 0.2605,
+      "step": 12348
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0018350154308118e-08,
+      "loss": 0.2709,
+      "step": 12349
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8350154308115773e-08,
+      "loss": 0.25,
+      "step": 12350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.668195846192343e-08,
+      "loss": 0.2729,
+      "step": 12351
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5013762615731088e-08,
+      "loss": 0.2675,
+      "step": 12352
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3345566769538745e-08,
+      "loss": 0.271,
+      "step": 12353
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.16773709233464e-08,
+      "loss": 0.2465,
+      "step": 12354
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0009175077154059e-08,
+      "loss": 0.2631,
+      "step": 12355
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.340979230961714e-09,
+      "loss": 0.2676,
+      "step": 12356
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.672783384769372e-09,
+      "loss": 0.2729,
+      "step": 12357
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.0045875385770295e-09,
+      "loss": 0.2847,
+      "step": 12358
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.336391692384686e-09,
+      "loss": 0.2744,
+      "step": 12359
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.668195846192343e-09,
+      "loss": 0.2714,
+      "step": 12360
+    },
+    {
+      "epoch": 3.0,
+      "step": 12360,
+      "total_flos": 1618222810046464.0,
+      "train_loss": 0.3937891273051018,
+      "train_runtime": 153709.928,
+      "train_samples_per_second": 10.294,
+      "train_steps_per_second": 0.08
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 12360,
+  "num_train_epochs": 3,
+  "save_steps": 100000,
+  "total_flos": 1618222810046464.0,
+  "trial_name": null,
+  "trial_params": null
+}