diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,27085 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.912621359223301,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999870550161813e-05,
+      "loss": 0.0711,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999741100323626e-05,
+      "loss": 0.0622,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999611650485437e-05,
+      "loss": 0.1359,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.99948220064725e-05,
+      "loss": 0.0756,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999352750809062e-05,
+      "loss": 0.1416,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999223300970875e-05,
+      "loss": 0.0614,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.999093851132686e-05,
+      "loss": 0.08,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998964401294499e-05,
+      "loss": 0.111,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99883495145631e-05,
+      "loss": 0.0603,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998705501618123e-05,
+      "loss": 0.0863,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998576051779936e-05,
+      "loss": 0.1034,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998446601941748e-05,
+      "loss": 0.0543,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99831715210356e-05,
+      "loss": 0.0955,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998187702265372e-05,
+      "loss": 0.0853,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998058252427185e-05,
+      "loss": 0.1786,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997928802588998e-05,
+      "loss": 0.1019,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99779935275081e-05,
+      "loss": 0.1239,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997669902912622e-05,
+      "loss": 0.1165,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997540453074434e-05,
+      "loss": 0.1117,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997411003236247e-05,
+      "loss": 0.0829,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99728155339806e-05,
+      "loss": 0.1164,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997152103559871e-05,
+      "loss": 0.0988,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.997022653721683e-05,
+      "loss": 0.0868,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996893203883496e-05,
+      "loss": 0.0842,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996763754045307e-05,
+      "loss": 0.1189,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99663430420712e-05,
+      "loss": 0.0691,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996504854368933e-05,
+      "loss": 0.0606,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996375404530744e-05,
+      "loss": 0.0533,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996245954692557e-05,
+      "loss": 0.0774,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99611650485437e-05,
+      "loss": 0.087,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995987055016182e-05,
+      "loss": 0.0515,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995857605177993e-05,
+      "loss": 0.0858,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995728155339806e-05,
+      "loss": 0.047,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995598705501619e-05,
+      "loss": 0.0885,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995469255663431e-05,
+      "loss": 0.072,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995339805825244e-05,
+      "loss": 0.0993,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995210355987055e-05,
+      "loss": 0.0608,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995080906148868e-05,
+      "loss": 0.0751,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99495145631068e-05,
+      "loss": 0.0574,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994822006472493e-05,
+      "loss": 0.0632,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994692556634305e-05,
+      "loss": 0.102,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994563106796116e-05,
+      "loss": 0.1176,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994433656957929e-05,
+      "loss": 0.1145,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994304207119741e-05,
+      "loss": 0.067,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994174757281554e-05,
+      "loss": 0.125,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994045307443367e-05,
+      "loss": 0.1159,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993915857605178e-05,
+      "loss": 0.0552,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99378640776699e-05,
+      "loss": 0.0969,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993656957928803e-05,
+      "loss": 0.0754,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993527508090616e-05,
+      "loss": 0.0677,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993398058252428e-05,
+      "loss": 0.0569,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99326860841424e-05,
+      "loss": 0.1022,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993139158576052e-05,
+      "loss": 0.1425,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993009708737865e-05,
+      "loss": 0.1391,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992880258899678e-05,
+      "loss": 0.1179,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992750809061489e-05,
+      "loss": 0.1089,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992621359223302e-05,
+      "loss": 0.1259,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992491909385113e-05,
+      "loss": 0.074,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992362459546926e-05,
+      "loss": 0.1047,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992233009708738e-05,
+      "loss": 0.0918,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.992103559870551e-05,
+      "loss": 0.0648,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991974110032362e-05,
+      "loss": 0.0714,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991844660194175e-05,
+      "loss": 0.114,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991715210355988e-05,
+      "loss": 0.105,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9915857605178e-05,
+      "loss": 0.1094,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991456310679612e-05,
+      "loss": 0.0608,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991326860841424e-05,
+      "loss": 0.0802,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.991197411003237e-05,
+      "loss": 0.0618,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99106796116505e-05,
+      "loss": 0.0715,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990938511326862e-05,
+      "loss": 0.085,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990809061488673e-05,
+      "loss": 0.088,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990679611650486e-05,
+      "loss": 0.0702,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990550161812299e-05,
+      "loss": 0.0748,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99042071197411e-05,
+      "loss": 0.1042,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990291262135923e-05,
+      "loss": 0.0809,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990161812297734e-05,
+      "loss": 0.059,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990032362459547e-05,
+      "loss": 0.1912,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98990291262136e-05,
+      "loss": 0.0707,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989773462783172e-05,
+      "loss": 0.0582,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989644012944985e-05,
+      "loss": 0.1362,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989514563106796e-05,
+      "loss": 0.079,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989385113268609e-05,
+      "loss": 0.0856,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989255663430421e-05,
+      "loss": 0.0571,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989126213592234e-05,
+      "loss": 0.0899,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988996763754047e-05,
+      "loss": 0.0506,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988867313915858e-05,
+      "loss": 0.0877,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98873786407767e-05,
+      "loss": 0.0767,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988608414239483e-05,
+      "loss": 0.0913,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988478964401296e-05,
+      "loss": 0.0692,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988349514563107e-05,
+      "loss": 0.0472,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988220064724919e-05,
+      "loss": 0.2374,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988090614886731e-05,
+      "loss": 0.0683,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987961165048544e-05,
+      "loss": 0.1562,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987831715210357e-05,
+      "loss": 0.0719,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987702265372169e-05,
+      "loss": 0.1143,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98757281553398e-05,
+      "loss": 0.0877,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987443365695793e-05,
+      "loss": 0.0735,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987313915857606e-05,
+      "loss": 0.0656,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987184466019418e-05,
+      "loss": 0.0846,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98705501618123e-05,
+      "loss": 0.0939,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986925566343042e-05,
+      "loss": 0.05,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986796116504855e-05,
+      "loss": 0.1024,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986666666666668e-05,
+      "loss": 0.0549,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98653721682848e-05,
+      "loss": 0.0714,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986407766990292e-05,
+      "loss": 0.1005,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986278317152103e-05,
+      "loss": 0.0751,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986148867313916e-05,
+      "loss": 0.0838,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986019417475728e-05,
+      "loss": 0.1063,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985889967637541e-05,
+      "loss": 0.1149,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985760517799354e-05,
+      "loss": 0.0811,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985631067961165e-05,
+      "loss": 0.084,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985501618122978e-05,
+      "loss": 0.0972,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98537216828479e-05,
+      "loss": 0.0485,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985242718446603e-05,
+      "loss": 0.0573,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.985113268608414e-05,
+      "loss": 0.0911,
+      "step": 115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984983818770227e-05,
+      "loss": 0.0685,
+      "step": 116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98485436893204e-05,
+      "loss": 0.0636,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984724919093852e-05,
+      "loss": 0.0599,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984595469255665e-05,
+      "loss": 0.0595,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984466019417476e-05,
+      "loss": 0.1437,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984336569579289e-05,
+      "loss": 0.1314,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984207119741101e-05,
+      "loss": 0.1637,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984077669902913e-05,
+      "loss": 0.1373,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983948220064725e-05,
+      "loss": 0.0736,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983818770226537e-05,
+      "loss": 0.1027,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98368932038835e-05,
+      "loss": 0.0769,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983559870550162e-05,
+      "loss": 0.0798,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983430420711975e-05,
+      "loss": 0.0758,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983300970873787e-05,
+      "loss": 0.0984,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983171521035599e-05,
+      "loss": 0.1013,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983042071197411e-05,
+      "loss": 0.0859,
+      "step": 131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982912621359224e-05,
+      "loss": 0.0608,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982783171521037e-05,
+      "loss": 0.0522,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982653721682849e-05,
+      "loss": 0.0674,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98252427184466e-05,
+      "loss": 0.058,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982394822006473e-05,
+      "loss": 0.0656,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982265372168286e-05,
+      "loss": 0.1346,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982135922330099e-05,
+      "loss": 0.1,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98200647249191e-05,
+      "loss": 0.133,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981877022653721e-05,
+      "loss": 0.0841,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981747572815534e-05,
+      "loss": 0.0756,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981618122977347e-05,
+      "loss": 0.1203,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981488673139159e-05,
+      "loss": 0.0844,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981359223300972e-05,
+      "loss": 0.058,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981229773462783e-05,
+      "loss": 0.1089,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981100323624596e-05,
+      "loss": 0.0475,
+      "step": 146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980970873786408e-05,
+      "loss": 0.0629,
+      "step": 147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980841423948221e-05,
+      "loss": 0.0754,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980711974110032e-05,
+      "loss": 0.3145,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980582524271845e-05,
+      "loss": 0.0855,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980453074433658e-05,
+      "loss": 0.1195,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.98032362459547e-05,
+      "loss": 0.072,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980194174757283e-05,
+      "loss": 0.086,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980064724919094e-05,
+      "loss": 0.0933,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979935275080906e-05,
+      "loss": 0.1039,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979805825242718e-05,
+      "loss": 0.0369,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979676375404531e-05,
+      "loss": 0.103,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979546925566344e-05,
+      "loss": 0.0513,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979417475728155e-05,
+      "loss": 0.0611,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979288025889968e-05,
+      "loss": 0.0788,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.97915857605178e-05,
+      "loss": 0.0637,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.979029126213593e-05,
+      "loss": 0.0594,
+      "step": 162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978899676375406e-05,
+      "loss": 0.1842,
+      "step": 163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978770226537217e-05,
+      "loss": 0.1363,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.97864077669903e-05,
+      "loss": 0.0818,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978511326860842e-05,
+      "loss": 0.0997,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978381877022655e-05,
+      "loss": 0.0404,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978252427184467e-05,
+      "loss": 0.1104,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.978122977346279e-05,
+      "loss": 0.0791,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977993527508091e-05,
+      "loss": 0.0427,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977864077669904e-05,
+      "loss": 0.07,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977734627831715e-05,
+      "loss": 0.0541,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977605177993528e-05,
+      "loss": 0.0831,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.97747572815534e-05,
+      "loss": 0.0852,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977346278317152e-05,
+      "loss": 0.0811,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977216828478965e-05,
+      "loss": 0.1247,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.977087378640777e-05,
+      "loss": 0.1735,
+      "step": 177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.97695792880259e-05,
+      "loss": 0.0662,
+      "step": 178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976828478964401e-05,
+      "loss": 0.0728,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976699029126214e-05,
+      "loss": 0.0747,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976569579288027e-05,
+      "loss": 0.0887,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976440129449839e-05,
+      "loss": 0.0777,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.97631067961165e-05,
+      "loss": 0.0549,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976181229773463e-05,
+      "loss": 0.1245,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.976051779935276e-05,
+      "loss": 0.101,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975922330097089e-05,
+      "loss": 0.0995,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975792880258901e-05,
+      "loss": 0.1216,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975663430420713e-05,
+      "loss": 0.2318,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975533980582524e-05,
+      "loss": 0.0423,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975404530744336e-05,
+      "loss": 0.0783,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975275080906149e-05,
+      "loss": 0.1257,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975145631067962e-05,
+      "loss": 0.0868,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.975016181229773e-05,
+      "loss": 0.1087,
+      "step": 193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974886731391586e-05,
+      "loss": 0.0591,
+      "step": 194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974757281553398e-05,
+      "loss": 0.0635,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974627831715211e-05,
+      "loss": 0.0977,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974498381877024e-05,
+      "loss": 0.0507,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974368932038835e-05,
+      "loss": 0.1566,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974239482200648e-05,
+      "loss": 0.0716,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97411003236246e-05,
+      "loss": 0.0785,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973980582524273e-05,
+      "loss": 0.0753,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973851132686086e-05,
+      "loss": 0.1081,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973721682847897e-05,
+      "loss": 0.0666,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973592233009708e-05,
+      "loss": 0.05,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973462783171521e-05,
+      "loss": 0.0748,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973333333333334e-05,
+      "loss": 0.1817,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973203883495146e-05,
+      "loss": 0.0667,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973074433656958e-05,
+      "loss": 0.0903,
+      "step": 208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97294498381877e-05,
+      "loss": 0.0894,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972815533980583e-05,
+      "loss": 0.0647,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972686084142396e-05,
+      "loss": 0.072,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972556634304208e-05,
+      "loss": 0.069,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97242718446602e-05,
+      "loss": 0.048,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972297734627832e-05,
+      "loss": 0.1615,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972168284789645e-05,
+      "loss": 0.0658,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.972038834951457e-05,
+      "loss": 0.0899,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971909385113269e-05,
+      "loss": 0.1954,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971779935275081e-05,
+      "loss": 0.0556,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971650485436894e-05,
+      "loss": 0.0867,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971521035598705e-05,
+      "loss": 0.0492,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971391585760518e-05,
+      "loss": 0.0674,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971262135922331e-05,
+      "loss": 0.0578,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971132686084142e-05,
+      "loss": 0.0677,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971003236245955e-05,
+      "loss": 0.0915,
+      "step": 224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970873786407767e-05,
+      "loss": 0.0481,
+      "step": 225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.97074433656958e-05,
+      "loss": 0.1228,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970614886731391e-05,
+      "loss": 0.0681,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970485436893204e-05,
+      "loss": 0.0623,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970355987055017e-05,
+      "loss": 0.1006,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970226537216829e-05,
+      "loss": 0.054,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.970097087378642e-05,
+      "loss": 0.1187,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969967637540453e-05,
+      "loss": 0.0645,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969838187702266e-05,
+      "loss": 0.0618,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969708737864079e-05,
+      "loss": 0.0758,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969579288025891e-05,
+      "loss": 0.0604,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969449838187704e-05,
+      "loss": 0.0997,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969320388349515e-05,
+      "loss": 0.0677,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969190938511326e-05,
+      "loss": 0.0691,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.969061488673139e-05,
+      "loss": 0.0559,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968932038834952e-05,
+      "loss": 0.093,
+      "step": 240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968802588996764e-05,
+      "loss": 0.0573,
+      "step": 241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968673139158576e-05,
+      "loss": 0.0484,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968543689320388e-05,
+      "loss": 0.0543,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968414239482201e-05,
+      "loss": 0.0548,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968284789644014e-05,
+      "loss": 0.1357,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968155339805826e-05,
+      "loss": 0.1181,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.968025889967638e-05,
+      "loss": 0.1341,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.96789644012945e-05,
+      "loss": 0.0853,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.967766990291263e-05,
+      "loss": 0.1193,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.967637540453076e-05,
+      "loss": 0.0727,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.967508090614887e-05,
+      "loss": 0.0758,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.9673786407767e-05,
+      "loss": 0.0786,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.967249190938511e-05,
+      "loss": 0.1106,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.967119741100324e-05,
+      "loss": 0.1935,
+      "step": 254
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966990291262136e-05,
+      "loss": 0.0886,
+      "step": 255
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966860841423949e-05,
+      "loss": 0.0641,
+      "step": 256
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.96673139158576e-05,
+      "loss": 0.1428,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966601941747573e-05,
+      "loss": 0.1002,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966472491909386e-05,
+      "loss": 0.0872,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966343042071198e-05,
+      "loss": 0.0747,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966213592233011e-05,
+      "loss": 0.1514,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.966084142394822e-05,
+      "loss": 0.1076,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965954692556635e-05,
+      "loss": 0.088,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965825242718447e-05,
+      "loss": 0.0948,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.96569579288026e-05,
+      "loss": 0.2122,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965566343042071e-05,
+      "loss": 0.0704,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965436893203884e-05,
+      "loss": 0.0877,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965307443365697e-05,
+      "loss": 0.0992,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965177993527508e-05,
+      "loss": 0.0511,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.965048543689321e-05,
+      "loss": 0.0955,
+      "step": 270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964919093851133e-05,
+      "loss": 0.0918,
+      "step": 271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964789644012945e-05,
+      "loss": 0.0775,
+      "step": 272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964660194174757e-05,
+      "loss": 0.1019,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.96453074433657e-05,
+      "loss": 0.1581,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964401294498383e-05,
+      "loss": 0.0692,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964271844660194e-05,
+      "loss": 0.1033,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964142394822007e-05,
+      "loss": 0.0873,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.964012944983819e-05,
+      "loss": 0.0849,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963883495145632e-05,
+      "loss": 0.0843,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963754045307445e-05,
+      "loss": 0.0692,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963624595469256e-05,
+      "loss": 0.0715,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963495145631069e-05,
+      "loss": 0.0894,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963365695792881e-05,
+      "loss": 0.1002,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963236245954694e-05,
+      "loss": 0.0768,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.963106796116507e-05,
+      "loss": 0.0707,
+      "step": 285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962977346278316e-05,
+      "loss": 0.0997,
+      "step": 286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962847896440129e-05,
+      "loss": 0.0842,
+      "step": 287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962718446601942e-05,
+      "loss": 0.0798,
+      "step": 288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962588996763754e-05,
+      "loss": 0.119,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962459546925567e-05,
+      "loss": 0.0946,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962330097087378e-05,
+      "loss": 0.0756,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962200647249191e-05,
+      "loss": 0.0424,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.962071197411004e-05,
+      "loss": 0.0778,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961941747572816e-05,
+      "loss": 0.0702,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961812297734629e-05,
+      "loss": 0.0865,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.96168284789644e-05,
+      "loss": 0.0568,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961553398058253e-05,
+      "loss": 0.0995,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961423948220066e-05,
+      "loss": 0.1454,
+      "step": 298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961294498381878e-05,
+      "loss": 0.0923,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.96116504854369e-05,
+      "loss": 0.1496,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.961035598705502e-05,
+      "loss": 0.0584,
+      "step": 301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960906148867314e-05,
+      "loss": 0.071,
+      "step": 302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960776699029126e-05,
+      "loss": 0.0988,
+      "step": 303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960647249190939e-05,
+      "loss": 0.111,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960517799352752e-05,
+      "loss": 0.0803,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960388349514563e-05,
+      "loss": 0.1252,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960258899676376e-05,
+      "loss": 0.0589,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960129449838188e-05,
+      "loss": 0.0758,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 0.0514,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959870550161812e-05,
+      "loss": 0.108,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959741100323625e-05,
+      "loss": 0.0549,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959611650485437e-05,
+      "loss": 0.0798,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.95948220064725e-05,
+      "loss": 0.0936,
+      "step": 313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959352750809063e-05,
+      "loss": 0.0846,
+      "step": 314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959223300970874e-05,
+      "loss": 0.1208,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.959093851132687e-05,
+      "loss": 0.1013,
+      "step": 316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.9589644012945e-05,
+      "loss": 0.0828,
+      "step": 317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958834951456311e-05,
+      "loss": 0.068,
+      "step": 318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958705501618123e-05,
+      "loss": 0.0401,
+      "step": 319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958576051779935e-05,
+      "loss": 0.0558,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958446601941747e-05,
+      "loss": 0.0425,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.95831715210356e-05,
+      "loss": 0.0761,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958187702265373e-05,
+      "loss": 0.0502,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.958058252427185e-05,
+      "loss": 0.0495,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957928802588997e-05,
+      "loss": 0.0948,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957799352750809e-05,
+      "loss": 0.0355,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957669902912622e-05,
+      "loss": 0.0599,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957540453074435e-05,
+      "loss": 0.0645,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957411003236247e-05,
+      "loss": 0.0735,
+      "step": 329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957281553398059e-05,
+      "loss": 0.1488,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957152103559871e-05,
+      "loss": 0.0638,
+      "step": 331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.957022653721684e-05,
+      "loss": 0.0495,
+      "step": 332
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956893203883497e-05,
+      "loss": 0.0689,
+      "step": 333
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956763754045308e-05,
+      "loss": 0.0599,
+      "step": 334
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956634304207119e-05,
+      "loss": 0.058,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956504854368932e-05,
+      "loss": 0.1031,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956375404530744e-05,
+      "loss": 0.0998,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.956245954692557e-05,
+      "loss": 0.1063,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.95611650485437e-05,
+      "loss": 0.0693,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955987055016181e-05,
+      "loss": 0.0704,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955857605177994e-05,
+      "loss": 0.0776,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955728155339806e-05,
+      "loss": 0.0657,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955598705501619e-05,
+      "loss": 0.1452,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.95546925566343e-05,
+      "loss": 0.0563,
+      "step": 344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955339805825243e-05,
+      "loss": 0.0761,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955210355987056e-05,
+      "loss": 0.0466,
+      "step": 346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.955080906148868e-05,
+      "loss": 0.0878,
+      "step": 347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954951456310681e-05,
+      "loss": 0.1105,
+      "step": 348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954822006472492e-05,
+      "loss": 0.0575,
+      "step": 349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954692556634305e-05,
+      "loss": 0.234,
+      "step": 350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954563106796116e-05,
+      "loss": 0.2075,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954433656957929e-05,
+      "loss": 0.0653,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954304207119742e-05,
+      "loss": 0.0768,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954174757281554e-05,
+      "loss": 0.0771,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.954045307443366e-05,
+      "loss": 0.0687,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953915857605178e-05,
+      "loss": 0.0736,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953786407766991e-05,
+      "loss": 0.0927,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953656957928804e-05,
+      "loss": 0.0747,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953527508090615e-05,
+      "loss": 0.158,
+      "step": 359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953398058252427e-05,
+      "loss": 0.0666,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.95326860841424e-05,
+      "loss": 0.039,
+      "step": 361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953139158576053e-05,
+      "loss": 0.0853,
+      "step": 362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.953009708737865e-05,
+      "loss": 0.1046,
+      "step": 363
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.952880258899677e-05,
+      "loss": 0.1247,
+      "step": 364
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.95275080906149e-05,
+      "loss": 0.0951,
+      "step": 365
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.952621359223302e-05,
+      "loss": 0.0662,
+      "step": 366
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.952491909385113e-05,
+      "loss": 0.0492,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.952362459546926e-05,
+      "loss": 0.0506,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.952233009708737e-05,
+      "loss": 0.0514,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.95210355987055e-05,
+      "loss": 0.0957,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951974110032363e-05,
+      "loss": 0.0872,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951844660194175e-05,
+      "loss": 0.0631,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951715210355988e-05,
+      "loss": 0.0502,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951585760517799e-05,
+      "loss": 0.0611,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951456310679612e-05,
+      "loss": 0.153,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951326860841425e-05,
+      "loss": 0.0961,
+      "step": 376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.951197411003237e-05,
+      "loss": 0.0848,
+      "step": 377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.95106796116505e-05,
+      "loss": 0.0597,
+      "step": 378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950938511326861e-05,
+      "loss": 0.057,
+      "step": 379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950809061488674e-05,
+      "loss": 0.0642,
+      "step": 380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950679611650487e-05,
+      "loss": 0.0562,
+      "step": 381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950550161812299e-05,
+      "loss": 0.1342,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.95042071197411e-05,
+      "loss": 0.0747,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950291262135922e-05,
+      "loss": 0.0799,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950161812297734e-05,
+      "loss": 0.0803,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.950032362459547e-05,
+      "loss": 0.0552,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.94990291262136e-05,
+      "loss": 0.1246,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949773462783172e-05,
+      "loss": 0.0724,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949644012944984e-05,
+      "loss": 0.0856,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949514563106796e-05,
+      "loss": 0.0708,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949385113268609e-05,
+      "loss": 0.0557,
+      "step": 391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949255663430422e-05,
+      "loss": 0.0878,
+      "step": 392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.949126213592233e-05,
+      "loss": 0.0902,
+      "step": 393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948996763754046e-05,
+      "loss": 0.0605,
+      "step": 394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948867313915858e-05,
+      "loss": 0.1036,
+      "step": 395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948737864077671e-05,
+      "loss": 0.0678,
+      "step": 396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948608414239484e-05,
+      "loss": 0.0621,
+      "step": 397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948478964401295e-05,
+      "loss": 0.069,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948349514563108e-05,
+      "loss": 0.0981,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948220064724919e-05,
+      "loss": 0.0695,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.948090614886732e-05,
+      "loss": 0.0771,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947961165048544e-05,
+      "loss": 0.0712,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947831715210356e-05,
+      "loss": 0.0507,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947702265372168e-05,
+      "loss": 0.0666,
+      "step": 404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947572815533981e-05,
+      "loss": 0.0619,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947443365695794e-05,
+      "loss": 0.0698,
+      "step": 406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947313915857606e-05,
+      "loss": 0.058,
+      "step": 407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.947184466019417e-05,
+      "loss": 0.0717,
+      "step": 408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.94705501618123e-05,
+      "loss": 0.0696,
+      "step": 409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946925566343043e-05,
+      "loss": 0.0661,
+      "step": 410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946796116504855e-05,
+      "loss": 0.1097,
+      "step": 411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946666666666668e-05,
+      "loss": 0.1104,
+      "step": 412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.94653721682848e-05,
+      "loss": 0.0877,
+      "step": 413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946407766990292e-05,
+      "loss": 0.0603,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946278317152105e-05,
+      "loss": 0.0585,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946148867313916e-05,
+      "loss": 0.0815,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.946019417475729e-05,
+      "loss": 0.0589,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.94588996763754e-05,
+      "loss": 0.0831,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.945760517799353e-05,
+      "loss": 0.0805,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.945631067961165e-05,
+      "loss": 0.0716,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.945501618122978e-05,
+      "loss": 0.0449,
+      "step": 421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.94537216828479e-05,
+      "loss": 0.0853,
+      "step": 422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.945242718446602e-05,
+      "loss": 0.0687,
+      "step": 423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.945113268608415e-05,
+      "loss": 0.0984,
+      "step": 424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944983818770227e-05,
+      "loss": 0.0669,
+      "step": 425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.94485436893204e-05,
+      "loss": 0.158,
+      "step": 426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944724919093851e-05,
+      "loss": 0.0853,
+      "step": 427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944595469255664e-05,
+      "loss": 0.0501,
+      "step": 428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944466019417477e-05,
+      "loss": 0.0644,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944336569579289e-05,
+      "loss": 0.0445,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944207119741102e-05,
+      "loss": 0.0535,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.944077669902913e-05,
+      "loss": 0.1226,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943948220064724e-05,
+      "loss": 0.0932,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943818770226537e-05,
+      "loss": 0.0489,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.94368932038835e-05,
+      "loss": 0.0575,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943559870550162e-05,
+      "loss": 0.0776,
+      "step": 436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943430420711974e-05,
+      "loss": 0.0682,
+      "step": 437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943300970873786e-05,
+      "loss": 0.0888,
+      "step": 438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943171521035599e-05,
+      "loss": 0.1788,
+      "step": 439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.943042071197412e-05,
+      "loss": 0.0935,
+      "step": 440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942912621359224e-05,
+      "loss": 0.0857,
+      "step": 441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942783171521036e-05,
+      "loss": 0.046,
+      "step": 442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942653721682848e-05,
+      "loss": 0.1296,
+      "step": 443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942524271844661e-05,
+      "loss": 0.0662,
+      "step": 444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942394822006474e-05,
+      "loss": 0.0686,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942265372168286e-05,
+      "loss": 0.0376,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942135922330098e-05,
+      "loss": 0.0767,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.942006472491909e-05,
+      "loss": 0.0579,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941877022653722e-05,
+      "loss": 0.1007,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941747572815534e-05,
+      "loss": 0.0555,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941618122977347e-05,
+      "loss": 0.1031,
+      "step": 451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941488673139158e-05,
+      "loss": 0.0892,
+      "step": 452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941359223300971e-05,
+      "loss": 0.0426,
+      "step": 453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941229773462784e-05,
+      "loss": 0.0909,
+      "step": 454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.941100323624596e-05,
+      "loss": 0.1092,
+      "step": 455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940970873786409e-05,
+      "loss": 0.0761,
+      "step": 456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.94084142394822e-05,
+      "loss": 0.1079,
+      "step": 457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940711974110033e-05,
+      "loss": 0.0948,
+      "step": 458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940582524271845e-05,
+      "loss": 0.0784,
+      "step": 459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940453074433658e-05,
+      "loss": 0.0776,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.94032362459547e-05,
+      "loss": 0.0818,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940194174757282e-05,
+      "loss": 0.0645,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.940064724919095e-05,
+      "loss": 0.0428,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939935275080907e-05,
+      "loss": 0.0509,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939805825242719e-05,
+      "loss": 0.0678,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939676375404531e-05,
+      "loss": 0.0568,
+      "step": 466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939546925566343e-05,
+      "loss": 0.0816,
+      "step": 467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939417475728155e-05,
+      "loss": 0.0706,
+      "step": 468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939288025889968e-05,
+      "loss": 0.1132,
+      "step": 469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.93915857605178e-05,
+      "loss": 0.0712,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.939029126213593e-05,
+      "loss": 0.0514,
+      "step": 471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938899676375405e-05,
+      "loss": 0.0839,
+      "step": 472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938770226537217e-05,
+      "loss": 0.1413,
+      "step": 473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.93864077669903e-05,
+      "loss": 0.0546,
+      "step": 474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938511326860843e-05,
+      "loss": 0.0365,
+      "step": 475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938381877022654e-05,
+      "loss": 0.0613,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938252427184467e-05,
+      "loss": 0.065,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.938122977346279e-05,
+      "loss": 0.0771,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937993527508092e-05,
+      "loss": 0.0618,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937864077669904e-05,
+      "loss": 0.0721,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937734627831716e-05,
+      "loss": 0.0779,
+      "step": 481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937605177993527e-05,
+      "loss": 0.0738,
+      "step": 482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.93747572815534e-05,
+      "loss": 0.123,
+      "step": 483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937346278317152e-05,
+      "loss": 0.0491,
+      "step": 484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937216828478965e-05,
+      "loss": 0.0863,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.937087378640776e-05,
+      "loss": 0.0988,
+      "step": 486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936957928802589e-05,
+      "loss": 0.0763,
+      "step": 487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936828478964402e-05,
+      "loss": 0.0755,
+      "step": 488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936699029126214e-05,
+      "loss": 0.0621,
+      "step": 489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936569579288027e-05,
+      "loss": 0.0832,
+      "step": 490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936440129449838e-05,
+      "loss": 0.0643,
+      "step": 491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936310679611651e-05,
+      "loss": 0.1803,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936181229773464e-05,
+      "loss": 0.0318,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.936051779935276e-05,
+      "loss": 0.135,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935922330097089e-05,
+      "loss": 0.078,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.9357928802589e-05,
+      "loss": 0.0619,
+      "step": 496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935663430420712e-05,
+      "loss": 0.0684,
+      "step": 497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935533980582524e-05,
+      "loss": 0.0749,
+      "step": 498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935404530744337e-05,
+      "loss": 0.0617,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.93527508090615e-05,
+      "loss": 0.0442,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935145631067961e-05,
+      "loss": 0.0621,
+      "step": 501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.935016181229774e-05,
+      "loss": 0.0497,
+      "step": 502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934886731391586e-05,
+      "loss": 0.0915,
+      "step": 503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934757281553399e-05,
+      "loss": 0.0429,
+      "step": 504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934627831715211e-05,
+      "loss": 0.0432,
+      "step": 505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934498381877023e-05,
+      "loss": 0.0955,
+      "step": 506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934368932038835e-05,
+      "loss": 0.056,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934239482200648e-05,
+      "loss": 0.0725,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.934110032362461e-05,
+      "loss": 0.0617,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933980582524272e-05,
+      "loss": 0.073,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933851132686085e-05,
+      "loss": 0.0823,
+      "step": 511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933721682847897e-05,
+      "loss": 0.0524,
+      "step": 512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.93359223300971e-05,
+      "loss": 0.082,
+      "step": 513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933462783171521e-05,
+      "loss": 0.1257,
+      "step": 514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933333333333334e-05,
+      "loss": 0.0965,
+      "step": 515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933203883495145e-05,
+      "loss": 0.0666,
+      "step": 516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.933074433656958e-05,
+      "loss": 0.0442,
+      "step": 517
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.93294498381877e-05,
+      "loss": 0.0921,
+      "step": 518
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932815533980583e-05,
+      "loss": 0.0527,
+      "step": 519
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932686084142395e-05,
+      "loss": 0.0568,
+      "step": 520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932556634304207e-05,
+      "loss": 0.05,
+      "step": 521
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.93242718446602e-05,
+      "loss": 0.072,
+      "step": 522
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932297734627833e-05,
+      "loss": 0.0549,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932168284789645e-05,
+      "loss": 0.0903,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.932038834951457e-05,
+      "loss": 0.0738,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931909385113269e-05,
+      "loss": 0.0506,
+      "step": 526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931779935275082e-05,
+      "loss": 0.0657,
+      "step": 527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931650485436894e-05,
+      "loss": 0.0669,
+      "step": 528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931521035598707e-05,
+      "loss": 0.0615,
+      "step": 529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931391585760517e-05,
+      "loss": 0.036,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.93126213592233e-05,
+      "loss": 0.099,
+      "step": 531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931132686084142e-05,
+      "loss": 0.0802,
+      "step": 532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.931003236245955e-05,
+      "loss": 0.0648,
+      "step": 533
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930873786407768e-05,
+      "loss": 0.0573,
+      "step": 534
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930744336569579e-05,
+      "loss": 0.0973,
+      "step": 535
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930614886731392e-05,
+      "loss": 0.0722,
+      "step": 536
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930485436893204e-05,
+      "loss": 0.0967,
+      "step": 537
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930355987055017e-05,
+      "loss": 0.0739,
+      "step": 538
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.93022653721683e-05,
+      "loss": 0.0839,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.930097087378641e-05,
+      "loss": 0.0497,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929967637540454e-05,
+      "loss": 0.0707,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929838187702266e-05,
+      "loss": 0.0463,
+      "step": 542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929708737864079e-05,
+      "loss": 0.066,
+      "step": 543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.92957928802589e-05,
+      "loss": 0.0727,
+      "step": 544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929449838187703e-05,
+      "loss": 0.0481,
+      "step": 545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929320388349514e-05,
+      "loss": 0.0602,
+      "step": 546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.929190938511327e-05,
+      "loss": 0.0394,
+      "step": 547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.92906148867314e-05,
+      "loss": 0.0738,
+      "step": 548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928932038834952e-05,
+      "loss": 0.0764,
+      "step": 549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928802588996763e-05,
+      "loss": 0.0896,
+      "step": 550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928673139158576e-05,
+      "loss": 0.0543,
+      "step": 551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928543689320389e-05,
+      "loss": 0.0677,
+      "step": 552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928414239482201e-05,
+      "loss": 0.0862,
+      "step": 553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928284789644013e-05,
+      "loss": 0.0846,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928155339805825e-05,
+      "loss": 0.0666,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.928025889967638e-05,
+      "loss": 0.068,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927896440129451e-05,
+      "loss": 0.0719,
+      "step": 557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927766990291263e-05,
+      "loss": 0.0953,
+      "step": 558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927637540453075e-05,
+      "loss": 0.0844,
+      "step": 559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927508090614887e-05,
+      "loss": 0.0641,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.9273786407767e-05,
+      "loss": 0.097,
+      "step": 561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927249190938513e-05,
+      "loss": 0.0834,
+      "step": 562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.927119741100324e-05,
+      "loss": 0.0738,
+      "step": 563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926990291262137e-05,
+      "loss": 0.0859,
+      "step": 564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926860841423948e-05,
+      "loss": 0.0468,
+      "step": 565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.92673139158576e-05,
+      "loss": 0.0589,
+      "step": 566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926601941747573e-05,
+      "loss": 0.0679,
+      "step": 567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926472491909386e-05,
+      "loss": 0.0484,
+      "step": 568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926343042071197e-05,
+      "loss": 0.055,
+      "step": 569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.92621359223301e-05,
+      "loss": 0.0386,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.926084142394823e-05,
+      "loss": 0.067,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925954692556635e-05,
+      "loss": 0.0689,
+      "step": 572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925825242718448e-05,
+      "loss": 0.0742,
+      "step": 573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925695792880259e-05,
+      "loss": 0.0493,
+      "step": 574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925566343042072e-05,
+      "loss": 0.0488,
+      "step": 575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925436893203884e-05,
+      "loss": 0.0327,
+      "step": 576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925307443365697e-05,
+      "loss": 0.0514,
+      "step": 577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.925177993527508e-05,
+      "loss": 0.0645,
+      "step": 578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.92504854368932e-05,
+      "loss": 0.0776,
+      "step": 579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924919093851132e-05,
+      "loss": 0.052,
+      "step": 580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924789644012945e-05,
+      "loss": 0.0674,
+      "step": 581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924660194174758e-05,
+      "loss": 0.1018,
+      "step": 582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.92453074433657e-05,
+      "loss": 0.047,
+      "step": 583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924401294498382e-05,
+      "loss": 0.0644,
+      "step": 584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924271844660194e-05,
+      "loss": 0.0444,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.924142394822007e-05,
+      "loss": 0.0438,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.92401294498382e-05,
+      "loss": 0.051,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923883495145631e-05,
+      "loss": 0.0725,
+      "step": 588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923754045307444e-05,
+      "loss": 0.1122,
+      "step": 589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923624595469256e-05,
+      "loss": 0.0832,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923495145631069e-05,
+      "loss": 0.0576,
+      "step": 591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923365695792882e-05,
+      "loss": 0.1084,
+      "step": 592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923236245954693e-05,
+      "loss": 0.0457,
+      "step": 593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.923106796116506e-05,
+      "loss": 0.0837,
+      "step": 594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922977346278317e-05,
+      "loss": 0.0575,
+      "step": 595
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.92284789644013e-05,
+      "loss": 0.079,
+      "step": 596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922718446601942e-05,
+      "loss": 0.0731,
+      "step": 597
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922588996763755e-05,
+      "loss": 0.0591,
+      "step": 598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922459546925566e-05,
+      "loss": 0.0609,
+      "step": 599
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922330097087379e-05,
+      "loss": 0.0436,
+      "step": 600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922200647249191e-05,
+      "loss": 0.0624,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.922071197411004e-05,
+      "loss": 0.0715,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921941747572815e-05,
+      "loss": 0.047,
+      "step": 603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921812297734628e-05,
+      "loss": 0.1064,
+      "step": 604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921682847896441e-05,
+      "loss": 0.0769,
+      "step": 605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921553398058253e-05,
+      "loss": 0.0399,
+      "step": 606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921423948220066e-05,
+      "loss": 0.059,
+      "step": 607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921294498381877e-05,
+      "loss": 0.0747,
+      "step": 608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.92116504854369e-05,
+      "loss": 0.081,
+      "step": 609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.921035598705503e-05,
+      "loss": 0.0601,
+      "step": 610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920906148867314e-05,
+      "loss": 0.0888,
+      "step": 611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920776699029127e-05,
+      "loss": 0.0468,
+      "step": 612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920647249190938e-05,
+      "loss": 0.0513,
+      "step": 613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.92051779935275e-05,
+      "loss": 0.0439,
+      "step": 614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920388349514563e-05,
+      "loss": 0.0855,
+      "step": 615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920258899676376e-05,
+      "loss": 0.0519,
+      "step": 616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.920129449838189e-05,
+      "loss": 0.1506,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.92e-05,
+      "loss": 0.0522,
+      "step": 618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919870550161813e-05,
+      "loss": 0.0594,
+      "step": 619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919741100323625e-05,
+      "loss": 0.056,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919611650485438e-05,
+      "loss": 0.0798,
+      "step": 621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.91948220064725e-05,
+      "loss": 0.1011,
+      "step": 622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919352750809062e-05,
+      "loss": 0.0993,
+      "step": 623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919223300970874e-05,
+      "loss": 0.0837,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.919093851132687e-05,
+      "loss": 0.0601,
+      "step": 625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.9189644012945e-05,
+      "loss": 0.0487,
+      "step": 626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918834951456311e-05,
+      "loss": 0.1191,
+      "step": 627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918705501618122e-05,
+      "loss": 0.0414,
+      "step": 628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918576051779935e-05,
+      "loss": 0.1064,
+      "step": 629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918446601941748e-05,
+      "loss": 0.0509,
+      "step": 630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.91831715210356e-05,
+      "loss": 0.0564,
+      "step": 631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918187702265373e-05,
+      "loss": 0.0561,
+      "step": 632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.918058252427184e-05,
+      "loss": 0.1064,
+      "step": 633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917928802588997e-05,
+      "loss": 0.1186,
+      "step": 634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.91779935275081e-05,
+      "loss": 0.0596,
+      "step": 635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917669902912622e-05,
+      "loss": 0.0468,
+      "step": 636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917540453074434e-05,
+      "loss": 0.1067,
+      "step": 637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917411003236246e-05,
+      "loss": 0.0609,
+      "step": 638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917281553398059e-05,
+      "loss": 0.0719,
+      "step": 639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917152103559872e-05,
+      "loss": 0.0301,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.917022653721684e-05,
+      "loss": 0.0948,
+      "step": 641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916893203883496e-05,
+      "loss": 0.0814,
+      "step": 642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916763754045308e-05,
+      "loss": 0.0358,
+      "step": 643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.91663430420712e-05,
+      "loss": 0.0568,
+      "step": 644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916504854368932e-05,
+      "loss": 0.1406,
+      "step": 645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916375404530745e-05,
+      "loss": 0.0767,
+      "step": 646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916245954692556e-05,
+      "loss": 0.0933,
+      "step": 647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.916116504854369e-05,
+      "loss": 0.1341,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915987055016181e-05,
+      "loss": 0.0623,
+      "step": 649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915857605177994e-05,
+      "loss": 0.1041,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915728155339807e-05,
+      "loss": 0.0611,
+      "step": 651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915598705501618e-05,
+      "loss": 0.0721,
+      "step": 652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915469255663431e-05,
+      "loss": 0.1013,
+      "step": 653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915339805825243e-05,
+      "loss": 0.0609,
+      "step": 654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915210355987056e-05,
+      "loss": 0.0662,
+      "step": 655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.915080906148869e-05,
+      "loss": 0.0945,
+      "step": 656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.91495145631068e-05,
+      "loss": 0.0774,
+      "step": 657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914822006472493e-05,
+      "loss": 0.0746,
+      "step": 658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914692556634305e-05,
+      "loss": 0.0564,
+      "step": 659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914563106796117e-05,
+      "loss": 0.073,
+      "step": 660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914433656957929e-05,
+      "loss": 0.1301,
+      "step": 661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.91430420711974e-05,
+      "loss": 0.0602,
+      "step": 662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914174757281553e-05,
+      "loss": 0.0948,
+      "step": 663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.914045307443366e-05,
+      "loss": 0.0502,
+      "step": 664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913915857605179e-05,
+      "loss": 0.0507,
+      "step": 665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913786407766991e-05,
+      "loss": 0.0831,
+      "step": 666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913656957928803e-05,
+      "loss": 0.0503,
+      "step": 667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913527508090615e-05,
+      "loss": 0.0788,
+      "step": 668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913398058252428e-05,
+      "loss": 0.0924,
+      "step": 669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.91326860841424e-05,
+      "loss": 0.0668,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913139158576052e-05,
+      "loss": 0.0726,
+      "step": 671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.913009708737864e-05,
+      "loss": 0.0581,
+      "step": 672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912880258899677e-05,
+      "loss": 0.08,
+      "step": 673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91275080906149e-05,
+      "loss": 0.108,
+      "step": 674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912621359223302e-05,
+      "loss": 0.1409,
+      "step": 675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912491909385114e-05,
+      "loss": 0.0702,
+      "step": 676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912362459546925e-05,
+      "loss": 0.0899,
+      "step": 677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.912233009708738e-05,
+      "loss": 0.0944,
+      "step": 678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.91210355987055e-05,
+      "loss": 0.0501,
+      "step": 679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911974110032363e-05,
+      "loss": 0.1506,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911844660194174e-05,
+      "loss": 0.062,
+      "step": 681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911715210355987e-05,
+      "loss": 0.0638,
+      "step": 682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.9115857605178e-05,
+      "loss": 0.0673,
+      "step": 683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911456310679612e-05,
+      "loss": 0.0837,
+      "step": 684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911326860841425e-05,
+      "loss": 0.11,
+      "step": 685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911197411003236e-05,
+      "loss": 0.0638,
+      "step": 686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911067961165049e-05,
+      "loss": 0.0673,
+      "step": 687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910938511326862e-05,
+      "loss": 0.0646,
+      "step": 688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910809061488674e-05,
+      "loss": 0.059,
+      "step": 689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910679611650487e-05,
+      "loss": 0.0484,
+      "step": 690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910550161812298e-05,
+      "loss": 0.0646,
+      "step": 691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910420711974111e-05,
+      "loss": 0.1035,
+      "step": 692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910291262135922e-05,
+      "loss": 0.1112,
+      "step": 693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910161812297735e-05,
+      "loss": 0.0718,
+      "step": 694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.910032362459547e-05,
+      "loss": 0.0722,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909902912621359e-05,
+      "loss": 0.0578,
+      "step": 696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909773462783171e-05,
+      "loss": 0.049,
+      "step": 697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909644012944984e-05,
+      "loss": 0.0342,
+      "step": 698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909514563106797e-05,
+      "loss": 0.0399,
+      "step": 699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.90938511326861e-05,
+      "loss": 0.0551,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909255663430421e-05,
+      "loss": 0.0771,
+      "step": 701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.909126213592233e-05,
+      "loss": 0.0405,
+      "step": 702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908996763754046e-05,
+      "loss": 0.0779,
+      "step": 703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908867313915859e-05,
+      "loss": 0.1113,
+      "step": 704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90873786407767e-05,
+      "loss": 0.0675,
+      "step": 705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908608414239483e-05,
+      "loss": 0.0723,
+      "step": 706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908478964401295e-05,
+      "loss": 0.0649,
+      "step": 707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908349514563108e-05,
+      "loss": 0.046,
+      "step": 708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908220064724919e-05,
+      "loss": 0.04,
+      "step": 709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.908090614886732e-05,
+      "loss": 0.1368,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907961165048543e-05,
+      "loss": 0.08,
+      "step": 711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907831715210356e-05,
+      "loss": 0.0541,
+      "step": 712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907702265372169e-05,
+      "loss": 0.0597,
+      "step": 713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907572815533981e-05,
+      "loss": 0.0623,
+      "step": 714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907443365695794e-05,
+      "loss": 0.0378,
+      "step": 715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907313915857605e-05,
+      "loss": 0.0827,
+      "step": 716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.907184466019418e-05,
+      "loss": 0.0617,
+      "step": 717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.90705501618123e-05,
+      "loss": 0.0461,
+      "step": 718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906925566343043e-05,
+      "loss": 0.0684,
+      "step": 719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906796116504854e-05,
+      "loss": 0.0522,
+      "step": 720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906666666666667e-05,
+      "loss": 0.0326,
+      "step": 721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90653721682848e-05,
+      "loss": 0.0636,
+      "step": 722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906407766990292e-05,
+      "loss": 0.0952,
+      "step": 723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906278317152105e-05,
+      "loss": 0.0487,
+      "step": 724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906148867313916e-05,
+      "loss": 0.0629,
+      "step": 725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.906019417475728e-05,
+      "loss": 0.0742,
+      "step": 726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90588996763754e-05,
+      "loss": 0.0497,
+      "step": 727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905760517799353e-05,
+      "loss": 0.0835,
+      "step": 728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905631067961166e-05,
+      "loss": 0.0339,
+      "step": 729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905501618122977e-05,
+      "loss": 0.0442,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.90537216828479e-05,
+      "loss": 0.0545,
+      "step": 731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905242718446602e-05,
+      "loss": 0.0624,
+      "step": 732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.905113268608415e-05,
+      "loss": 0.0755,
+      "step": 733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904983818770228e-05,
+      "loss": 0.0615,
+      "step": 734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904854368932039e-05,
+      "loss": 0.0648,
+      "step": 735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904724919093852e-05,
+      "loss": 0.0655,
+      "step": 736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904595469255664e-05,
+      "loss": 0.0641,
+      "step": 737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904466019417477e-05,
+      "loss": 0.0677,
+      "step": 738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.90433656957929e-05,
+      "loss": 0.0331,
+      "step": 739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904207119741101e-05,
+      "loss": 0.0774,
+      "step": 740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.904077669902914e-05,
+      "loss": 0.0834,
+      "step": 741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903948220064725e-05,
+      "loss": 0.0417,
+      "step": 742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903818770226537e-05,
+      "loss": 0.108,
+      "step": 743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.90368932038835e-05,
+      "loss": 0.0952,
+      "step": 744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903559870550161e-05,
+      "loss": 0.0842,
+      "step": 745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903430420711974e-05,
+      "loss": 0.0584,
+      "step": 746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903300970873787e-05,
+      "loss": 0.0795,
+      "step": 747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.9031715210356e-05,
+      "loss": 0.0352,
+      "step": 748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.903042071197412e-05,
+      "loss": 0.0618,
+      "step": 749
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902912621359223e-05,
+      "loss": 0.0456,
+      "step": 750
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902783171521036e-05,
+      "loss": 0.0742,
+      "step": 751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902653721682849e-05,
+      "loss": 0.1197,
+      "step": 752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902524271844661e-05,
+      "loss": 0.083,
+      "step": 753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902394822006473e-05,
+      "loss": 0.0872,
+      "step": 754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902265372168285e-05,
+      "loss": 0.0823,
+      "step": 755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.902135922330098e-05,
+      "loss": 0.0535,
+      "step": 756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.90200647249191e-05,
+      "loss": 0.0484,
+      "step": 757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901877022653722e-05,
+      "loss": 0.0962,
+      "step": 758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901747572815535e-05,
+      "loss": 0.0754,
+      "step": 759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901618122977346e-05,
+      "loss": 0.0688,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901488673139159e-05,
+      "loss": 0.0676,
+      "step": 761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901359223300971e-05,
+      "loss": 0.0464,
+      "step": 762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901229773462784e-05,
+      "loss": 0.0791,
+      "step": 763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.901100323624595e-05,
+      "loss": 0.0612,
+      "step": 764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900970873786408e-05,
+      "loss": 0.0337,
+      "step": 765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90084142394822e-05,
+      "loss": 0.0894,
+      "step": 766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900711974110033e-05,
+      "loss": 0.0587,
+      "step": 767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900582524271846e-05,
+      "loss": 0.0588,
+      "step": 768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900453074433657e-05,
+      "loss": 0.0529,
+      "step": 769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.90032362459547e-05,
+      "loss": 0.0545,
+      "step": 770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900194174757282e-05,
+      "loss": 0.0657,
+      "step": 771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.900064724919095e-05,
+      "loss": 0.089,
+      "step": 772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899935275080908e-05,
+      "loss": 0.0574,
+      "step": 773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899805825242718e-05,
+      "loss": 0.058,
+      "step": 774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89967637540453e-05,
+      "loss": 0.0834,
+      "step": 775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899546925566343e-05,
+      "loss": 0.073,
+      "step": 776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899417475728156e-05,
+      "loss": 0.0352,
+      "step": 777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899288025889968e-05,
+      "loss": 0.082,
+      "step": 778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.89915857605178e-05,
+      "loss": 0.1105,
+      "step": 779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.899029126213592e-05,
+      "loss": 0.0939,
+      "step": 780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898899676375405e-05,
+      "loss": 0.0449,
+      "step": 781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898770226537218e-05,
+      "loss": 0.0322,
+      "step": 782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89864077669903e-05,
+      "loss": 0.0537,
+      "step": 783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898511326860842e-05,
+      "loss": 0.0933,
+      "step": 784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898381877022654e-05,
+      "loss": 0.0441,
+      "step": 785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.898252427184467e-05,
+      "loss": 0.0513,
+      "step": 786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89812297734628e-05,
+      "loss": 0.0973,
+      "step": 787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897993527508091e-05,
+      "loss": 0.043,
+      "step": 788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897864077669904e-05,
+      "loss": 0.073,
+      "step": 789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897734627831716e-05,
+      "loss": 0.0495,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897605177993527e-05,
+      "loss": 0.0549,
+      "step": 791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.89747572815534e-05,
+      "loss": 0.1102,
+      "step": 792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897346278317153e-05,
+      "loss": 0.05,
+      "step": 793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897216828478964e-05,
+      "loss": 0.0516,
+      "step": 794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.897087378640777e-05,
+      "loss": 0.0717,
+      "step": 795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89695792880259e-05,
+      "loss": 0.0386,
+      "step": 796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896828478964402e-05,
+      "loss": 0.0775,
+      "step": 797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896699029126213e-05,
+      "loss": 0.0428,
+      "step": 798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896569579288026e-05,
+      "loss": 0.0815,
+      "step": 799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896440129449839e-05,
+      "loss": 0.0286,
+      "step": 800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896310679611651e-05,
+      "loss": 0.0914,
+      "step": 801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896181229773464e-05,
+      "loss": 0.1202,
+      "step": 802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896051779935275e-05,
+      "loss": 0.0742,
+      "step": 803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895922330097088e-05,
+      "loss": 0.0551,
+      "step": 804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8957928802589e-05,
+      "loss": 0.0973,
+      "step": 805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895663430420713e-05,
+      "loss": 0.0436,
+      "step": 806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895533980582525e-05,
+      "loss": 0.0387,
+      "step": 807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895404530744337e-05,
+      "loss": 0.118,
+      "step": 808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895275080906149e-05,
+      "loss": 0.0735,
+      "step": 809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895145631067961e-05,
+      "loss": 0.0965,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895016181229774e-05,
+      "loss": 0.0668,
+      "step": 811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894886731391587e-05,
+      "loss": 0.0746,
+      "step": 812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894757281553398e-05,
+      "loss": 0.0457,
+      "step": 813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.89462783171521e-05,
+      "loss": 0.0734,
+      "step": 814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894498381877023e-05,
+      "loss": 0.0538,
+      "step": 815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894368932038836e-05,
+      "loss": 0.0495,
+      "step": 816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.894239482200648e-05,
+      "loss": 0.0941,
+      "step": 817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.89411003236246e-05,
+      "loss": 0.0606,
+      "step": 818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893980582524272e-05,
+      "loss": 0.0436,
+      "step": 819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893851132686085e-05,
+      "loss": 0.0984,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893721682847898e-05,
+      "loss": 0.0673,
+      "step": 821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893592233009709e-05,
+      "loss": 0.0562,
+      "step": 822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.89346278317152e-05,
+      "loss": 0.0826,
+      "step": 823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893333333333333e-05,
+      "loss": 0.112,
+      "step": 824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893203883495146e-05,
+      "loss": 0.0593,
+      "step": 825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.893074433656958e-05,
+      "loss": 0.0557,
+      "step": 826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892944983818771e-05,
+      "loss": 0.0622,
+      "step": 827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892815533980582e-05,
+      "loss": 0.0747,
+      "step": 828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892686084142395e-05,
+      "loss": 0.0412,
+      "step": 829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892556634304208e-05,
+      "loss": 0.0726,
+      "step": 830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89242718446602e-05,
+      "loss": 0.0515,
+      "step": 831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892297734627833e-05,
+      "loss": 0.0679,
+      "step": 832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892168284789644e-05,
+      "loss": 0.0995,
+      "step": 833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.892038834951457e-05,
+      "loss": 0.0598,
+      "step": 834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89190938511327e-05,
+      "loss": 0.0462,
+      "step": 835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891779935275082e-05,
+      "loss": 0.0677,
+      "step": 836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891650485436894e-05,
+      "loss": 0.0482,
+      "step": 837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891521035598706e-05,
+      "loss": 0.0426,
+      "step": 838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891391585760517e-05,
+      "loss": 0.0737,
+      "step": 839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.89126213592233e-05,
+      "loss": 0.0853,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891132686084143e-05,
+      "loss": 0.0598,
+      "step": 841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.891003236245955e-05,
+      "loss": 0.0777,
+      "step": 842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890873786407767e-05,
+      "loss": 0.0882,
+      "step": 843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.89074433656958e-05,
+      "loss": 0.1516,
+      "step": 844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890614886731392e-05,
+      "loss": 0.119,
+      "step": 845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890485436893205e-05,
+      "loss": 0.04,
+      "step": 846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890355987055016e-05,
+      "loss": 0.0595,
+      "step": 847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890226537216829e-05,
+      "loss": 0.0479,
+      "step": 848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.890097087378641e-05,
+      "loss": 0.0703,
+      "step": 849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889967637540454e-05,
+      "loss": 0.0541,
+      "step": 850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889838187702267e-05,
+      "loss": 0.0446,
+      "step": 851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889708737864078e-05,
+      "loss": 0.0899,
+      "step": 852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.88957928802589e-05,
+      "loss": 0.044,
+      "step": 853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889449838187703e-05,
+      "loss": 0.0493,
+      "step": 854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889320388349516e-05,
+      "loss": 0.0769,
+      "step": 855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889190938511327e-05,
+      "loss": 0.0895,
+      "step": 856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.889061488673139e-05,
+      "loss": 0.0355,
+      "step": 857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888932038834951e-05,
+      "loss": 0.0602,
+      "step": 858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888802588996764e-05,
+      "loss": 0.0503,
+      "step": 859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888673139158577e-05,
+      "loss": 0.0695,
+      "step": 860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888543689320389e-05,
+      "loss": 0.0474,
+      "step": 861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.8884142394822e-05,
+      "loss": 0.0873,
+      "step": 862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888284789644013e-05,
+      "loss": 0.0385,
+      "step": 863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888155339805826e-05,
+      "loss": 0.0725,
+      "step": 864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.888025889967638e-05,
+      "loss": 0.0901,
+      "step": 865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887896440129451e-05,
+      "loss": 0.0756,
+      "step": 866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887766990291262e-05,
+      "loss": 0.0598,
+      "step": 867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887637540453075e-05,
+      "loss": 0.045,
+      "step": 868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887508090614888e-05,
+      "loss": 0.063,
+      "step": 869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.8873786407767e-05,
+      "loss": 0.0368,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887249190938512e-05,
+      "loss": 0.0486,
+      "step": 871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.887119741100323e-05,
+      "loss": 0.0917,
+      "step": 872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886990291262136e-05,
+      "loss": 0.0732,
+      "step": 873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886860841423948e-05,
+      "loss": 0.0483,
+      "step": 874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886731391585761e-05,
+      "loss": 0.0375,
+      "step": 875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886601941747574e-05,
+      "loss": 0.0916,
+      "step": 876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886472491909385e-05,
+      "loss": 0.0517,
+      "step": 877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886343042071198e-05,
+      "loss": 0.0769,
+      "step": 878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.88621359223301e-05,
+      "loss": 0.066,
+      "step": 879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.886084142394823e-05,
+      "loss": 0.0663,
+      "step": 880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885954692556634e-05,
+      "loss": 0.0957,
+      "step": 881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885825242718447e-05,
+      "loss": 0.0591,
+      "step": 882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.88569579288026e-05,
+      "loss": 0.0676,
+      "step": 883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885566343042072e-05,
+      "loss": 0.0548,
+      "step": 884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885436893203885e-05,
+      "loss": 0.0677,
+      "step": 885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885307443365696e-05,
+      "loss": 0.0842,
+      "step": 886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.885177993527509e-05,
+      "loss": 0.088,
+      "step": 887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.88504854368932e-05,
+      "loss": 0.0587,
+      "step": 888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884919093851133e-05,
+      "loss": 0.0826,
+      "step": 889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884789644012945e-05,
+      "loss": 0.0829,
+      "step": 890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884660194174757e-05,
+      "loss": 0.0618,
+      "step": 891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88453074433657e-05,
+      "loss": 0.0897,
+      "step": 892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884401294498382e-05,
+      "loss": 0.0855,
+      "step": 893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884271844660195e-05,
+      "loss": 0.0832,
+      "step": 894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884142394822007e-05,
+      "loss": 0.0578,
+      "step": 895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884012944983819e-05,
+      "loss": 0.0577,
+      "step": 896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883883495145631e-05,
+      "loss": 0.0387,
+      "step": 897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883754045307444e-05,
+      "loss": 0.0864,
+      "step": 898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883624595469257e-05,
+      "loss": 0.0818,
+      "step": 899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88349514563107e-05,
+      "loss": 0.0609,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88336569579288e-05,
+      "loss": 0.038,
+      "step": 901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883236245954693e-05,
+      "loss": 0.0965,
+      "step": 902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.883106796116506e-05,
+      "loss": 0.042,
+      "step": 903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882977346278319e-05,
+      "loss": 0.0657,
+      "step": 904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.88284789644013e-05,
+      "loss": 0.0523,
+      "step": 905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882718446601941e-05,
+      "loss": 0.054,
+      "step": 906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882588996763754e-05,
+      "loss": 0.0582,
+      "step": 907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882459546925567e-05,
+      "loss": 0.0707,
+      "step": 908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882330097087379e-05,
+      "loss": 0.0412,
+      "step": 909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882200647249192e-05,
+      "loss": 0.0618,
+      "step": 910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.882071197411003e-05,
+      "loss": 0.0822,
+      "step": 911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881941747572816e-05,
+      "loss": 0.0823,
+      "step": 912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881812297734628e-05,
+      "loss": 0.053,
+      "step": 913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881682847896441e-05,
+      "loss": 0.0629,
+      "step": 914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881553398058252e-05,
+      "loss": 0.0457,
+      "step": 915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881423948220065e-05,
+      "loss": 0.0926,
+      "step": 916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881294498381878e-05,
+      "loss": 0.0323,
+      "step": 917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.88116504854369e-05,
+      "loss": 0.0684,
+      "step": 918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.881035598705503e-05,
+      "loss": 0.0825,
+      "step": 919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880906148867314e-05,
+      "loss": 0.0555,
+      "step": 920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880776699029126e-05,
+      "loss": 0.1019,
+      "step": 921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880647249190938e-05,
+      "loss": 0.1084,
+      "step": 922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880517799352751e-05,
+      "loss": 0.0631,
+      "step": 923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880388349514564e-05,
+      "loss": 0.0618,
+      "step": 924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880258899676376e-05,
+      "loss": 0.0548,
+      "step": 925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.880129449838188e-05,
+      "loss": 0.0668,
+      "step": 926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.88e-05,
+      "loss": 0.0371,
+      "step": 927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879870550161813e-05,
+      "loss": 0.131,
+      "step": 928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879741100323626e-05,
+      "loss": 0.0533,
+      "step": 929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879611650485437e-05,
+      "loss": 0.0678,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.87948220064725e-05,
+      "loss": 0.0683,
+      "step": 931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879352750809062e-05,
+      "loss": 0.0539,
+      "step": 932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879223300970875e-05,
+      "loss": 0.0304,
+      "step": 933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.879093851132688e-05,
+      "loss": 0.0522,
+      "step": 934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878964401294499e-05,
+      "loss": 0.0799,
+      "step": 935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878834951456311e-05,
+      "loss": 0.0384,
+      "step": 936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878705501618123e-05,
+      "loss": 0.0596,
+      "step": 937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878576051779935e-05,
+      "loss": 0.0365,
+      "step": 938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878446601941748e-05,
+      "loss": 0.0592,
+      "step": 939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87831715210356e-05,
+      "loss": 0.0801,
+      "step": 940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878187702265372e-05,
+      "loss": 0.0772,
+      "step": 941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.878058252427185e-05,
+      "loss": 0.0808,
+      "step": 942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877928802588997e-05,
+      "loss": 0.0843,
+      "step": 943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87779935275081e-05,
+      "loss": 0.0661,
+      "step": 944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877669902912621e-05,
+      "loss": 0.1222,
+      "step": 945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877540453074434e-05,
+      "loss": 0.0332,
+      "step": 946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877411003236247e-05,
+      "loss": 0.0642,
+      "step": 947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87728155339806e-05,
+      "loss": 0.0596,
+      "step": 948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.87715210355987e-05,
+      "loss": 0.1209,
+      "step": 949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.877022653721683e-05,
+      "loss": 0.0529,
+      "step": 950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876893203883496e-05,
+      "loss": 0.0549,
+      "step": 951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876763754045309e-05,
+      "loss": 0.0643,
+      "step": 952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.87663430420712e-05,
+      "loss": 0.0837,
+      "step": 953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876504854368933e-05,
+      "loss": 0.0486,
+      "step": 954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876375404530744e-05,
+      "loss": 0.0596,
+      "step": 955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876245954692557e-05,
+      "loss": 0.0547,
+      "step": 956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.876116504854369e-05,
+      "loss": 0.0257,
+      "step": 957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875987055016182e-05,
+      "loss": 0.075,
+      "step": 958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875857605177995e-05,
+      "loss": 0.089,
+      "step": 959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875728155339806e-05,
+      "loss": 0.1249,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875598705501618e-05,
+      "loss": 0.0836,
+      "step": 961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875469255663431e-05,
+      "loss": 0.0582,
+      "step": 962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875339805825244e-05,
+      "loss": 0.0433,
+      "step": 963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875210355987055e-05,
+      "loss": 0.0677,
+      "step": 964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.875080906148868e-05,
+      "loss": 0.0414,
+      "step": 965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.87495145631068e-05,
+      "loss": 0.0434,
+      "step": 966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874822006472493e-05,
+      "loss": 0.0799,
+      "step": 967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874692556634306e-05,
+      "loss": 0.0765,
+      "step": 968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874563106796117e-05,
+      "loss": 0.0713,
+      "step": 969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874433656957928e-05,
+      "loss": 0.0259,
+      "step": 970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874304207119741e-05,
+      "loss": 0.1004,
+      "step": 971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874174757281554e-05,
+      "loss": 0.0475,
+      "step": 972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.874045307443366e-05,
+      "loss": 0.0533,
+      "step": 973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873915857605178e-05,
+      "loss": 0.0709,
+      "step": 974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.87378640776699e-05,
+      "loss": 0.064,
+      "step": 975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873656957928803e-05,
+      "loss": 0.0369,
+      "step": 976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873527508090616e-05,
+      "loss": 0.0751,
+      "step": 977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873398058252428e-05,
+      "loss": 0.0585,
+      "step": 978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.87326860841424e-05,
+      "loss": 0.0451,
+      "step": 979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873139158576052e-05,
+      "loss": 0.0472,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.873009708737865e-05,
+      "loss": 0.1507,
+      "step": 981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872880258899678e-05,
+      "loss": 0.0685,
+      "step": 982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.87275080906149e-05,
+      "loss": 0.0398,
+      "step": 983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872621359223301e-05,
+      "loss": 0.0696,
+      "step": 984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872491909385114e-05,
+      "loss": 0.0688,
+      "step": 985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872362459546925e-05,
+      "loss": 0.0402,
+      "step": 986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872233009708738e-05,
+      "loss": 0.0725,
+      "step": 987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.872103559870551e-05,
+      "loss": 0.0746,
+      "step": 988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871974110032362e-05,
+      "loss": 0.0757,
+      "step": 989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871844660194175e-05,
+      "loss": 0.0832,
+      "step": 990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871715210355987e-05,
+      "loss": 0.0714,
+      "step": 991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.8715857605178e-05,
+      "loss": 0.0948,
+      "step": 992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871456310679613e-05,
+      "loss": 0.0807,
+      "step": 993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871326860841424e-05,
+      "loss": 0.059,
+      "step": 994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.871197411003237e-05,
+      "loss": 0.1009,
+      "step": 995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.87106796116505e-05,
+      "loss": 0.0444,
+      "step": 996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870938511326862e-05,
+      "loss": 0.0642,
+      "step": 997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870809061488673e-05,
+      "loss": 0.0532,
+      "step": 998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870679611650486e-05,
+      "loss": 0.1145,
+      "step": 999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870550161812299e-05,
+      "loss": 0.0386,
+      "step": 1000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870420711974111e-05,
+      "loss": 0.0624,
+      "step": 1001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870291262135923e-05,
+      "loss": 0.0512,
+      "step": 1002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870161812297735e-05,
+      "loss": 0.0812,
+      "step": 1003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.870032362459547e-05,
+      "loss": 0.0581,
+      "step": 1004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869902912621359e-05,
+      "loss": 0.063,
+      "step": 1005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869773462783172e-05,
+      "loss": 0.0348,
+      "step": 1006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869644012944985e-05,
+      "loss": 0.0843,
+      "step": 1007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869514563106796e-05,
+      "loss": 0.0746,
+      "step": 1008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869385113268608e-05,
+      "loss": 0.0772,
+      "step": 1009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869255663430421e-05,
+      "loss": 0.0397,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.869126213592234e-05,
+      "loss": 0.0748,
+      "step": 1011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868996763754046e-05,
+      "loss": 0.0766,
+      "step": 1012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868867313915858e-05,
+      "loss": 0.0721,
+      "step": 1013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.86873786407767e-05,
+      "loss": 0.0325,
+      "step": 1014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868608414239483e-05,
+      "loss": 0.0259,
+      "step": 1015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868478964401296e-05,
+      "loss": 0.0523,
+      "step": 1016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868349514563108e-05,
+      "loss": 0.1067,
+      "step": 1017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.86822006472492e-05,
+      "loss": 0.0681,
+      "step": 1018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.868090614886731e-05,
+      "loss": 0.0529,
+      "step": 1019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867961165048544e-05,
+      "loss": 0.0394,
+      "step": 1020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867831715210356e-05,
+      "loss": 0.0337,
+      "step": 1021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867702265372169e-05,
+      "loss": 0.0758,
+      "step": 1022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.86757281553398e-05,
+      "loss": 0.0581,
+      "step": 1023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867443365695793e-05,
+      "loss": 0.0678,
+      "step": 1024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867313915857606e-05,
+      "loss": 0.0928,
+      "step": 1025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867184466019418e-05,
+      "loss": 0.0687,
+      "step": 1026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.867055016181231e-05,
+      "loss": 0.0391,
+      "step": 1027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866925566343042e-05,
+      "loss": 0.0327,
+      "step": 1028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866796116504855e-05,
+      "loss": 0.0715,
+      "step": 1029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866666666666668e-05,
+      "loss": 0.0448,
+      "step": 1030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.86653721682848e-05,
+      "loss": 0.039,
+      "step": 1031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866407766990291e-05,
+      "loss": 0.1586,
+      "step": 1032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866278317152104e-05,
+      "loss": 0.0681,
+      "step": 1033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866148867313917e-05,
+      "loss": 0.0523,
+      "step": 1034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.866019417475728e-05,
+      "loss": 0.0352,
+      "step": 1035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865889967637541e-05,
+      "loss": 0.0311,
+      "step": 1036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865760517799353e-05,
+      "loss": 0.0937,
+      "step": 1037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865631067961165e-05,
+      "loss": 0.0625,
+      "step": 1038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865501618122977e-05,
+      "loss": 0.0643,
+      "step": 1039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.86537216828479e-05,
+      "loss": 0.1006,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865242718446603e-05,
+      "loss": 0.0628,
+      "step": 1041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.865113268608414e-05,
+      "loss": 0.0565,
+      "step": 1042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864983818770227e-05,
+      "loss": 0.0903,
+      "step": 1043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864854368932039e-05,
+      "loss": 0.1194,
+      "step": 1044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864724919093852e-05,
+      "loss": 0.0646,
+      "step": 1045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864595469255665e-05,
+      "loss": 0.0344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864466019417476e-05,
+      "loss": 0.0855,
+      "step": 1047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864336569579289e-05,
+      "loss": 0.0977,
+      "step": 1048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864207119741101e-05,
+      "loss": 0.1221,
+      "step": 1049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.864077669902914e-05,
+      "loss": 0.0727,
+      "step": 1050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863948220064725e-05,
+      "loss": 0.0509,
+      "step": 1051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863818770226538e-05,
+      "loss": 0.0754,
+      "step": 1052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863689320388349e-05,
+      "loss": 0.1003,
+      "step": 1053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863559870550162e-05,
+      "loss": 0.0919,
+      "step": 1054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863430420711974e-05,
+      "loss": 0.0719,
+      "step": 1055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863300970873787e-05,
+      "loss": 0.0758,
+      "step": 1056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863171521035598e-05,
+      "loss": 0.0568,
+      "step": 1057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.863042071197411e-05,
+      "loss": 0.0484,
+      "step": 1058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862912621359224e-05,
+      "loss": 0.0701,
+      "step": 1059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862783171521036e-05,
+      "loss": 0.1152,
+      "step": 1060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862653721682849e-05,
+      "loss": 0.0893,
+      "step": 1061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.86252427184466e-05,
+      "loss": 0.089,
+      "step": 1062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862394822006473e-05,
+      "loss": 0.0722,
+      "step": 1063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862265372168286e-05,
+      "loss": 0.0473,
+      "step": 1064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.862135922330098e-05,
+      "loss": 0.0525,
+      "step": 1065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.86200647249191e-05,
+      "loss": 0.077,
+      "step": 1066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861877022653722e-05,
+      "loss": 0.0815,
+      "step": 1067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861747572815534e-05,
+      "loss": 0.0465,
+      "step": 1068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861618122977346e-05,
+      "loss": 0.0883,
+      "step": 1069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861488673139159e-05,
+      "loss": 0.0825,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861359223300972e-05,
+      "loss": 0.0632,
+      "step": 1071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861229773462783e-05,
+      "loss": 0.0819,
+      "step": 1072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.861100323624596e-05,
+      "loss": 0.0483,
+      "step": 1073
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860970873786408e-05,
+      "loss": 0.0918,
+      "step": 1074
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860841423948221e-05,
+      "loss": 0.0465,
+      "step": 1075
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860711974110034e-05,
+      "loss": 0.0833,
+      "step": 1076
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860582524271845e-05,
+      "loss": 0.0554,
+      "step": 1077
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860453074433658e-05,
+      "loss": 0.0486,
+      "step": 1078
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.86032362459547e-05,
+      "loss": 0.0394,
+      "step": 1079
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860194174757283e-05,
+      "loss": 0.0526,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.860064724919094e-05,
+      "loss": 0.0553,
+      "step": 1081
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859935275080907e-05,
+      "loss": 0.051,
+      "step": 1082
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.85980582524272e-05,
+      "loss": 0.0384,
+      "step": 1083
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859676375404531e-05,
+      "loss": 0.0503,
+      "step": 1084
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859546925566343e-05,
+      "loss": 0.077,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859417475728156e-05,
+      "loss": 0.0663,
+      "step": 1086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859288025889967e-05,
+      "loss": 0.0853,
+      "step": 1087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.85915857605178e-05,
+      "loss": 0.1039,
+      "step": 1088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.859029126213593e-05,
+      "loss": 0.0453,
+      "step": 1089
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858899676375405e-05,
+      "loss": 0.0483,
+      "step": 1090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858770226537217e-05,
+      "loss": 0.0428,
+      "step": 1091
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858640776699029e-05,
+      "loss": 0.0636,
+      "step": 1092
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858511326860842e-05,
+      "loss": 0.0518,
+      "step": 1093
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858381877022655e-05,
+      "loss": 0.0602,
+      "step": 1094
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858252427184467e-05,
+      "loss": 0.0638,
+      "step": 1095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.858122977346279e-05,
+      "loss": 0.0416,
+      "step": 1096
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857993527508091e-05,
+      "loss": 0.0822,
+      "step": 1097
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857864077669904e-05,
+      "loss": 0.0781,
+      "step": 1098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857734627831717e-05,
+      "loss": 0.0397,
+      "step": 1099
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857605177993528e-05,
+      "loss": 0.0529,
+      "step": 1100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857475728155339e-05,
+      "loss": 0.048,
+      "step": 1101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857346278317152e-05,
+      "loss": 0.085,
+      "step": 1102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857216828478964e-05,
+      "loss": 0.0804,
+      "step": 1103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.857087378640777e-05,
+      "loss": 0.0528,
+      "step": 1104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.85695792880259e-05,
+      "loss": 0.0442,
+      "step": 1105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856828478964401e-05,
+      "loss": 0.0485,
+      "step": 1106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856699029126214e-05,
+      "loss": 0.042,
+      "step": 1107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856569579288026e-05,
+      "loss": 0.0526,
+      "step": 1108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856440129449839e-05,
+      "loss": 0.0708,
+      "step": 1109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856310679611652e-05,
+      "loss": 0.0526,
+      "step": 1110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856181229773463e-05,
+      "loss": 0.0288,
+      "step": 1111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.856051779935276e-05,
+      "loss": 0.0567,
+      "step": 1112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855922330097088e-05,
+      "loss": 0.0484,
+      "step": 1113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855792880258901e-05,
+      "loss": 0.071,
+      "step": 1114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855663430420712e-05,
+      "loss": 0.0695,
+      "step": 1115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855533980582524e-05,
+      "loss": 0.055,
+      "step": 1116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855404530744336e-05,
+      "loss": 0.0748,
+      "step": 1117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855275080906149e-05,
+      "loss": 0.0695,
+      "step": 1118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855145631067962e-05,
+      "loss": 0.0458,
+      "step": 1119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.855016181229774e-05,
+      "loss": 0.1587,
+      "step": 1120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854886731391586e-05,
+      "loss": 0.0485,
+      "step": 1121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854757281553398e-05,
+      "loss": 0.079,
+      "step": 1122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854627831715211e-05,
+      "loss": 0.0435,
+      "step": 1123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854498381877024e-05,
+      "loss": 0.0898,
+      "step": 1124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854368932038835e-05,
+      "loss": 0.0494,
+      "step": 1125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.854239482200648e-05,
+      "loss": 0.0633,
+      "step": 1126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.85411003236246e-05,
+      "loss": 0.0325,
+      "step": 1127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853980582524273e-05,
+      "loss": 0.0564,
+      "step": 1128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853851132686085e-05,
+      "loss": 0.0802,
+      "step": 1129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853721682847897e-05,
+      "loss": 0.0428,
+      "step": 1130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.85359223300971e-05,
+      "loss": 0.0421,
+      "step": 1131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853462783171522e-05,
+      "loss": 0.0435,
+      "step": 1132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853333333333333e-05,
+      "loss": 0.0504,
+      "step": 1133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853203883495146e-05,
+      "loss": 0.0946,
+      "step": 1134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.853074433656957e-05,
+      "loss": 0.0727,
+      "step": 1135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.85294498381877e-05,
+      "loss": 0.0578,
+      "step": 1136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852815533980583e-05,
+      "loss": 0.0584,
+      "step": 1137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852686084142395e-05,
+      "loss": 0.0632,
+      "step": 1138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852556634304208e-05,
+      "loss": 0.0292,
+      "step": 1139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852427184466019e-05,
+      "loss": 0.0491,
+      "step": 1140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852297734627832e-05,
+      "loss": 0.0533,
+      "step": 1141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852168284789645e-05,
+      "loss": 0.0304,
+      "step": 1142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.852038834951457e-05,
+      "loss": 0.0827,
+      "step": 1143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.85190938511327e-05,
+      "loss": 0.0254,
+      "step": 1144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851779935275081e-05,
+      "loss": 0.0376,
+      "step": 1145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851650485436894e-05,
+      "loss": 0.0596,
+      "step": 1146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851521035598707e-05,
+      "loss": 0.0676,
+      "step": 1147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851391585760519e-05,
+      "loss": 0.0363,
+      "step": 1148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.85126213592233e-05,
+      "loss": 0.0821,
+      "step": 1149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851132686084142e-05,
+      "loss": 0.0536,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.851003236245954e-05,
+      "loss": 0.0274,
+      "step": 1151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850873786407767e-05,
+      "loss": 0.1172,
+      "step": 1152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.85074433656958e-05,
+      "loss": 0.039,
+      "step": 1153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850614886731392e-05,
+      "loss": 0.0643,
+      "step": 1154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850485436893204e-05,
+      "loss": 0.061,
+      "step": 1155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850355987055016e-05,
+      "loss": 0.0484,
+      "step": 1156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850226537216829e-05,
+      "loss": 0.0629,
+      "step": 1157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.850097087378642e-05,
+      "loss": 0.0486,
+      "step": 1158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849967637540453e-05,
+      "loss": 0.0495,
+      "step": 1159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849838187702266e-05,
+      "loss": 0.0729,
+      "step": 1160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849708737864078e-05,
+      "loss": 0.0324,
+      "step": 1161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849579288025891e-05,
+      "loss": 0.0475,
+      "step": 1162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849449838187704e-05,
+      "loss": 0.0434,
+      "step": 1163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849320388349515e-05,
+      "loss": 0.0505,
+      "step": 1164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849190938511326e-05,
+      "loss": 0.0733,
+      "step": 1165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.849061488673139e-05,
+      "loss": 0.0604,
+      "step": 1166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848932038834952e-05,
+      "loss": 0.0608,
+      "step": 1167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848802588996764e-05,
+      "loss": 0.0386,
+      "step": 1168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848673139158577e-05,
+      "loss": 0.0616,
+      "step": 1169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848543689320388e-05,
+      "loss": 0.0475,
+      "step": 1170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848414239482201e-05,
+      "loss": 0.0405,
+      "step": 1171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848284789644014e-05,
+      "loss": 0.0667,
+      "step": 1172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848155339805826e-05,
+      "loss": 0.0608,
+      "step": 1173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.848025889967638e-05,
+      "loss": 0.0563,
+      "step": 1174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.84789644012945e-05,
+      "loss": 0.0473,
+      "step": 1175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847766990291263e-05,
+      "loss": 0.0425,
+      "step": 1176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847637540453075e-05,
+      "loss": 0.0969,
+      "step": 1177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847508090614888e-05,
+      "loss": 0.0737,
+      "step": 1178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.8473786407767e-05,
+      "loss": 0.0505,
+      "step": 1179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847249190938512e-05,
+      "loss": 0.0686,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.847119741100325e-05,
+      "loss": 0.0721,
+      "step": 1181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846990291262136e-05,
+      "loss": 0.0486,
+      "step": 1182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846860841423949e-05,
+      "loss": 0.0493,
+      "step": 1183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.84673139158576e-05,
+      "loss": 0.0565,
+      "step": 1184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846601941747573e-05,
+      "loss": 0.0497,
+      "step": 1185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846472491909385e-05,
+      "loss": 0.0494,
+      "step": 1186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846343042071198e-05,
+      "loss": 0.0836,
+      "step": 1187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.84621359223301e-05,
+      "loss": 0.0616,
+      "step": 1188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.846084142394822e-05,
+      "loss": 0.0482,
+      "step": 1189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845954692556635e-05,
+      "loss": 0.0712,
+      "step": 1190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845825242718447e-05,
+      "loss": 0.0424,
+      "step": 1191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.84569579288026e-05,
+      "loss": 0.0579,
+      "step": 1192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845566343042073e-05,
+      "loss": 0.0532,
+      "step": 1193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845436893203884e-05,
+      "loss": 0.0856,
+      "step": 1194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845307443365697e-05,
+      "loss": 0.0668,
+      "step": 1195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845177993527509e-05,
+      "loss": 0.0861,
+      "step": 1196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.845048543689322e-05,
+      "loss": 0.0448,
+      "step": 1197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844919093851133e-05,
+      "loss": 0.0965,
+      "step": 1198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844789644012944e-05,
+      "loss": 0.0435,
+      "step": 1199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844660194174757e-05,
+      "loss": 0.0701,
+      "step": 1200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.84453074433657e-05,
+      "loss": 0.0248,
+      "step": 1201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844401294498382e-05,
+      "loss": 0.0316,
+      "step": 1202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844271844660195e-05,
+      "loss": 0.0652,
+      "step": 1203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844142394822006e-05,
+      "loss": 0.0485,
+      "step": 1204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.844012944983819e-05,
+      "loss": 0.0363,
+      "step": 1205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843883495145632e-05,
+      "loss": 0.0337,
+      "step": 1206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843754045307444e-05,
+      "loss": 0.0362,
+      "step": 1207
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843624595469256e-05,
+      "loss": 0.0438,
+      "step": 1208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843495145631068e-05,
+      "loss": 0.0538,
+      "step": 1209
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843365695792881e-05,
+      "loss": 0.0633,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843236245954694e-05,
+      "loss": 0.0722,
+      "step": 1211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.843106796116506e-05,
+      "loss": 0.0562,
+      "step": 1212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842977346278318e-05,
+      "loss": 0.0294,
+      "step": 1213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842847896440129e-05,
+      "loss": 0.0734,
+      "step": 1214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842718446601942e-05,
+      "loss": 0.0452,
+      "step": 1215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842588996763754e-05,
+      "loss": 0.0392,
+      "step": 1216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842459546925567e-05,
+      "loss": 0.082,
+      "step": 1217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842330097087378e-05,
+      "loss": 0.1053,
+      "step": 1218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842200647249191e-05,
+      "loss": 0.0988,
+      "step": 1219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.842071197411004e-05,
+      "loss": 0.0356,
+      "step": 1220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841941747572816e-05,
+      "loss": 0.059,
+      "step": 1221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841812297734629e-05,
+      "loss": 0.0451,
+      "step": 1222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.84168284789644e-05,
+      "loss": 0.0754,
+      "step": 1223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841553398058253e-05,
+      "loss": 0.0286,
+      "step": 1224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841423948220065e-05,
+      "loss": 0.0869,
+      "step": 1225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841294498381878e-05,
+      "loss": 0.0448,
+      "step": 1226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841165048543691e-05,
+      "loss": 0.0669,
+      "step": 1227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.841035598705502e-05,
+      "loss": 0.0549,
+      "step": 1228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840906148867315e-05,
+      "loss": 0.0331,
+      "step": 1229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840776699029126e-05,
+      "loss": 0.0573,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840647249190939e-05,
+      "loss": 0.0646,
+      "step": 1231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840517799352751e-05,
+      "loss": 0.05,
+      "step": 1232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840388349514563e-05,
+      "loss": 0.2505,
+      "step": 1233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840258899676375e-05,
+      "loss": 0.0612,
+      "step": 1234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.840129449838188e-05,
+      "loss": 0.0699,
+      "step": 1235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84e-05,
+      "loss": 0.0416,
+      "step": 1236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839870550161813e-05,
+      "loss": 0.0824,
+      "step": 1237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839741100323625e-05,
+      "loss": 0.0352,
+      "step": 1238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839611650485437e-05,
+      "loss": 0.0358,
+      "step": 1239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.83948220064725e-05,
+      "loss": 0.0282,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839352750809063e-05,
+      "loss": 0.0537,
+      "step": 1241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839223300970874e-05,
+      "loss": 0.1,
+      "step": 1242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.839093851132687e-05,
+      "loss": 0.0481,
+      "step": 1243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838964401294499e-05,
+      "loss": 0.0675,
+      "step": 1244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838834951456312e-05,
+      "loss": 0.0591,
+      "step": 1245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838705501618125e-05,
+      "loss": 0.0884,
+      "step": 1246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838576051779936e-05,
+      "loss": 0.0912,
+      "step": 1247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838446601941747e-05,
+      "loss": 0.0442,
+      "step": 1248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.83831715210356e-05,
+      "loss": 0.1109,
+      "step": 1249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838187702265372e-05,
+      "loss": 0.0378,
+      "step": 1250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838058252427185e-05,
+      "loss": 0.0586,
+      "step": 1251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837928802588996e-05,
+      "loss": 0.0335,
+      "step": 1252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837799352750809e-05,
+      "loss": 0.0545,
+      "step": 1253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837669902912622e-05,
+      "loss": 0.0353,
+      "step": 1254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837540453074434e-05,
+      "loss": 0.0714,
+      "step": 1255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837411003236247e-05,
+      "loss": 0.0253,
+      "step": 1256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837281553398058e-05,
+      "loss": 0.0544,
+      "step": 1257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837152103559871e-05,
+      "loss": 0.0845,
+      "step": 1258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.837022653721684e-05,
+      "loss": 0.069,
+      "step": 1259
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836893203883496e-05,
+      "loss": 0.0391,
+      "step": 1260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836763754045309e-05,
+      "loss": 0.0232,
+      "step": 1261
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.83663430420712e-05,
+      "loss": 0.0651,
+      "step": 1262
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836504854368932e-05,
+      "loss": 0.0803,
+      "step": 1263
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836375404530744e-05,
+      "loss": 0.04,
+      "step": 1264
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.836245954692557e-05,
+      "loss": 0.1043,
+      "step": 1265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.83611650485437e-05,
+      "loss": 0.0554,
+      "step": 1266
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835987055016181e-05,
+      "loss": 0.0521,
+      "step": 1267
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835857605177994e-05,
+      "loss": 0.0282,
+      "step": 1268
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835728155339806e-05,
+      "loss": 0.026,
+      "step": 1269
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835598705501619e-05,
+      "loss": 0.0669,
+      "step": 1270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835469255663432e-05,
+      "loss": 0.0486,
+      "step": 1271
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835339805825243e-05,
+      "loss": 0.0662,
+      "step": 1272
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835210355987055e-05,
+      "loss": 0.0785,
+      "step": 1273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.835080906148868e-05,
+      "loss": 0.0982,
+      "step": 1274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834951456310681e-05,
+      "loss": 0.0907,
+      "step": 1275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834822006472492e-05,
+      "loss": 0.0285,
+      "step": 1276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834692556634305e-05,
+      "loss": 0.1043,
+      "step": 1277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834563106796117e-05,
+      "loss": 0.0931,
+      "step": 1278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834433656957929e-05,
+      "loss": 0.0725,
+      "step": 1279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834304207119741e-05,
+      "loss": 0.0717,
+      "step": 1280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834174757281554e-05,
+      "loss": 0.0317,
+      "step": 1281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.834045307443365e-05,
+      "loss": 0.0579,
+      "step": 1282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833915857605178e-05,
+      "loss": 0.0883,
+      "step": 1283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.83378640776699e-05,
+      "loss": 0.0785,
+      "step": 1284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833656957928803e-05,
+      "loss": 0.0483,
+      "step": 1285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833527508090616e-05,
+      "loss": 0.0494,
+      "step": 1286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833398058252427e-05,
+      "loss": 0.0494,
+      "step": 1287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.83326860841424e-05,
+      "loss": 0.0268,
+      "step": 1288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833139158576053e-05,
+      "loss": 0.0405,
+      "step": 1289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.833009708737865e-05,
+      "loss": 0.0376,
+      "step": 1290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832880258899677e-05,
+      "loss": 0.0653,
+      "step": 1291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832750809061489e-05,
+      "loss": 0.064,
+      "step": 1292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832621359223302e-05,
+      "loss": 0.0449,
+      "step": 1293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832491909385115e-05,
+      "loss": 0.0648,
+      "step": 1294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832362459546927e-05,
+      "loss": 0.0746,
+      "step": 1295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.832233009708738e-05,
+      "loss": 0.0257,
+      "step": 1296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.83210355987055e-05,
+      "loss": 0.0586,
+      "step": 1297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831974110032362e-05,
+      "loss": 0.0385,
+      "step": 1298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831844660194175e-05,
+      "loss": 0.0498,
+      "step": 1299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831715210355988e-05,
+      "loss": 0.0493,
+      "step": 1300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831585760517799e-05,
+      "loss": 0.0549,
+      "step": 1301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831456310679612e-05,
+      "loss": 0.0846,
+      "step": 1302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831326860841424e-05,
+      "loss": 0.0465,
+      "step": 1303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.831197411003237e-05,
+      "loss": 0.0521,
+      "step": 1304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.83106796116505e-05,
+      "loss": 0.0674,
+      "step": 1305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830938511326861e-05,
+      "loss": 0.0389,
+      "step": 1306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830809061488674e-05,
+      "loss": 0.0453,
+      "step": 1307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830679611650486e-05,
+      "loss": 0.0486,
+      "step": 1308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830550161812299e-05,
+      "loss": 0.0305,
+      "step": 1309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.83042071197411e-05,
+      "loss": 0.0535,
+      "step": 1310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830291262135923e-05,
+      "loss": 0.0733,
+      "step": 1311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830161812297734e-05,
+      "loss": 0.0556,
+      "step": 1312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.830032362459547e-05,
+      "loss": 0.0429,
+      "step": 1313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.82990291262136e-05,
+      "loss": 0.0538,
+      "step": 1314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829773462783172e-05,
+      "loss": 0.0446,
+      "step": 1315
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829644012944984e-05,
+      "loss": 0.0856,
+      "step": 1316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829514563106796e-05,
+      "loss": 0.0362,
+      "step": 1317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829385113268609e-05,
+      "loss": 0.0406,
+      "step": 1318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829255663430422e-05,
+      "loss": 0.0697,
+      "step": 1319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.829126213592234e-05,
+      "loss": 0.1058,
+      "step": 1320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828996763754045e-05,
+      "loss": 0.07,
+      "step": 1321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828867313915858e-05,
+      "loss": 0.0584,
+      "step": 1322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828737864077671e-05,
+      "loss": 0.0512,
+      "step": 1323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828608414239483e-05,
+      "loss": 0.0575,
+      "step": 1324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828478964401295e-05,
+      "loss": 0.0782,
+      "step": 1325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828349514563107e-05,
+      "loss": 0.0778,
+      "step": 1326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82822006472492e-05,
+      "loss": 0.0514,
+      "step": 1327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828090614886731e-05,
+      "loss": 0.0936,
+      "step": 1328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827961165048544e-05,
+      "loss": 0.0346,
+      "step": 1329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827831715210357e-05,
+      "loss": 0.058,
+      "step": 1330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827702265372168e-05,
+      "loss": 0.0403,
+      "step": 1331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82757281553398e-05,
+      "loss": 0.0692,
+      "step": 1332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827443365695793e-05,
+      "loss": 0.0455,
+      "step": 1333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827313915857606e-05,
+      "loss": 0.0733,
+      "step": 1334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.827184466019417e-05,
+      "loss": 0.0514,
+      "step": 1335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82705501618123e-05,
+      "loss": 0.0783,
+      "step": 1336
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826925566343043e-05,
+      "loss": 0.0302,
+      "step": 1337
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826796116504855e-05,
+      "loss": 0.2193,
+      "step": 1338
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826666666666668e-05,
+      "loss": 0.0785,
+      "step": 1339
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826537216828479e-05,
+      "loss": 0.0436,
+      "step": 1340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826407766990292e-05,
+      "loss": 0.054,
+      "step": 1341
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826278317152105e-05,
+      "loss": 0.1,
+      "step": 1342
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826148867313917e-05,
+      "loss": 0.035,
+      "step": 1343
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.826019417475728e-05,
+      "loss": 0.0603,
+      "step": 1344
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.82588996763754e-05,
+      "loss": 0.0426,
+      "step": 1345
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825760517799352e-05,
+      "loss": 0.0435,
+      "step": 1346
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825631067961165e-05,
+      "loss": 0.0792,
+      "step": 1347
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825501618122978e-05,
+      "loss": 0.0368,
+      "step": 1348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.82537216828479e-05,
+      "loss": 0.0534,
+      "step": 1349
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825242718446602e-05,
+      "loss": 0.0514,
+      "step": 1350
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.825113268608414e-05,
+      "loss": 0.0542,
+      "step": 1351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824983818770227e-05,
+      "loss": 0.045,
+      "step": 1352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.82485436893204e-05,
+      "loss": 0.0669,
+      "step": 1353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824724919093852e-05,
+      "loss": 0.0992,
+      "step": 1354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824595469255664e-05,
+      "loss": 0.0528,
+      "step": 1355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824466019417476e-05,
+      "loss": 0.047,
+      "step": 1356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824336569579289e-05,
+      "loss": 0.0291,
+      "step": 1357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824207119741102e-05,
+      "loss": 0.0319,
+      "step": 1358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824077669902913e-05,
+      "loss": 0.0505,
+      "step": 1359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823948220064726e-05,
+      "loss": 0.122,
+      "step": 1360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823818770226537e-05,
+      "loss": 0.0817,
+      "step": 1361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.82368932038835e-05,
+      "loss": 0.0312,
+      "step": 1362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823559870550162e-05,
+      "loss": 0.0481,
+      "step": 1363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823430420711975e-05,
+      "loss": 0.0565,
+      "step": 1364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823300970873786e-05,
+      "loss": 0.0506,
+      "step": 1365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823171521035599e-05,
+      "loss": 0.0622,
+      "step": 1366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823042071197412e-05,
+      "loss": 0.0919,
+      "step": 1367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822912621359224e-05,
+      "loss": 0.0381,
+      "step": 1368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822783171521035e-05,
+      "loss": 0.0535,
+      "step": 1369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822653721682848e-05,
+      "loss": 0.0557,
+      "step": 1370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822524271844661e-05,
+      "loss": 0.0607,
+      "step": 1371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822394822006473e-05,
+      "loss": 0.0304,
+      "step": 1372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822265372168286e-05,
+      "loss": 0.0463,
+      "step": 1373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.822135922330097e-05,
+      "loss": 0.0596,
+      "step": 1374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.82200647249191e-05,
+      "loss": 0.0804,
+      "step": 1375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821877022653723e-05,
+      "loss": 0.0555,
+      "step": 1376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821747572815534e-05,
+      "loss": 0.1173,
+      "step": 1377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821618122977347e-05,
+      "loss": 0.0626,
+      "step": 1378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821488673139158e-05,
+      "loss": 0.0453,
+      "step": 1379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.82135922330097e-05,
+      "loss": 0.061,
+      "step": 1380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821229773462783e-05,
+      "loss": 0.0328,
+      "step": 1381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.821100323624596e-05,
+      "loss": 0.0402,
+      "step": 1382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820970873786409e-05,
+      "loss": 0.059,
+      "step": 1383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.82084142394822e-05,
+      "loss": 0.0724,
+      "step": 1384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820711974110033e-05,
+      "loss": 0.0294,
+      "step": 1385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820582524271845e-05,
+      "loss": 0.0445,
+      "step": 1386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820453074433658e-05,
+      "loss": 0.0587,
+      "step": 1387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.82032362459547e-05,
+      "loss": 0.0628,
+      "step": 1388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820194174757282e-05,
+      "loss": 0.045,
+      "step": 1389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.820064724919095e-05,
+      "loss": 0.0447,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819935275080907e-05,
+      "loss": 0.0438,
+      "step": 1391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.81980582524272e-05,
+      "loss": 0.0937,
+      "step": 1392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819676375404531e-05,
+      "loss": 0.0632,
+      "step": 1393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819546925566342e-05,
+      "loss": 0.0398,
+      "step": 1394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819417475728155e-05,
+      "loss": 0.0521,
+      "step": 1395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819288025889968e-05,
+      "loss": 0.037,
+      "step": 1396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.81915857605178e-05,
+      "loss": 0.0738,
+      "step": 1397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.819029126213593e-05,
+      "loss": 0.104,
+      "step": 1398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818899676375404e-05,
+      "loss": 0.0543,
+      "step": 1399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818770226537217e-05,
+      "loss": 0.0835,
+      "step": 1400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.81864077669903e-05,
+      "loss": 0.0747,
+      "step": 1401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818511326860842e-05,
+      "loss": 0.0735,
+      "step": 1402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818381877022654e-05,
+      "loss": 0.0509,
+      "step": 1403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818252427184466e-05,
+      "loss": 0.0372,
+      "step": 1404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.818122977346279e-05,
+      "loss": 0.0394,
+      "step": 1405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817993527508092e-05,
+      "loss": 0.0369,
+      "step": 1406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817864077669904e-05,
+      "loss": 0.0497,
+      "step": 1407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817734627831716e-05,
+      "loss": 0.0291,
+      "step": 1408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817605177993528e-05,
+      "loss": 0.0639,
+      "step": 1409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.81747572815534e-05,
+      "loss": 0.0666,
+      "step": 1410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817346278317152e-05,
+      "loss": 0.0335,
+      "step": 1411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817216828478965e-05,
+      "loss": 0.0945,
+      "step": 1412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.817087378640778e-05,
+      "loss": 0.0456,
+      "step": 1413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816957928802589e-05,
+      "loss": 0.0449,
+      "step": 1414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816828478964401e-05,
+      "loss": 0.062,
+      "step": 1415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816699029126214e-05,
+      "loss": 0.0711,
+      "step": 1416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816569579288027e-05,
+      "loss": 0.057,
+      "step": 1417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816440129449838e-05,
+      "loss": 0.045,
+      "step": 1418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816310679611651e-05,
+      "loss": 0.0542,
+      "step": 1419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816181229773463e-05,
+      "loss": 0.0934,
+      "step": 1420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.816051779935276e-05,
+      "loss": 0.0551,
+      "step": 1421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815922330097089e-05,
+      "loss": 0.0424,
+      "step": 1422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.8157928802589e-05,
+      "loss": 0.0359,
+      "step": 1423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815663430420713e-05,
+      "loss": 0.0673,
+      "step": 1424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815533980582525e-05,
+      "loss": 0.0546,
+      "step": 1425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815404530744337e-05,
+      "loss": 0.1362,
+      "step": 1426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81527508090615e-05,
+      "loss": 0.0514,
+      "step": 1427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.81514563106796e-05,
+      "loss": 0.0405,
+      "step": 1428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.815016181229773e-05,
+      "loss": 0.0283,
+      "step": 1429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814886731391586e-05,
+      "loss": 0.0388,
+      "step": 1430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814757281553399e-05,
+      "loss": 0.0578,
+      "step": 1431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814627831715211e-05,
+      "loss": 0.0559,
+      "step": 1432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814498381877023e-05,
+      "loss": 0.0435,
+      "step": 1433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814368932038835e-05,
+      "loss": 0.0607,
+      "step": 1434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.814239482200648e-05,
+      "loss": 0.0456,
+      "step": 1435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.81411003236246e-05,
+      "loss": 0.0624,
+      "step": 1436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813980582524273e-05,
+      "loss": 0.0952,
+      "step": 1437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813851132686085e-05,
+      "loss": 0.039,
+      "step": 1438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813721682847897e-05,
+      "loss": 0.0391,
+      "step": 1439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.81359223300971e-05,
+      "loss": 0.0521,
+      "step": 1440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813462783171522e-05,
+      "loss": 0.0517,
+      "step": 1441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813333333333334e-05,
+      "loss": 0.0401,
+      "step": 1442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813203883495145e-05,
+      "loss": 0.0458,
+      "step": 1443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.813074433656958e-05,
+      "loss": 0.0433,
+      "step": 1444
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81294498381877e-05,
+      "loss": 0.0565,
+      "step": 1445
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812815533980583e-05,
+      "loss": 0.0821,
+      "step": 1446
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812686084142396e-05,
+      "loss": 0.0732,
+      "step": 1447
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812556634304207e-05,
+      "loss": 0.062,
+      "step": 1448
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.81242718446602e-05,
+      "loss": 0.0702,
+      "step": 1449
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812297734627832e-05,
+      "loss": 0.0565,
+      "step": 1450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812168284789645e-05,
+      "loss": 0.0301,
+      "step": 1451
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.812038834951456e-05,
+      "loss": 0.0406,
+      "step": 1452
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811909385113269e-05,
+      "loss": 0.1379,
+      "step": 1453
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811779935275082e-05,
+      "loss": 0.0553,
+      "step": 1454
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811650485436894e-05,
+      "loss": 0.0477,
+      "step": 1455
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811521035598707e-05,
+      "loss": 0.0556,
+      "step": 1456
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811391585760518e-05,
+      "loss": 0.0615,
+      "step": 1457
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811262135922331e-05,
+      "loss": 0.0333,
+      "step": 1458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811132686084142e-05,
+      "loss": 0.0501,
+      "step": 1459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811003236245955e-05,
+      "loss": 0.0647,
+      "step": 1460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810873786407768e-05,
+      "loss": 0.0452,
+      "step": 1461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810744336569579e-05,
+      "loss": 0.0628,
+      "step": 1462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810614886731391e-05,
+      "loss": 0.074,
+      "step": 1463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810485436893204e-05,
+      "loss": 0.0502,
+      "step": 1464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810355987055017e-05,
+      "loss": 0.0425,
+      "step": 1465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.81022653721683e-05,
+      "loss": 0.0771,
+      "step": 1466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.810097087378641e-05,
+      "loss": 0.0468,
+      "step": 1467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809967637540453e-05,
+      "loss": 0.0564,
+      "step": 1468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809838187702266e-05,
+      "loss": 0.0473,
+      "step": 1469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809708737864079e-05,
+      "loss": 0.0394,
+      "step": 1470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809579288025891e-05,
+      "loss": 0.0418,
+      "step": 1471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809449838187703e-05,
+      "loss": 0.0709,
+      "step": 1472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809320388349515e-05,
+      "loss": 0.0398,
+      "step": 1473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.809190938511328e-05,
+      "loss": 0.0591,
+      "step": 1474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.80906148867314e-05,
+      "loss": 0.0216,
+      "step": 1475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808932038834952e-05,
+      "loss": 0.1273,
+      "step": 1476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808802588996763e-05,
+      "loss": 0.0312,
+      "step": 1477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808673139158576e-05,
+      "loss": 0.0854,
+      "step": 1478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808543689320389e-05,
+      "loss": 0.0932,
+      "step": 1479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808414239482201e-05,
+      "loss": 0.0428,
+      "step": 1480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808284789644014e-05,
+      "loss": 0.0679,
+      "step": 1481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808155339805825e-05,
+      "loss": 0.0532,
+      "step": 1482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.808025889967638e-05,
+      "loss": 0.0471,
+      "step": 1483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80789644012945e-05,
+      "loss": 0.0667,
+      "step": 1484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807766990291263e-05,
+      "loss": 0.0424,
+      "step": 1485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807637540453075e-05,
+      "loss": 0.0659,
+      "step": 1486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807508090614887e-05,
+      "loss": 0.0615,
+      "step": 1487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.8073786407767e-05,
+      "loss": 0.0472,
+      "step": 1488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807249190938512e-05,
+      "loss": 0.0876,
+      "step": 1489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.807119741100325e-05,
+      "loss": 0.0536,
+      "step": 1490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806990291262136e-05,
+      "loss": 0.0554,
+      "step": 1491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806860841423948e-05,
+      "loss": 0.0494,
+      "step": 1492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80673139158576e-05,
+      "loss": 0.0754,
+      "step": 1493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806601941747573e-05,
+      "loss": 0.038,
+      "step": 1494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806472491909386e-05,
+      "loss": 0.048,
+      "step": 1495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806343042071197e-05,
+      "loss": 0.1264,
+      "step": 1496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80621359223301e-05,
+      "loss": 0.0936,
+      "step": 1497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.806084142394822e-05,
+      "loss": 0.0901,
+      "step": 1498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805954692556635e-05,
+      "loss": 0.0743,
+      "step": 1499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805825242718448e-05,
+      "loss": 0.1006,
+      "step": 1500
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.07756045460700989,
+      "eval_mean_accuracy": 0.6345811958558637,
+      "eval_mean_iou": 0.5775972270662748,
+      "eval_overall_accuracy": 0.9742011623382568,
+      "eval_per_category_accuracy": [
+        0.9951330224475959,
+        0.656829640904865,
+        0.4315481676207792,
+        0.4548139524502147
+      ],
+      "eval_per_category_iou": [
+        0.9752981519225001,
+        0.5298292832376184,
+        0.40812493900344177,
+        0.3971365341015387
+      ],
+      "eval_runtime": 188.9787,
+      "eval_samples_per_second": 10.583,
+      "eval_steps_per_second": 1.323,
+      "step": 1500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805695792880259e-05,
+      "loss": 0.0806,
+      "step": 1501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805566343042072e-05,
+      "loss": 0.0827,
+      "step": 1502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805436893203884e-05,
+      "loss": 0.1009,
+      "step": 1503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805307443365697e-05,
+      "loss": 0.0531,
+      "step": 1504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.80517799352751e-05,
+      "loss": 0.0744,
+      "step": 1505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.805048543689321e-05,
+      "loss": 0.0517,
+      "step": 1506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804919093851132e-05,
+      "loss": 0.0989,
+      "step": 1507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804789644012945e-05,
+      "loss": 0.0702,
+      "step": 1508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804660194174758e-05,
+      "loss": 0.0409,
+      "step": 1509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.80453074433657e-05,
+      "loss": 0.0869,
+      "step": 1510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804401294498381e-05,
+      "loss": 0.0545,
+      "step": 1511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804271844660194e-05,
+      "loss": 0.0689,
+      "step": 1512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.804142394822007e-05,
+      "loss": 0.0585,
+      "step": 1513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.80401294498382e-05,
+      "loss": 0.0512,
+      "step": 1514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803883495145632e-05,
+      "loss": 0.1633,
+      "step": 1515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803754045307443e-05,
+      "loss": 0.0688,
+      "step": 1516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803624595469256e-05,
+      "loss": 0.0872,
+      "step": 1517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803495145631069e-05,
+      "loss": 0.0592,
+      "step": 1518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803365695792881e-05,
+      "loss": 0.0716,
+      "step": 1519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803236245954693e-05,
+      "loss": 0.0594,
+      "step": 1520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.803106796116505e-05,
+      "loss": 0.0572,
+      "step": 1521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802977346278318e-05,
+      "loss": 0.0526,
+      "step": 1522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802847896440131e-05,
+      "loss": 0.0602,
+      "step": 1523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802718446601942e-05,
+      "loss": 0.066,
+      "step": 1524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802588996763755e-05,
+      "loss": 0.0482,
+      "step": 1525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802459546925566e-05,
+      "loss": 0.0377,
+      "step": 1526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802330097087379e-05,
+      "loss": 0.0532,
+      "step": 1527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802200647249191e-05,
+      "loss": 0.05,
+      "step": 1528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.802071197411004e-05,
+      "loss": 0.0797,
+      "step": 1529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801941747572817e-05,
+      "loss": 0.0594,
+      "step": 1530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801812297734628e-05,
+      "loss": 0.0914,
+      "step": 1531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80168284789644e-05,
+      "loss": 0.0895,
+      "step": 1532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801553398058253e-05,
+      "loss": 0.1111,
+      "step": 1533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801423948220066e-05,
+      "loss": 0.1089,
+      "step": 1534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801294498381877e-05,
+      "loss": 0.0434,
+      "step": 1535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.80116504854369e-05,
+      "loss": 0.1033,
+      "step": 1536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.801035598705502e-05,
+      "loss": 0.0727,
+      "step": 1537
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800906148867315e-05,
+      "loss": 0.1107,
+      "step": 1538
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800776699029128e-05,
+      "loss": 0.0709,
+      "step": 1539
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800647249190939e-05,
+      "loss": 0.0968,
+      "step": 1540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.80051779935275e-05,
+      "loss": 0.0767,
+      "step": 1541
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800388349514563e-05,
+      "loss": 0.065,
+      "step": 1542
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800258899676376e-05,
+      "loss": 0.08,
+      "step": 1543
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.800129449838188e-05,
+      "loss": 0.0526,
+      "step": 1544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.8e-05,
+      "loss": 0.1482,
+      "step": 1545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799870550161812e-05,
+      "loss": 0.0545,
+      "step": 1546
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799741100323625e-05,
+      "loss": 0.0756,
+      "step": 1547
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799611650485438e-05,
+      "loss": 0.0521,
+      "step": 1548
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.79948220064725e-05,
+      "loss": 0.0298,
+      "step": 1549
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799352750809062e-05,
+      "loss": 0.0322,
+      "step": 1550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799223300970874e-05,
+      "loss": 0.0668,
+      "step": 1551
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.799093851132687e-05,
+      "loss": 0.0387,
+      "step": 1552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.7989644012945e-05,
+      "loss": 0.0522,
+      "step": 1553
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798834951456312e-05,
+      "loss": 0.1572,
+      "step": 1554
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798705501618124e-05,
+      "loss": 0.0553,
+      "step": 1555
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798576051779935e-05,
+      "loss": 0.033,
+      "step": 1556
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798446601941748e-05,
+      "loss": 0.0312,
+      "step": 1557
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.79831715210356e-05,
+      "loss": 0.0532,
+      "step": 1558
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798187702265373e-05,
+      "loss": 0.0524,
+      "step": 1559
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.798058252427184e-05,
+      "loss": 0.0254,
+      "step": 1560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797928802588997e-05,
+      "loss": 0.0423,
+      "step": 1561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.79779935275081e-05,
+      "loss": 0.0726,
+      "step": 1562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797669902912622e-05,
+      "loss": 0.0545,
+      "step": 1563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797540453074435e-05,
+      "loss": 0.049,
+      "step": 1564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797411003236246e-05,
+      "loss": 0.0504,
+      "step": 1565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797281553398059e-05,
+      "loss": 0.0482,
+      "step": 1566
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797152103559871e-05,
+      "loss": 0.0356,
+      "step": 1567
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.797022653721684e-05,
+      "loss": 0.047,
+      "step": 1568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796893203883495e-05,
+      "loss": 0.0355,
+      "step": 1569
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796763754045308e-05,
+      "loss": 0.0367,
+      "step": 1570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.79663430420712e-05,
+      "loss": 0.0405,
+      "step": 1571
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796504854368933e-05,
+      "loss": 0.0503,
+      "step": 1572
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796375404530745e-05,
+      "loss": 0.0573,
+      "step": 1573
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796245954692557e-05,
+      "loss": 0.0593,
+      "step": 1574
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.796116504854369e-05,
+      "loss": 0.0836,
+      "step": 1575
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795987055016181e-05,
+      "loss": 0.0365,
+      "step": 1576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795857605177994e-05,
+      "loss": 0.0672,
+      "step": 1577
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795728155339807e-05,
+      "loss": 0.0623,
+      "step": 1578
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795598705501618e-05,
+      "loss": 0.0573,
+      "step": 1579
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.79546925566343e-05,
+      "loss": 0.066,
+      "step": 1580
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795339805825243e-05,
+      "loss": 0.0682,
+      "step": 1581
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795210355987056e-05,
+      "loss": 0.0505,
+      "step": 1582
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.795080906148869e-05,
+      "loss": 0.06,
+      "step": 1583
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79495145631068e-05,
+      "loss": 0.0395,
+      "step": 1584
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794822006472492e-05,
+      "loss": 0.0569,
+      "step": 1585
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794692556634305e-05,
+      "loss": 0.0269,
+      "step": 1586
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794563106796118e-05,
+      "loss": 0.0723,
+      "step": 1587
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79443365695793e-05,
+      "loss": 0.0304,
+      "step": 1588
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79430420711974e-05,
+      "loss": 0.0385,
+      "step": 1589
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794174757281553e-05,
+      "loss": 0.0435,
+      "step": 1590
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.794045307443366e-05,
+      "loss": 0.0564,
+      "step": 1591
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793915857605178e-05,
+      "loss": 0.0712,
+      "step": 1592
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793786407766991e-05,
+      "loss": 0.0226,
+      "step": 1593
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793656957928802e-05,
+      "loss": 0.072,
+      "step": 1594
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793527508090615e-05,
+      "loss": 0.0392,
+      "step": 1595
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793398058252428e-05,
+      "loss": 0.0596,
+      "step": 1596
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.79326860841424e-05,
+      "loss": 0.0462,
+      "step": 1597
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793139158576053e-05,
+      "loss": 0.0385,
+      "step": 1598
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.793009708737864e-05,
+      "loss": 0.0556,
+      "step": 1599
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792880258899677e-05,
+      "loss": 0.0339,
+      "step": 1600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79275080906149e-05,
+      "loss": 0.0811,
+      "step": 1601
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792621359223302e-05,
+      "loss": 0.0384,
+      "step": 1602
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792491909385114e-05,
+      "loss": 0.0374,
+      "step": 1603
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792362459546926e-05,
+      "loss": 0.0426,
+      "step": 1604
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.792233009708738e-05,
+      "loss": 0.0628,
+      "step": 1605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.79210355987055e-05,
+      "loss": 0.0445,
+      "step": 1606
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791974110032363e-05,
+      "loss": 0.0427,
+      "step": 1607
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791844660194175e-05,
+      "loss": 0.0458,
+      "step": 1608
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791715210355987e-05,
+      "loss": 0.029,
+      "step": 1609
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.7915857605178e-05,
+      "loss": 0.0513,
+      "step": 1610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791456310679612e-05,
+      "loss": 0.0423,
+      "step": 1611
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791326860841425e-05,
+      "loss": 0.0771,
+      "step": 1612
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791197411003236e-05,
+      "loss": 0.0477,
+      "step": 1613
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.791067961165049e-05,
+      "loss": 0.0504,
+      "step": 1614
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790938511326861e-05,
+      "loss": 0.028,
+      "step": 1615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790809061488674e-05,
+      "loss": 0.0482,
+      "step": 1616
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790679611650487e-05,
+      "loss": 0.0279,
+      "step": 1617
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790550161812298e-05,
+      "loss": 0.0554,
+      "step": 1618
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.79042071197411e-05,
+      "loss": 0.0298,
+      "step": 1619
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790291262135923e-05,
+      "loss": 0.0505,
+      "step": 1620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790161812297735e-05,
+      "loss": 0.03,
+      "step": 1621
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.790032362459547e-05,
+      "loss": 0.0474,
+      "step": 1622
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.78990291262136e-05,
+      "loss": 0.0361,
+      "step": 1623
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789773462783171e-05,
+      "loss": 0.0795,
+      "step": 1624
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789644012944984e-05,
+      "loss": 0.0625,
+      "step": 1625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789514563106797e-05,
+      "loss": 0.0532,
+      "step": 1626
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789385113268609e-05,
+      "loss": 0.0268,
+      "step": 1627
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.78925566343042e-05,
+      "loss": 0.0327,
+      "step": 1628
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.789126213592233e-05,
+      "loss": 0.0474,
+      "step": 1629
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788996763754046e-05,
+      "loss": 0.0675,
+      "step": 1630
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788867313915859e-05,
+      "loss": 0.0406,
+      "step": 1631
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788737864077671e-05,
+      "loss": 0.0543,
+      "step": 1632
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788608414239482e-05,
+      "loss": 0.0355,
+      "step": 1633
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788478964401295e-05,
+      "loss": 0.0583,
+      "step": 1634
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788349514563108e-05,
+      "loss": 0.0407,
+      "step": 1635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.78822006472492e-05,
+      "loss": 0.044,
+      "step": 1636
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.788090614886732e-05,
+      "loss": 0.1118,
+      "step": 1637
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787961165048543e-05,
+      "loss": 0.0331,
+      "step": 1638
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787831715210356e-05,
+      "loss": 0.0553,
+      "step": 1639
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787702265372168e-05,
+      "loss": 0.0537,
+      "step": 1640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787572815533981e-05,
+      "loss": 0.0455,
+      "step": 1641
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787443365695794e-05,
+      "loss": 0.0561,
+      "step": 1642
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787313915857605e-05,
+      "loss": 0.0585,
+      "step": 1643
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.787184466019418e-05,
+      "loss": 0.1284,
+      "step": 1644
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.78705501618123e-05,
+      "loss": 0.0277,
+      "step": 1645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786925566343043e-05,
+      "loss": 0.0296,
+      "step": 1646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786796116504856e-05,
+      "loss": 0.0726,
+      "step": 1647
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786666666666667e-05,
+      "loss": 0.0801,
+      "step": 1648
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78653721682848e-05,
+      "loss": 0.0603,
+      "step": 1649
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786407766990292e-05,
+      "loss": 0.048,
+      "step": 1650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786278317152105e-05,
+      "loss": 0.0543,
+      "step": 1651
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786148867313916e-05,
+      "loss": 0.0484,
+      "step": 1652
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.786019417475729e-05,
+      "loss": 0.0541,
+      "step": 1653
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78588996763754e-05,
+      "loss": 0.0418,
+      "step": 1654
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785760517799353e-05,
+      "loss": 0.0428,
+      "step": 1655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785631067961165e-05,
+      "loss": 0.0525,
+      "step": 1656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785501618122978e-05,
+      "loss": 0.0303,
+      "step": 1657
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.78537216828479e-05,
+      "loss": 0.0431,
+      "step": 1658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785242718446602e-05,
+      "loss": 0.0525,
+      "step": 1659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.785113268608415e-05,
+      "loss": 0.057,
+      "step": 1660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784983818770227e-05,
+      "loss": 0.0464,
+      "step": 1661
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784854368932039e-05,
+      "loss": 0.0337,
+      "step": 1662
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784724919093851e-05,
+      "loss": 0.0438,
+      "step": 1663
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784595469255664e-05,
+      "loss": 0.0521,
+      "step": 1664
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784466019417477e-05,
+      "loss": 0.0451,
+      "step": 1665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.78433656957929e-05,
+      "loss": 0.0549,
+      "step": 1666
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.7842071197411e-05,
+      "loss": 0.061,
+      "step": 1667
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.784077669902913e-05,
+      "loss": 0.0421,
+      "step": 1668
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783948220064726e-05,
+      "loss": 0.0485,
+      "step": 1669
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783818770226537e-05,
+      "loss": 0.0422,
+      "step": 1670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.78368932038835e-05,
+      "loss": 0.0877,
+      "step": 1671
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783559870550161e-05,
+      "loss": 0.0345,
+      "step": 1672
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783430420711974e-05,
+      "loss": 0.0683,
+      "step": 1673
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783300970873787e-05,
+      "loss": 0.0371,
+      "step": 1674
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783171521035599e-05,
+      "loss": 0.1149,
+      "step": 1675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.783042071197412e-05,
+      "loss": 0.0587,
+      "step": 1676
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782912621359223e-05,
+      "loss": 0.0262,
+      "step": 1677
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782783171521036e-05,
+      "loss": 0.0425,
+      "step": 1678
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782653721682849e-05,
+      "loss": 0.0447,
+      "step": 1679
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782524271844661e-05,
+      "loss": 0.0496,
+      "step": 1680
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782394822006474e-05,
+      "loss": 0.0339,
+      "step": 1681
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782265372168285e-05,
+      "loss": 0.0321,
+      "step": 1682
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.782135922330098e-05,
+      "loss": 0.0936,
+      "step": 1683
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.78200647249191e-05,
+      "loss": 0.0824,
+      "step": 1684
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781877022653723e-05,
+      "loss": 0.0655,
+      "step": 1685
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781747572815534e-05,
+      "loss": 0.0529,
+      "step": 1686
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781618122977346e-05,
+      "loss": 0.0371,
+      "step": 1687
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781488673139158e-05,
+      "loss": 0.0392,
+      "step": 1688
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781359223300971e-05,
+      "loss": 0.0398,
+      "step": 1689
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781229773462784e-05,
+      "loss": 0.0988,
+      "step": 1690
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.781100323624596e-05,
+      "loss": 0.0537,
+      "step": 1691
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780970873786408e-05,
+      "loss": 0.0642,
+      "step": 1692
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.78084142394822e-05,
+      "loss": 0.0806,
+      "step": 1693
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780711974110033e-05,
+      "loss": 0.0501,
+      "step": 1694
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780582524271846e-05,
+      "loss": 0.068,
+      "step": 1695
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780453074433657e-05,
+      "loss": 0.0701,
+      "step": 1696
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.78032362459547e-05,
+      "loss": 0.0464,
+      "step": 1697
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780194174757282e-05,
+      "loss": 0.0481,
+      "step": 1698
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.780064724919095e-05,
+      "loss": 0.0554,
+      "step": 1699
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779935275080908e-05,
+      "loss": 0.0992,
+      "step": 1700
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779805825242719e-05,
+      "loss": 0.0316,
+      "step": 1701
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779676375404532e-05,
+      "loss": 0.0605,
+      "step": 1702
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779546925566343e-05,
+      "loss": 0.0784,
+      "step": 1703
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779417475728155e-05,
+      "loss": 0.0386,
+      "step": 1704
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779288025889968e-05,
+      "loss": 0.081,
+      "step": 1705
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.77915857605178e-05,
+      "loss": 0.0566,
+      "step": 1706
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.779029126213592e-05,
+      "loss": 0.0499,
+      "step": 1707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778899676375405e-05,
+      "loss": 0.0472,
+      "step": 1708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778770226537217e-05,
+      "loss": 0.0595,
+      "step": 1709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.77864077669903e-05,
+      "loss": 0.0337,
+      "step": 1710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778511326860841e-05,
+      "loss": 0.1299,
+      "step": 1711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778381877022654e-05,
+      "loss": 0.032,
+      "step": 1712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.778252427184467e-05,
+      "loss": 0.0417,
+      "step": 1713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.77812297734628e-05,
+      "loss": 0.0706,
+      "step": 1714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777993527508092e-05,
+      "loss": 0.0332,
+      "step": 1715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777864077669903e-05,
+      "loss": 0.0713,
+      "step": 1716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777734627831716e-05,
+      "loss": 0.0555,
+      "step": 1717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777605177993529e-05,
+      "loss": 0.0344,
+      "step": 1718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.77747572815534e-05,
+      "loss": 0.0356,
+      "step": 1719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777346278317153e-05,
+      "loss": 0.0688,
+      "step": 1720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777216828478964e-05,
+      "loss": 0.0506,
+      "step": 1721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.777087378640777e-05,
+      "loss": 0.0618,
+      "step": 1722
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776957928802589e-05,
+      "loss": 0.0702,
+      "step": 1723
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776828478964402e-05,
+      "loss": 0.0588,
+      "step": 1724
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776699029126215e-05,
+      "loss": 0.042,
+      "step": 1725
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776569579288026e-05,
+      "loss": 0.086,
+      "step": 1726
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776440129449839e-05,
+      "loss": 0.0511,
+      "step": 1727
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776310679611651e-05,
+      "loss": 0.0579,
+      "step": 1728
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776181229773464e-05,
+      "loss": 0.049,
+      "step": 1729
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.776051779935275e-05,
+      "loss": 0.0395,
+      "step": 1730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775922330097088e-05,
+      "loss": 0.039,
+      "step": 1731
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.7757928802589e-05,
+      "loss": 0.0259,
+      "step": 1732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775663430420713e-05,
+      "loss": 0.051,
+      "step": 1733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775533980582526e-05,
+      "loss": 0.0465,
+      "step": 1734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775404530744337e-05,
+      "loss": 0.0658,
+      "step": 1735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775275080906148e-05,
+      "loss": 0.0594,
+      "step": 1736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775145631067961e-05,
+      "loss": 0.0391,
+      "step": 1737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.775016181229774e-05,
+      "loss": 0.0341,
+      "step": 1738
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774886731391586e-05,
+      "loss": 0.0606,
+      "step": 1739
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774757281553398e-05,
+      "loss": 0.0797,
+      "step": 1740
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.77462783171521e-05,
+      "loss": 0.0449,
+      "step": 1741
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774498381877023e-05,
+      "loss": 0.0475,
+      "step": 1742
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774368932038836e-05,
+      "loss": 0.0436,
+      "step": 1743
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.774239482200648e-05,
+      "loss": 0.05,
+      "step": 1744
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.77411003236246e-05,
+      "loss": 0.0616,
+      "step": 1745
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773980582524272e-05,
+      "loss": 0.1009,
+      "step": 1746
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773851132686085e-05,
+      "loss": 0.0474,
+      "step": 1747
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773721682847898e-05,
+      "loss": 0.0509,
+      "step": 1748
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.77359223300971e-05,
+      "loss": 0.0703,
+      "step": 1749
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773462783171522e-05,
+      "loss": 0.0646,
+      "step": 1750
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773333333333334e-05,
+      "loss": 0.0464,
+      "step": 1751
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773203883495145e-05,
+      "loss": 0.0215,
+      "step": 1752
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.773074433656958e-05,
+      "loss": 0.0359,
+      "step": 1753
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772944983818771e-05,
+      "loss": 0.0432,
+      "step": 1754
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772815533980582e-05,
+      "loss": 0.0788,
+      "step": 1755
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772686084142395e-05,
+      "loss": 0.0395,
+      "step": 1756
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772556634304207e-05,
+      "loss": 0.025,
+      "step": 1757
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.77242718446602e-05,
+      "loss": 0.0788,
+      "step": 1758
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772297734627833e-05,
+      "loss": 0.0574,
+      "step": 1759
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772168284789644e-05,
+      "loss": 0.0582,
+      "step": 1760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.772038834951457e-05,
+      "loss": 0.0344,
+      "step": 1761
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.77190938511327e-05,
+      "loss": 0.03,
+      "step": 1762
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771779935275082e-05,
+      "loss": 0.0828,
+      "step": 1763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771650485436893e-05,
+      "loss": 0.0659,
+      "step": 1764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771521035598706e-05,
+      "loss": 0.0387,
+      "step": 1765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771391585760519e-05,
+      "loss": 0.072,
+      "step": 1766
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771262135922331e-05,
+      "loss": 0.0236,
+      "step": 1767
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771132686084143e-05,
+      "loss": 0.0705,
+      "step": 1768
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.771003236245955e-05,
+      "loss": 0.0722,
+      "step": 1769
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770873786407767e-05,
+      "loss": 0.0322,
+      "step": 1770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770744336569579e-05,
+      "loss": 0.024,
+      "step": 1771
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770614886731392e-05,
+      "loss": 0.0627,
+      "step": 1772
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770485436893205e-05,
+      "loss": 0.1333,
+      "step": 1773
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770355987055017e-05,
+      "loss": 0.0248,
+      "step": 1774
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770226537216828e-05,
+      "loss": 0.0793,
+      "step": 1775
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.770097087378641e-05,
+      "loss": 0.051,
+      "step": 1776
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769967637540454e-05,
+      "loss": 0.0399,
+      "step": 1777
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769838187702266e-05,
+      "loss": 0.0385,
+      "step": 1778
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769708737864078e-05,
+      "loss": 0.0401,
+      "step": 1779
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.76957928802589e-05,
+      "loss": 0.0373,
+      "step": 1780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769449838187703e-05,
+      "loss": 0.0627,
+      "step": 1781
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769320388349516e-05,
+      "loss": 0.0288,
+      "step": 1782
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.769190938511328e-05,
+      "loss": 0.0754,
+      "step": 1783
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.76906148867314e-05,
+      "loss": 0.0613,
+      "step": 1784
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768932038834951e-05,
+      "loss": 0.0544,
+      "step": 1785
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768802588996764e-05,
+      "loss": 0.043,
+      "step": 1786
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768673139158576e-05,
+      "loss": 0.0507,
+      "step": 1787
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768543689320389e-05,
+      "loss": 0.0396,
+      "step": 1788
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.7684142394822e-05,
+      "loss": 0.0815,
+      "step": 1789
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768284789644013e-05,
+      "loss": 0.03,
+      "step": 1790
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768155339805826e-05,
+      "loss": 0.0351,
+      "step": 1791
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.768025889967638e-05,
+      "loss": 0.0223,
+      "step": 1792
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767896440129451e-05,
+      "loss": 0.0308,
+      "step": 1793
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767766990291262e-05,
+      "loss": 0.065,
+      "step": 1794
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767637540453075e-05,
+      "loss": 0.0425,
+      "step": 1795
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767508090614888e-05,
+      "loss": 0.0491,
+      "step": 1796
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.7673786407767e-05,
+      "loss": 0.0767,
+      "step": 1797
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767249190938513e-05,
+      "loss": 0.0416,
+      "step": 1798
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.767119741100324e-05,
+      "loss": 0.0376,
+      "step": 1799
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766990291262137e-05,
+      "loss": 0.0389,
+      "step": 1800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766860841423948e-05,
+      "loss": 0.0763,
+      "step": 1801
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766731391585761e-05,
+      "loss": 0.0719,
+      "step": 1802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766601941747573e-05,
+      "loss": 0.0693,
+      "step": 1803
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766472491909385e-05,
+      "loss": 0.0255,
+      "step": 1804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766343042071197e-05,
+      "loss": 0.0331,
+      "step": 1805
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.76621359223301e-05,
+      "loss": 0.0415,
+      "step": 1806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.766084142394823e-05,
+      "loss": 0.0452,
+      "step": 1807
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765954692556635e-05,
+      "loss": 0.0354,
+      "step": 1808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765825242718447e-05,
+      "loss": 0.0651,
+      "step": 1809
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.76569579288026e-05,
+      "loss": 0.0601,
+      "step": 1810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765566343042072e-05,
+      "loss": 0.0546,
+      "step": 1811
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765436893203885e-05,
+      "loss": 0.0496,
+      "step": 1812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765307443365696e-05,
+      "loss": 0.06,
+      "step": 1813
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765177993527509e-05,
+      "loss": 0.0551,
+      "step": 1814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.765048543689321e-05,
+      "loss": 0.0446,
+      "step": 1815
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764919093851134e-05,
+      "loss": 0.0774,
+      "step": 1816
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764789644012945e-05,
+      "loss": 0.0674,
+      "step": 1817
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764660194174758e-05,
+      "loss": 0.0496,
+      "step": 1818
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764530744336569e-05,
+      "loss": 0.0484,
+      "step": 1819
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764401294498382e-05,
+      "loss": 0.0774,
+      "step": 1820
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764271844660195e-05,
+      "loss": 0.0716,
+      "step": 1821
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764142394822007e-05,
+      "loss": 0.0551,
+      "step": 1822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.764012944983818e-05,
+      "loss": 0.0511,
+      "step": 1823
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763883495145631e-05,
+      "loss": 0.0685,
+      "step": 1824
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763754045307444e-05,
+      "loss": 0.0544,
+      "step": 1825
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763624595469256e-05,
+      "loss": 0.0539,
+      "step": 1826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763495145631069e-05,
+      "loss": 0.0563,
+      "step": 1827
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.76336569579288e-05,
+      "loss": 0.0649,
+      "step": 1828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763236245954693e-05,
+      "loss": 0.0485,
+      "step": 1829
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.763106796116506e-05,
+      "loss": 0.036,
+      "step": 1830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762977346278318e-05,
+      "loss": 0.0592,
+      "step": 1831
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762847896440131e-05,
+      "loss": 0.059,
+      "step": 1832
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762718446601941e-05,
+      "loss": 0.0307,
+      "step": 1833
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762588996763754e-05,
+      "loss": 0.0444,
+      "step": 1834
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762459546925566e-05,
+      "loss": 0.0367,
+      "step": 1835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762330097087379e-05,
+      "loss": 0.0387,
+      "step": 1836
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762200647249192e-05,
+      "loss": 0.0465,
+      "step": 1837
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.762071197411003e-05,
+      "loss": 0.0683,
+      "step": 1838
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761941747572816e-05,
+      "loss": 0.0476,
+      "step": 1839
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761812297734628e-05,
+      "loss": 0.0566,
+      "step": 1840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761682847896441e-05,
+      "loss": 0.0331,
+      "step": 1841
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761553398058254e-05,
+      "loss": 0.0348,
+      "step": 1842
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761423948220065e-05,
+      "loss": 0.0752,
+      "step": 1843
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761294498381878e-05,
+      "loss": 0.0449,
+      "step": 1844
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.76116504854369e-05,
+      "loss": 0.0932,
+      "step": 1845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.761035598705503e-05,
+      "loss": 0.0631,
+      "step": 1846
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760906148867314e-05,
+      "loss": 0.036,
+      "step": 1847
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760776699029127e-05,
+      "loss": 0.0351,
+      "step": 1848
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760647249190938e-05,
+      "loss": 0.049,
+      "step": 1849
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760517799352751e-05,
+      "loss": 0.0667,
+      "step": 1850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760388349514563e-05,
+      "loss": 0.0417,
+      "step": 1851
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760258899676376e-05,
+      "loss": 0.1036,
+      "step": 1852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.760129449838187e-05,
+      "loss": 0.0541,
+      "step": 1853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.76e-05,
+      "loss": 0.0736,
+      "step": 1854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759870550161813e-05,
+      "loss": 0.0353,
+      "step": 1855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759741100323625e-05,
+      "loss": 0.0496,
+      "step": 1856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759611650485437e-05,
+      "loss": 0.0829,
+      "step": 1857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.75948220064725e-05,
+      "loss": 0.0415,
+      "step": 1858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759352750809062e-05,
+      "loss": 0.0291,
+      "step": 1859
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759223300970875e-05,
+      "loss": 0.0596,
+      "step": 1860
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.759093851132687e-05,
+      "loss": 0.1418,
+      "step": 1861
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758964401294499e-05,
+      "loss": 0.0424,
+      "step": 1862
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758834951456311e-05,
+      "loss": 0.0505,
+      "step": 1863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758705501618124e-05,
+      "loss": 0.0412,
+      "step": 1864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758576051779937e-05,
+      "loss": 0.0896,
+      "step": 1865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758446601941748e-05,
+      "loss": 0.0278,
+      "step": 1866
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.75831715210356e-05,
+      "loss": 0.0654,
+      "step": 1867
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758187702265372e-05,
+      "loss": 0.0684,
+      "step": 1868
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.758058252427185e-05,
+      "loss": 0.0589,
+      "step": 1869
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757928802588997e-05,
+      "loss": 0.038,
+      "step": 1870
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.75779935275081e-05,
+      "loss": 0.0345,
+      "step": 1871
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757669902912621e-05,
+      "loss": 0.042,
+      "step": 1872
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757540453074434e-05,
+      "loss": 0.0734,
+      "step": 1873
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757411003236246e-05,
+      "loss": 0.0389,
+      "step": 1874
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757281553398059e-05,
+      "loss": 0.0499,
+      "step": 1875
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757152103559872e-05,
+      "loss": 0.0329,
+      "step": 1876
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.757022653721683e-05,
+      "loss": 0.0606,
+      "step": 1877
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756893203883496e-05,
+      "loss": 0.0535,
+      "step": 1878
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756763754045308e-05,
+      "loss": 0.0594,
+      "step": 1879
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756634304207121e-05,
+      "loss": 0.0711,
+      "step": 1880
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756504854368932e-05,
+      "loss": 0.0358,
+      "step": 1881
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756375404530744e-05,
+      "loss": 0.0242,
+      "step": 1882
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756245954692556e-05,
+      "loss": 0.0321,
+      "step": 1883
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.756116504854369e-05,
+      "loss": 0.0516,
+      "step": 1884
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755987055016182e-05,
+      "loss": 0.0358,
+      "step": 1885
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755857605177994e-05,
+      "loss": 0.0391,
+      "step": 1886
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755728155339806e-05,
+      "loss": 0.0407,
+      "step": 1887
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755598705501618e-05,
+      "loss": 0.034,
+      "step": 1888
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755469255663431e-05,
+      "loss": 0.0294,
+      "step": 1889
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755339805825244e-05,
+      "loss": 0.0595,
+      "step": 1890
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755210355987056e-05,
+      "loss": 0.0745,
+      "step": 1891
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.755080906148868e-05,
+      "loss": 0.0453,
+      "step": 1892
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.75495145631068e-05,
+      "loss": 0.0554,
+      "step": 1893
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754822006472493e-05,
+      "loss": 0.0299,
+      "step": 1894
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754692556634306e-05,
+      "loss": 0.0339,
+      "step": 1895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754563106796117e-05,
+      "loss": 0.0714,
+      "step": 1896
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.75443365695793e-05,
+      "loss": 0.0397,
+      "step": 1897
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754304207119741e-05,
+      "loss": 0.0768,
+      "step": 1898
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754174757281553e-05,
+      "loss": 0.0367,
+      "step": 1899
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.754045307443366e-05,
+      "loss": 0.0541,
+      "step": 1900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753915857605179e-05,
+      "loss": 0.0583,
+      "step": 1901
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.75378640776699e-05,
+      "loss": 0.0357,
+      "step": 1902
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753656957928803e-05,
+      "loss": 0.0442,
+      "step": 1903
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753527508090615e-05,
+      "loss": 0.0721,
+      "step": 1904
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753398058252428e-05,
+      "loss": 0.0618,
+      "step": 1905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.75326860841424e-05,
+      "loss": 0.0793,
+      "step": 1906
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753139158576052e-05,
+      "loss": 0.0689,
+      "step": 1907
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.753009708737865e-05,
+      "loss": 0.0272,
+      "step": 1908
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752880258899677e-05,
+      "loss": 0.0252,
+      "step": 1909
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.75275080906149e-05,
+      "loss": 0.0491,
+      "step": 1910
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752621359223301e-05,
+      "loss": 0.0966,
+      "step": 1911
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752491909385114e-05,
+      "loss": 0.0404,
+      "step": 1912
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752362459546927e-05,
+      "loss": 0.0394,
+      "step": 1913
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.752233009708739e-05,
+      "loss": 0.0345,
+      "step": 1914
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.75210355987055e-05,
+      "loss": 0.0525,
+      "step": 1915
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751974110032362e-05,
+      "loss": 0.0507,
+      "step": 1916
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751844660194175e-05,
+      "loss": 0.051,
+      "step": 1917
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751715210355987e-05,
+      "loss": 0.0423,
+      "step": 1918
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.7515857605178e-05,
+      "loss": 0.0355,
+      "step": 1919
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751456310679612e-05,
+      "loss": 0.0381,
+      "step": 1920
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751326860841424e-05,
+      "loss": 0.0226,
+      "step": 1921
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751197411003236e-05,
+      "loss": 0.0414,
+      "step": 1922
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.751067961165049e-05,
+      "loss": 0.0431,
+      "step": 1923
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750938511326862e-05,
+      "loss": 0.0419,
+      "step": 1924
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750809061488674e-05,
+      "loss": 0.0318,
+      "step": 1925
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750679611650486e-05,
+      "loss": 0.0448,
+      "step": 1926
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750550161812298e-05,
+      "loss": 0.0664,
+      "step": 1927
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750420711974111e-05,
+      "loss": 0.052,
+      "step": 1928
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750291262135924e-05,
+      "loss": 0.0564,
+      "step": 1929
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750161812297735e-05,
+      "loss": 0.084,
+      "step": 1930
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.750032362459546e-05,
+      "loss": 0.0334,
+      "step": 1931
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749902912621359e-05,
+      "loss": 0.104,
+      "step": 1932
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749773462783172e-05,
+      "loss": 0.0297,
+      "step": 1933
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749644012944984e-05,
+      "loss": 0.0651,
+      "step": 1934
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749514563106797e-05,
+      "loss": 0.0515,
+      "step": 1935
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749385113268608e-05,
+      "loss": 0.0529,
+      "step": 1936
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749255663430421e-05,
+      "loss": 0.0472,
+      "step": 1937
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.749126213592234e-05,
+      "loss": 0.0723,
+      "step": 1938
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748996763754046e-05,
+      "loss": 0.0346,
+      "step": 1939
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748867313915858e-05,
+      "loss": 0.0581,
+      "step": 1940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.74873786407767e-05,
+      "loss": 0.0455,
+      "step": 1941
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748608414239483e-05,
+      "loss": 0.0492,
+      "step": 1942
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748478964401296e-05,
+      "loss": 0.0544,
+      "step": 1943
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748349514563108e-05,
+      "loss": 0.0519,
+      "step": 1944
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.74822006472492e-05,
+      "loss": 0.0679,
+      "step": 1945
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.748090614886732e-05,
+      "loss": 0.0392,
+      "step": 1946
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747961165048543e-05,
+      "loss": 0.0492,
+      "step": 1947
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747831715210356e-05,
+      "loss": 0.0599,
+      "step": 1948
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747702265372169e-05,
+      "loss": 0.0545,
+      "step": 1949
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.74757281553398e-05,
+      "loss": 0.0647,
+      "step": 1950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747443365695793e-05,
+      "loss": 0.0323,
+      "step": 1951
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747313915857605e-05,
+      "loss": 0.046,
+      "step": 1952
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747184466019418e-05,
+      "loss": 0.0351,
+      "step": 1953
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.747055016181231e-05,
+      "loss": 0.0306,
+      "step": 1954
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746925566343042e-05,
+      "loss": 0.0477,
+      "step": 1955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746796116504855e-05,
+      "loss": 0.0281,
+      "step": 1956
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746666666666667e-05,
+      "loss": 0.0581,
+      "step": 1957
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.74653721682848e-05,
+      "loss": 0.0435,
+      "step": 1958
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746407766990293e-05,
+      "loss": 0.0742,
+      "step": 1959
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746278317152104e-05,
+      "loss": 0.0607,
+      "step": 1960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746148867313917e-05,
+      "loss": 0.0617,
+      "step": 1961
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.746019417475729e-05,
+      "loss": 0.0869,
+      "step": 1962
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745889967637542e-05,
+      "loss": 0.0334,
+      "step": 1963
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745760517799353e-05,
+      "loss": 0.0542,
+      "step": 1964
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745631067961165e-05,
+      "loss": 0.0256,
+      "step": 1965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745501618122977e-05,
+      "loss": 0.0458,
+      "step": 1966
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.74537216828479e-05,
+      "loss": 0.0401,
+      "step": 1967
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745242718446602e-05,
+      "loss": 0.0727,
+      "step": 1968
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.745113268608415e-05,
+      "loss": 0.0354,
+      "step": 1969
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744983818770226e-05,
+      "loss": 0.0444,
+      "step": 1970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744854368932039e-05,
+      "loss": 0.0641,
+      "step": 1971
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744724919093852e-05,
+      "loss": 0.0693,
+      "step": 1972
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744595469255664e-05,
+      "loss": 0.0652,
+      "step": 1973
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744466019417476e-05,
+      "loss": 0.0801,
+      "step": 1974
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744336569579288e-05,
+      "loss": 0.0415,
+      "step": 1975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744207119741101e-05,
+      "loss": 0.0574,
+      "step": 1976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.744077669902914e-05,
+      "loss": 0.0548,
+      "step": 1977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743948220064726e-05,
+      "loss": 0.0634,
+      "step": 1978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743818770226538e-05,
+      "loss": 0.0485,
+      "step": 1979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743689320388349e-05,
+      "loss": 0.0817,
+      "step": 1980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743559870550162e-05,
+      "loss": 0.0431,
+      "step": 1981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743430420711974e-05,
+      "loss": 0.0441,
+      "step": 1982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743300970873787e-05,
+      "loss": 0.0417,
+      "step": 1983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.7431715210356e-05,
+      "loss": 0.0383,
+      "step": 1984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.743042071197411e-05,
+      "loss": 0.0359,
+      "step": 1985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742912621359224e-05,
+      "loss": 0.0326,
+      "step": 1986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742783171521036e-05,
+      "loss": 0.0342,
+      "step": 1987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742653721682849e-05,
+      "loss": 0.0549,
+      "step": 1988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.74252427184466e-05,
+      "loss": 0.0673,
+      "step": 1989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742394822006473e-05,
+      "loss": 0.0616,
+      "step": 1990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742265372168286e-05,
+      "loss": 0.0678,
+      "step": 1991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742135922330098e-05,
+      "loss": 0.0487,
+      "step": 1992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.742006472491911e-05,
+      "loss": 0.0348,
+      "step": 1993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741877022653722e-05,
+      "loss": 0.0583,
+      "step": 1994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741747572815535e-05,
+      "loss": 0.0389,
+      "step": 1995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741618122977346e-05,
+      "loss": 0.0518,
+      "step": 1996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741488673139159e-05,
+      "loss": 0.0396,
+      "step": 1997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741359223300971e-05,
+      "loss": 0.056,
+      "step": 1998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741229773462783e-05,
+      "loss": 0.0395,
+      "step": 1999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.741100323624595e-05,
+      "loss": 0.0513,
+      "step": 2000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740970873786408e-05,
+      "loss": 0.0321,
+      "step": 2001
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740841423948221e-05,
+      "loss": 0.0393,
+      "step": 2002
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740711974110033e-05,
+      "loss": 0.0394,
+      "step": 2003
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740582524271845e-05,
+      "loss": 0.0308,
+      "step": 2004
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740453074433657e-05,
+      "loss": 0.0611,
+      "step": 2005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.74032362459547e-05,
+      "loss": 0.054,
+      "step": 2006
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740194174757283e-05,
+      "loss": 0.0345,
+      "step": 2007
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.740064724919095e-05,
+      "loss": 0.015,
+      "step": 2008
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739935275080907e-05,
+      "loss": 0.0295,
+      "step": 2009
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739805825242719e-05,
+      "loss": 0.049,
+      "step": 2010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739676375404532e-05,
+      "loss": 0.0744,
+      "step": 2011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739546925566343e-05,
+      "loss": 0.0366,
+      "step": 2012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739417475728156e-05,
+      "loss": 0.033,
+      "step": 2013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739288025889967e-05,
+      "loss": 0.0538,
+      "step": 2014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.73915857605178e-05,
+      "loss": 0.0322,
+      "step": 2015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.739029126213592e-05,
+      "loss": 0.0365,
+      "step": 2016
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738899676375405e-05,
+      "loss": 0.0499,
+      "step": 2017
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738770226537218e-05,
+      "loss": 0.0439,
+      "step": 2018
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738640776699029e-05,
+      "loss": 0.0533,
+      "step": 2019
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738511326860842e-05,
+      "loss": 0.0414,
+      "step": 2020
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738381877022654e-05,
+      "loss": 0.081,
+      "step": 2021
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738252427184467e-05,
+      "loss": 0.0494,
+      "step": 2022
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.738122977346278e-05,
+      "loss": 0.0409,
+      "step": 2023
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737993527508091e-05,
+      "loss": 0.0802,
+      "step": 2024
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737864077669904e-05,
+      "loss": 0.0548,
+      "step": 2025
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737734627831716e-05,
+      "loss": 0.0703,
+      "step": 2026
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737605177993529e-05,
+      "loss": 0.0605,
+      "step": 2027
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.73747572815534e-05,
+      "loss": 0.0491,
+      "step": 2028
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737346278317152e-05,
+      "loss": 0.0359,
+      "step": 2029
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737216828478964e-05,
+      "loss": 0.057,
+      "step": 2030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.737087378640777e-05,
+      "loss": 0.0776,
+      "step": 2031
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.73695792880259e-05,
+      "loss": 0.051,
+      "step": 2032
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736828478964401e-05,
+      "loss": 0.0464,
+      "step": 2033
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736699029126214e-05,
+      "loss": 0.0403,
+      "step": 2034
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736569579288026e-05,
+      "loss": 0.0569,
+      "step": 2035
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736440129449839e-05,
+      "loss": 0.0437,
+      "step": 2036
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736310679611652e-05,
+      "loss": 0.0616,
+      "step": 2037
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736181229773463e-05,
+      "loss": 0.0576,
+      "step": 2038
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.736051779935276e-05,
+      "loss": 0.0484,
+      "step": 2039
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735922330097088e-05,
+      "loss": 0.0823,
+      "step": 2040
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735792880258901e-05,
+      "loss": 0.0394,
+      "step": 2041
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735663430420713e-05,
+      "loss": 0.0486,
+      "step": 2042
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735533980582525e-05,
+      "loss": 0.0385,
+      "step": 2043
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735404530744337e-05,
+      "loss": 0.044,
+      "step": 2044
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735275080906149e-05,
+      "loss": 0.0506,
+      "step": 2045
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735145631067961e-05,
+      "loss": 0.0597,
+      "step": 2046
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.735016181229774e-05,
+      "loss": 0.0609,
+      "step": 2047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734886731391585e-05,
+      "loss": 0.0597,
+      "step": 2048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734757281553398e-05,
+      "loss": 0.0646,
+      "step": 2049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734627831715211e-05,
+      "loss": 0.0497,
+      "step": 2050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734498381877023e-05,
+      "loss": 0.0305,
+      "step": 2051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734368932038836e-05,
+      "loss": 0.0465,
+      "step": 2052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.734239482200647e-05,
+      "loss": 0.0386,
+      "step": 2053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.73411003236246e-05,
+      "loss": 0.0279,
+      "step": 2054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733980582524273e-05,
+      "loss": 0.0612,
+      "step": 2055
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733851132686085e-05,
+      "loss": 0.0281,
+      "step": 2056
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733721682847897e-05,
+      "loss": 0.0328,
+      "step": 2057
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733592233009709e-05,
+      "loss": 0.0329,
+      "step": 2058
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733462783171522e-05,
+      "loss": 0.0399,
+      "step": 2059
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733333333333335e-05,
+      "loss": 0.0435,
+      "step": 2060
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733203883495146e-05,
+      "loss": 0.0749,
+      "step": 2061
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.733074433656959e-05,
+      "loss": 0.067,
+      "step": 2062
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.73294498381877e-05,
+      "loss": 0.0364,
+      "step": 2063
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732815533980582e-05,
+      "loss": 0.0529,
+      "step": 2064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732686084142395e-05,
+      "loss": 0.0902,
+      "step": 2065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732556634304208e-05,
+      "loss": 0.0403,
+      "step": 2066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732427184466019e-05,
+      "loss": 0.0537,
+      "step": 2067
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732297734627832e-05,
+      "loss": 0.0708,
+      "step": 2068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732168284789644e-05,
+      "loss": 0.0376,
+      "step": 2069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.732038834951457e-05,
+      "loss": 0.0464,
+      "step": 2070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.73190938511327e-05,
+      "loss": 0.066,
+      "step": 2071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731779935275081e-05,
+      "loss": 0.0368,
+      "step": 2072
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731650485436894e-05,
+      "loss": 0.0878,
+      "step": 2073
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731521035598706e-05,
+      "loss": 0.0759,
+      "step": 2074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731391585760519e-05,
+      "loss": 0.0394,
+      "step": 2075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731262135922332e-05,
+      "loss": 0.0307,
+      "step": 2076
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731132686084143e-05,
+      "loss": 0.0395,
+      "step": 2077
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.731003236245954e-05,
+      "loss": 0.0316,
+      "step": 2078
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730873786407767e-05,
+      "loss": 0.0603,
+      "step": 2079
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.73074433656958e-05,
+      "loss": 0.0392,
+      "step": 2080
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730614886731392e-05,
+      "loss": 0.0564,
+      "step": 2081
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730485436893204e-05,
+      "loss": 0.0694,
+      "step": 2082
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730355987055016e-05,
+      "loss": 0.0698,
+      "step": 2083
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730226537216829e-05,
+      "loss": 0.0388,
+      "step": 2084
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730097087378642e-05,
+      "loss": 0.027,
+      "step": 2085
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729967637540454e-05,
+      "loss": 0.0422,
+      "step": 2086
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729838187702265e-05,
+      "loss": 0.1199,
+      "step": 2087
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729708737864078e-05,
+      "loss": 0.0518,
+      "step": 2088
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729579288025891e-05,
+      "loss": 0.0724,
+      "step": 2089
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729449838187703e-05,
+      "loss": 0.0602,
+      "step": 2090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729320388349515e-05,
+      "loss": 0.0468,
+      "step": 2091
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.729190938511327e-05,
+      "loss": 0.0596,
+      "step": 2092
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.72906148867314e-05,
+      "loss": 0.0386,
+      "step": 2093
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728932038834951e-05,
+      "loss": 0.0499,
+      "step": 2094
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728802588996764e-05,
+      "loss": 0.045,
+      "step": 2095
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728673139158577e-05,
+      "loss": 0.0435,
+      "step": 2096
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728543689320388e-05,
+      "loss": 0.0546,
+      "step": 2097
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728414239482201e-05,
+      "loss": 0.0344,
+      "step": 2098
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728284789644013e-05,
+      "loss": 0.0474,
+      "step": 2099
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728155339805826e-05,
+      "loss": 0.04,
+      "step": 2100
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.728025889967637e-05,
+      "loss": 0.0421,
+      "step": 2101
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.72789644012945e-05,
+      "loss": 0.0616,
+      "step": 2102
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727766990291263e-05,
+      "loss": 0.0393,
+      "step": 2103
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727637540453075e-05,
+      "loss": 0.0641,
+      "step": 2104
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727508090614888e-05,
+      "loss": 0.0527,
+      "step": 2105
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727378640776699e-05,
+      "loss": 0.0452,
+      "step": 2106
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727249190938512e-05,
+      "loss": 0.0535,
+      "step": 2107
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.727119741100325e-05,
+      "loss": 0.0625,
+      "step": 2108
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726990291262137e-05,
+      "loss": 0.0718,
+      "step": 2109
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726860841423949e-05,
+      "loss": 0.0394,
+      "step": 2110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726731391585761e-05,
+      "loss": 0.0621,
+      "step": 2111
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726601941747572e-05,
+      "loss": 0.0335,
+      "step": 2112
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726472491909385e-05,
+      "loss": 0.072,
+      "step": 2113
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726343042071198e-05,
+      "loss": 0.062,
+      "step": 2114
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72621359223301e-05,
+      "loss": 0.0656,
+      "step": 2115
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.726084142394822e-05,
+      "loss": 0.054,
+      "step": 2116
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725954692556634e-05,
+      "loss": 0.0307,
+      "step": 2117
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725825242718447e-05,
+      "loss": 0.0493,
+      "step": 2118
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.72569579288026e-05,
+      "loss": 0.0373,
+      "step": 2119
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725566343042072e-05,
+      "loss": 0.0459,
+      "step": 2120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725436893203884e-05,
+      "loss": 0.056,
+      "step": 2121
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725307443365696e-05,
+      "loss": 0.0454,
+      "step": 2122
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725177993527509e-05,
+      "loss": 0.0342,
+      "step": 2123
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.725048543689322e-05,
+      "loss": 0.0456,
+      "step": 2124
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724919093851133e-05,
+      "loss": 0.0276,
+      "step": 2125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724789644012944e-05,
+      "loss": 0.0294,
+      "step": 2126
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724660194174757e-05,
+      "loss": 0.0632,
+      "step": 2127
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.72453074433657e-05,
+      "loss": 0.0511,
+      "step": 2128
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724401294498382e-05,
+      "loss": 0.0649,
+      "step": 2129
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724271844660195e-05,
+      "loss": 0.0316,
+      "step": 2130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724142394822006e-05,
+      "loss": 0.0504,
+      "step": 2131
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.724012944983819e-05,
+      "loss": 0.0461,
+      "step": 2132
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723883495145632e-05,
+      "loss": 0.0319,
+      "step": 2133
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723754045307444e-05,
+      "loss": 0.0432,
+      "step": 2134
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723624595469257e-05,
+      "loss": 0.036,
+      "step": 2135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723495145631068e-05,
+      "loss": 0.048,
+      "step": 2136
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723365695792881e-05,
+      "loss": 0.0658,
+      "step": 2137
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723236245954693e-05,
+      "loss": 0.0294,
+      "step": 2138
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.723106796116506e-05,
+      "loss": 0.0409,
+      "step": 2139
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722977346278317e-05,
+      "loss": 0.0427,
+      "step": 2140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72284789644013e-05,
+      "loss": 0.0432,
+      "step": 2141
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722718446601943e-05,
+      "loss": 0.1008,
+      "step": 2142
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722588996763754e-05,
+      "loss": 0.0407,
+      "step": 2143
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722459546925567e-05,
+      "loss": 0.0356,
+      "step": 2144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72233009708738e-05,
+      "loss": 0.0613,
+      "step": 2145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72220064724919e-05,
+      "loss": 0.038,
+      "step": 2146
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.722071197411003e-05,
+      "loss": 0.07,
+      "step": 2147
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721941747572816e-05,
+      "loss": 0.0664,
+      "step": 2148
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721812297734629e-05,
+      "loss": 0.066,
+      "step": 2149
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72168284789644e-05,
+      "loss": 0.0567,
+      "step": 2150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721553398058253e-05,
+      "loss": 0.0503,
+      "step": 2151
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721423948220065e-05,
+      "loss": 0.0505,
+      "step": 2152
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721294498381878e-05,
+      "loss": 0.0586,
+      "step": 2153
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.72116504854369e-05,
+      "loss": 0.0442,
+      "step": 2154
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.721035598705502e-05,
+      "loss": 0.0487,
+      "step": 2155
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720906148867315e-05,
+      "loss": 0.0236,
+      "step": 2156
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720776699029127e-05,
+      "loss": 0.0257,
+      "step": 2157
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.72064724919094e-05,
+      "loss": 0.0338,
+      "step": 2158
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720517799352751e-05,
+      "loss": 0.0856,
+      "step": 2159
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720388349514562e-05,
+      "loss": 0.0633,
+      "step": 2160
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720258899676375e-05,
+      "loss": 0.0571,
+      "step": 2161
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.720129449838188e-05,
+      "loss": 0.0396,
+      "step": 2162
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.72e-05,
+      "loss": 0.0586,
+      "step": 2163
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719870550161813e-05,
+      "loss": 0.0412,
+      "step": 2164
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719741100323624e-05,
+      "loss": 0.0277,
+      "step": 2165
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719611650485437e-05,
+      "loss": 0.0486,
+      "step": 2166
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.71948220064725e-05,
+      "loss": 0.0526,
+      "step": 2167
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719352750809062e-05,
+      "loss": 0.0295,
+      "step": 2168
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719223300970875e-05,
+      "loss": 0.0516,
+      "step": 2169
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.719093851132686e-05,
+      "loss": 0.0426,
+      "step": 2170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718964401294499e-05,
+      "loss": 0.047,
+      "step": 2171
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718834951456312e-05,
+      "loss": 0.06,
+      "step": 2172
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718705501618124e-05,
+      "loss": 0.024,
+      "step": 2173
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718576051779936e-05,
+      "loss": 0.0324,
+      "step": 2174
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718446601941747e-05,
+      "loss": 0.0233,
+      "step": 2175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.71831715210356e-05,
+      "loss": 0.0418,
+      "step": 2176
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718187702265372e-05,
+      "loss": 0.0402,
+      "step": 2177
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.718058252427185e-05,
+      "loss": 0.0407,
+      "step": 2178
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717928802588998e-05,
+      "loss": 0.0793,
+      "step": 2179
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717799352750809e-05,
+      "loss": 0.0356,
+      "step": 2180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717669902912622e-05,
+      "loss": 0.0269,
+      "step": 2181
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717540453074434e-05,
+      "loss": 0.0403,
+      "step": 2182
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717411003236247e-05,
+      "loss": 0.0833,
+      "step": 2183
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717281553398058e-05,
+      "loss": 0.0643,
+      "step": 2184
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717152103559871e-05,
+      "loss": 0.0188,
+      "step": 2185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.717022653721683e-05,
+      "loss": 0.0339,
+      "step": 2186
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716893203883496e-05,
+      "loss": 0.0409,
+      "step": 2187
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716763754045309e-05,
+      "loss": 0.0314,
+      "step": 2188
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71663430420712e-05,
+      "loss": 0.0592,
+      "step": 2189
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716504854368933e-05,
+      "loss": 0.0566,
+      "step": 2190
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716375404530745e-05,
+      "loss": 0.0464,
+      "step": 2191
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.716245954692557e-05,
+      "loss": 0.0293,
+      "step": 2192
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71611650485437e-05,
+      "loss": 0.0367,
+      "step": 2193
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.71598705501618e-05,
+      "loss": 0.0384,
+      "step": 2194
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715857605177993e-05,
+      "loss": 0.0871,
+      "step": 2195
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715728155339806e-05,
+      "loss": 0.0221,
+      "step": 2196
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715598705501619e-05,
+      "loss": 0.0295,
+      "step": 2197
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715469255663431e-05,
+      "loss": 0.0427,
+      "step": 2198
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715339805825243e-05,
+      "loss": 0.0447,
+      "step": 2199
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715210355987055e-05,
+      "loss": 0.0338,
+      "step": 2200
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.715080906148868e-05,
+      "loss": 0.0701,
+      "step": 2201
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71495145631068e-05,
+      "loss": 0.0369,
+      "step": 2202
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714822006472493e-05,
+      "loss": 0.0433,
+      "step": 2203
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714692556634305e-05,
+      "loss": 0.0329,
+      "step": 2204
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714563106796117e-05,
+      "loss": 0.0299,
+      "step": 2205
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71443365695793e-05,
+      "loss": 0.0489,
+      "step": 2206
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714304207119743e-05,
+      "loss": 0.0357,
+      "step": 2207
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714174757281554e-05,
+      "loss": 0.053,
+      "step": 2208
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.714045307443365e-05,
+      "loss": 0.0559,
+      "step": 2209
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713915857605178e-05,
+      "loss": 0.0611,
+      "step": 2210
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71378640776699e-05,
+      "loss": 0.0734,
+      "step": 2211
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713656957928803e-05,
+      "loss": 0.029,
+      "step": 2212
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713527508090616e-05,
+      "loss": 0.0751,
+      "step": 2213
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713398058252427e-05,
+      "loss": 0.0458,
+      "step": 2214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.71326860841424e-05,
+      "loss": 0.0346,
+      "step": 2215
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713139158576052e-05,
+      "loss": 0.0952,
+      "step": 2216
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.713009708737865e-05,
+      "loss": 0.0566,
+      "step": 2217
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712880258899676e-05,
+      "loss": 0.0372,
+      "step": 2218
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712750809061489e-05,
+      "loss": 0.0452,
+      "step": 2219
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712621359223302e-05,
+      "loss": 0.053,
+      "step": 2220
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712491909385114e-05,
+      "loss": 0.0903,
+      "step": 2221
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712362459546927e-05,
+      "loss": 0.0551,
+      "step": 2222
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.712233009708738e-05,
+      "loss": 0.0307,
+      "step": 2223
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71210355987055e-05,
+      "loss": 0.0537,
+      "step": 2224
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711974110032362e-05,
+      "loss": 0.0775,
+      "step": 2225
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711844660194175e-05,
+      "loss": 0.0595,
+      "step": 2226
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711715210355988e-05,
+      "loss": 0.0478,
+      "step": 2227
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.7115857605178e-05,
+      "loss": 0.0261,
+      "step": 2228
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711456310679612e-05,
+      "loss": 0.0614,
+      "step": 2229
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711326860841424e-05,
+      "loss": 0.0298,
+      "step": 2230
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.711197411003237e-05,
+      "loss": 0.06,
+      "step": 2231
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.71106796116505e-05,
+      "loss": 0.0496,
+      "step": 2232
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710938511326861e-05,
+      "loss": 0.027,
+      "step": 2233
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710809061488673e-05,
+      "loss": 0.0411,
+      "step": 2234
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710679611650486e-05,
+      "loss": 0.0429,
+      "step": 2235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710550161812299e-05,
+      "loss": 0.0468,
+      "step": 2236
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710420711974111e-05,
+      "loss": 0.0878,
+      "step": 2237
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710291262135923e-05,
+      "loss": 0.0457,
+      "step": 2238
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710161812297735e-05,
+      "loss": 0.041,
+      "step": 2239
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.710032362459547e-05,
+      "loss": 0.0521,
+      "step": 2240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.70990291262136e-05,
+      "loss": 0.0392,
+      "step": 2241
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709773462783172e-05,
+      "loss": 0.0377,
+      "step": 2242
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709644012944983e-05,
+      "loss": 0.0505,
+      "step": 2243
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709514563106796e-05,
+      "loss": 0.0537,
+      "step": 2244
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709385113268609e-05,
+      "loss": 0.0614,
+      "step": 2245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709255663430421e-05,
+      "loss": 0.0371,
+      "step": 2246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.709126213592234e-05,
+      "loss": 0.0303,
+      "step": 2247
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708996763754045e-05,
+      "loss": 0.0481,
+      "step": 2248
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708867313915858e-05,
+      "loss": 0.0724,
+      "step": 2249
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70873786407767e-05,
+      "loss": 0.0483,
+      "step": 2250
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708608414239483e-05,
+      "loss": 0.0699,
+      "step": 2251
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708478964401296e-05,
+      "loss": 0.0383,
+      "step": 2252
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708349514563107e-05,
+      "loss": 0.0684,
+      "step": 2253
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70822006472492e-05,
+      "loss": 0.0764,
+      "step": 2254
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.708090614886733e-05,
+      "loss": 0.0447,
+      "step": 2255
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707961165048545e-05,
+      "loss": 0.0386,
+      "step": 2256
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707831715210356e-05,
+      "loss": 0.0795,
+      "step": 2257
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707702265372168e-05,
+      "loss": 0.0396,
+      "step": 2258
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70757281553398e-05,
+      "loss": 0.0517,
+      "step": 2259
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707443365695793e-05,
+      "loss": 0.1167,
+      "step": 2260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707313915857606e-05,
+      "loss": 0.06,
+      "step": 2261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.707184466019418e-05,
+      "loss": 0.0723,
+      "step": 2262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.70705501618123e-05,
+      "loss": 0.058,
+      "step": 2263
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706925566343042e-05,
+      "loss": 0.051,
+      "step": 2264
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706796116504855e-05,
+      "loss": 0.0295,
+      "step": 2265
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706666666666668e-05,
+      "loss": 0.0353,
+      "step": 2266
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706537216828479e-05,
+      "loss": 0.0454,
+      "step": 2267
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706407766990292e-05,
+      "loss": 0.0995,
+      "step": 2268
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706278317152104e-05,
+      "loss": 0.0622,
+      "step": 2269
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.706148867313917e-05,
+      "loss": 0.1094,
+      "step": 2270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70601941747573e-05,
+      "loss": 0.0516,
+      "step": 2271
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705889967637541e-05,
+      "loss": 0.0522,
+      "step": 2272
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705760517799352e-05,
+      "loss": 0.0822,
+      "step": 2273
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705631067961165e-05,
+      "loss": 0.0523,
+      "step": 2274
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705501618122978e-05,
+      "loss": 0.0685,
+      "step": 2275
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.70537216828479e-05,
+      "loss": 0.0365,
+      "step": 2276
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705242718446602e-05,
+      "loss": 0.0273,
+      "step": 2277
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.705113268608414e-05,
+      "loss": 0.088,
+      "step": 2278
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704983818770227e-05,
+      "loss": 0.0508,
+      "step": 2279
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.70485436893204e-05,
+      "loss": 0.0503,
+      "step": 2280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704724919093852e-05,
+      "loss": 0.0499,
+      "step": 2281
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704595469255663e-05,
+      "loss": 0.0441,
+      "step": 2282
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704466019417476e-05,
+      "loss": 0.0901,
+      "step": 2283
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704336569579289e-05,
+      "loss": 0.0302,
+      "step": 2284
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704207119741101e-05,
+      "loss": 0.0706,
+      "step": 2285
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.704077669902914e-05,
+      "loss": 0.0718,
+      "step": 2286
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703948220064725e-05,
+      "loss": 0.0382,
+      "step": 2287
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703818770226538e-05,
+      "loss": 0.0412,
+      "step": 2288
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.70368932038835e-05,
+      "loss": 0.0381,
+      "step": 2289
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703559870550162e-05,
+      "loss": 0.0759,
+      "step": 2290
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703430420711975e-05,
+      "loss": 0.066,
+      "step": 2291
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703300970873786e-05,
+      "loss": 0.0339,
+      "step": 2292
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703171521035599e-05,
+      "loss": 0.0633,
+      "step": 2293
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.703042071197411e-05,
+      "loss": 0.0493,
+      "step": 2294
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702912621359224e-05,
+      "loss": 0.0434,
+      "step": 2295
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702783171521037e-05,
+      "loss": 0.0834,
+      "step": 2296
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702653721682848e-05,
+      "loss": 0.0296,
+      "step": 2297
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.70252427184466e-05,
+      "loss": 0.0819,
+      "step": 2298
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702394822006473e-05,
+      "loss": 0.0481,
+      "step": 2299
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702265372168286e-05,
+      "loss": 0.0378,
+      "step": 2300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.702135922330097e-05,
+      "loss": 0.0289,
+      "step": 2301
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.70200647249191e-05,
+      "loss": 0.0378,
+      "step": 2302
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701877022653723e-05,
+      "loss": 0.0846,
+      "step": 2303
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701747572815535e-05,
+      "loss": 0.0488,
+      "step": 2304
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701618122977348e-05,
+      "loss": 0.0395,
+      "step": 2305
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701488673139159e-05,
+      "loss": 0.0487,
+      "step": 2306
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.70135922330097e-05,
+      "loss": 0.0484,
+      "step": 2307
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701229773462783e-05,
+      "loss": 0.035,
+      "step": 2308
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.701100323624596e-05,
+      "loss": 0.0574,
+      "step": 2309
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700970873786408e-05,
+      "loss": 0.0355,
+      "step": 2310
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.70084142394822e-05,
+      "loss": 0.0337,
+      "step": 2311
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700711974110032e-05,
+      "loss": 0.0563,
+      "step": 2312
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700582524271845e-05,
+      "loss": 0.0449,
+      "step": 2313
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700453074433658e-05,
+      "loss": 0.0524,
+      "step": 2314
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.70032362459547e-05,
+      "loss": 0.0604,
+      "step": 2315
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700194174757282e-05,
+      "loss": 0.0392,
+      "step": 2316
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.700064724919094e-05,
+      "loss": 0.0499,
+      "step": 2317
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699935275080907e-05,
+      "loss": 0.0654,
+      "step": 2318
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.69980582524272e-05,
+      "loss": 0.0709,
+      "step": 2319
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699676375404532e-05,
+      "loss": 0.0588,
+      "step": 2320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699546925566344e-05,
+      "loss": 0.0722,
+      "step": 2321
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699417475728155e-05,
+      "loss": 0.0374,
+      "step": 2322
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699288025889968e-05,
+      "loss": 0.0254,
+      "step": 2323
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.69915857605178e-05,
+      "loss": 0.0567,
+      "step": 2324
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.699029126213593e-05,
+      "loss": 0.0305,
+      "step": 2325
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698899676375404e-05,
+      "loss": 0.0285,
+      "step": 2326
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698770226537217e-05,
+      "loss": 0.0565,
+      "step": 2327
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.69864077669903e-05,
+      "loss": 0.047,
+      "step": 2328
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698511326860842e-05,
+      "loss": 0.0557,
+      "step": 2329
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698381877022655e-05,
+      "loss": 0.0807,
+      "step": 2330
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698252427184466e-05,
+      "loss": 0.0531,
+      "step": 2331
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.698122977346279e-05,
+      "loss": 0.0351,
+      "step": 2332
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697993527508091e-05,
+      "loss": 0.0586,
+      "step": 2333
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697864077669904e-05,
+      "loss": 0.0277,
+      "step": 2334
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697734627831715e-05,
+      "loss": 0.0475,
+      "step": 2335
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697605177993528e-05,
+      "loss": 0.0451,
+      "step": 2336
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697475728155341e-05,
+      "loss": 0.0423,
+      "step": 2337
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697346278317152e-05,
+      "loss": 0.0609,
+      "step": 2338
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697216828478965e-05,
+      "loss": 0.0537,
+      "step": 2339
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.697087378640777e-05,
+      "loss": 0.0622,
+      "step": 2340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696957928802589e-05,
+      "loss": 0.1072,
+      "step": 2341
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696828478964401e-05,
+      "loss": 0.0426,
+      "step": 2342
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696699029126214e-05,
+      "loss": 0.0395,
+      "step": 2343
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696569579288027e-05,
+      "loss": 0.0733,
+      "step": 2344
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696440129449839e-05,
+      "loss": 0.0478,
+      "step": 2345
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.69631067961165e-05,
+      "loss": 0.044,
+      "step": 2346
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696181229773463e-05,
+      "loss": 0.1052,
+      "step": 2347
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.696051779935276e-05,
+      "loss": 0.0447,
+      "step": 2348
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695922330097089e-05,
+      "loss": 0.0365,
+      "step": 2349
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.6957928802589e-05,
+      "loss": 0.0718,
+      "step": 2350
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695663430420713e-05,
+      "loss": 0.0404,
+      "step": 2351
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695533980582525e-05,
+      "loss": 0.0831,
+      "step": 2352
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695404530744338e-05,
+      "loss": 0.0598,
+      "step": 2353
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695275080906149e-05,
+      "loss": 0.0461,
+      "step": 2354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695145631067962e-05,
+      "loss": 0.0798,
+      "step": 2355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.695016181229773e-05,
+      "loss": 0.0692,
+      "step": 2356
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694886731391586e-05,
+      "loss": 0.0544,
+      "step": 2357
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694757281553398e-05,
+      "loss": 0.098,
+      "step": 2358
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694627831715211e-05,
+      "loss": 0.0418,
+      "step": 2359
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694498381877022e-05,
+      "loss": 0.0718,
+      "step": 2360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694368932038835e-05,
+      "loss": 0.0389,
+      "step": 2361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.694239482200648e-05,
+      "loss": 0.1461,
+      "step": 2362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.69411003236246e-05,
+      "loss": 0.0447,
+      "step": 2363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693980582524273e-05,
+      "loss": 0.046,
+      "step": 2364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693851132686084e-05,
+      "loss": 0.1452,
+      "step": 2365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693721682847897e-05,
+      "loss": 0.0388,
+      "step": 2366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.69359223300971e-05,
+      "loss": 0.0488,
+      "step": 2367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693462783171522e-05,
+      "loss": 0.0704,
+      "step": 2368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693333333333335e-05,
+      "loss": 0.0674,
+      "step": 2369
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693203883495146e-05,
+      "loss": 0.046,
+      "step": 2370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.693074433656958e-05,
+      "loss": 0.0354,
+      "step": 2371
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.69294498381877e-05,
+      "loss": 0.0424,
+      "step": 2372
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692815533980583e-05,
+      "loss": 0.0691,
+      "step": 2373
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692686084142396e-05,
+      "loss": 0.0389,
+      "step": 2374
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692556634304207e-05,
+      "loss": 0.0504,
+      "step": 2375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.69242718446602e-05,
+      "loss": 0.0384,
+      "step": 2376
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692297734627832e-05,
+      "loss": 0.046,
+      "step": 2377
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692168284789645e-05,
+      "loss": 0.0248,
+      "step": 2378
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.692038834951457e-05,
+      "loss": 0.0318,
+      "step": 2379
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691909385113269e-05,
+      "loss": 0.0776,
+      "step": 2380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691779935275081e-05,
+      "loss": 0.0697,
+      "step": 2381
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691650485436894e-05,
+      "loss": 0.0384,
+      "step": 2382
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691521035598707e-05,
+      "loss": 0.0434,
+      "step": 2383
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691391585760518e-05,
+      "loss": 0.0358,
+      "step": 2384
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691262135922331e-05,
+      "loss": 0.0377,
+      "step": 2385
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691132686084143e-05,
+      "loss": 0.046,
+      "step": 2386
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.691003236245955e-05,
+      "loss": 0.023,
+      "step": 2387
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690873786407767e-05,
+      "loss": 0.0508,
+      "step": 2388
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.69074433656958e-05,
+      "loss": 0.0469,
+      "step": 2389
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690614886731391e-05,
+      "loss": 0.0463,
+      "step": 2390
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690485436893204e-05,
+      "loss": 0.1121,
+      "step": 2391
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690355987055017e-05,
+      "loss": 0.0582,
+      "step": 2392
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.690226537216829e-05,
+      "loss": 0.0661,
+      "step": 2393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.69009708737864e-05,
+      "loss": 0.041,
+      "step": 2394
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689967637540453e-05,
+      "loss": 0.0323,
+      "step": 2395
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689838187702266e-05,
+      "loss": 0.0374,
+      "step": 2396
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689708737864079e-05,
+      "loss": 0.0455,
+      "step": 2397
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689579288025891e-05,
+      "loss": 0.1005,
+      "step": 2398
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689449838187703e-05,
+      "loss": 0.0237,
+      "step": 2399
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689320388349515e-05,
+      "loss": 0.1475,
+      "step": 2400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.689190938511328e-05,
+      "loss": 0.0384,
+      "step": 2401
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.68906148867314e-05,
+      "loss": 0.0529,
+      "step": 2402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688932038834952e-05,
+      "loss": 0.0527,
+      "step": 2403
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688802588996763e-05,
+      "loss": 0.0377,
+      "step": 2404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688673139158576e-05,
+      "loss": 0.0682,
+      "step": 2405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688543689320388e-05,
+      "loss": 0.0276,
+      "step": 2406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688414239482201e-05,
+      "loss": 0.045,
+      "step": 2407
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688284789644014e-05,
+      "loss": 0.053,
+      "step": 2408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688155339805825e-05,
+      "loss": 0.0339,
+      "step": 2409
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.688025889967638e-05,
+      "loss": 0.0617,
+      "step": 2410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.68789644012945e-05,
+      "loss": 0.0667,
+      "step": 2411
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687766990291263e-05,
+      "loss": 0.0279,
+      "step": 2412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687637540453076e-05,
+      "loss": 0.0557,
+      "step": 2413
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687508090614887e-05,
+      "loss": 0.0885,
+      "step": 2414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.6873786407767e-05,
+      "loss": 0.0588,
+      "step": 2415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687249190938512e-05,
+      "loss": 0.0831,
+      "step": 2416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.687119741100325e-05,
+      "loss": 0.0315,
+      "step": 2417
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686990291262136e-05,
+      "loss": 0.0492,
+      "step": 2418
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686860841423949e-05,
+      "loss": 0.0405,
+      "step": 2419
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68673139158576e-05,
+      "loss": 0.0534,
+      "step": 2420
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686601941747573e-05,
+      "loss": 0.0659,
+      "step": 2421
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686472491909386e-05,
+      "loss": 0.0674,
+      "step": 2422
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686343042071198e-05,
+      "loss": 0.0563,
+      "step": 2423
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68621359223301e-05,
+      "loss": 0.0423,
+      "step": 2424
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.686084142394822e-05,
+      "loss": 0.0658,
+      "step": 2425
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685954692556635e-05,
+      "loss": 0.0319,
+      "step": 2426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685825242718447e-05,
+      "loss": 0.1967,
+      "step": 2427
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685695792880259e-05,
+      "loss": 0.047,
+      "step": 2428
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685566343042071e-05,
+      "loss": 0.0493,
+      "step": 2429
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685436893203884e-05,
+      "loss": 0.0511,
+      "step": 2430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685307443365697e-05,
+      "loss": 0.0501,
+      "step": 2431
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68517799352751e-05,
+      "loss": 0.0476,
+      "step": 2432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685048543689321e-05,
+      "loss": 0.0545,
+      "step": 2433
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684919093851133e-05,
+      "loss": 0.0764,
+      "step": 2434
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684789644012946e-05,
+      "loss": 0.0937,
+      "step": 2435
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684660194174757e-05,
+      "loss": 0.0404,
+      "step": 2436
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.68453074433657e-05,
+      "loss": 0.0317,
+      "step": 2437
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684401294498383e-05,
+      "loss": 0.0398,
+      "step": 2438
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684271844660194e-05,
+      "loss": 0.0303,
+      "step": 2439
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684142394822007e-05,
+      "loss": 0.0284,
+      "step": 2440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.684012944983819e-05,
+      "loss": 0.0408,
+      "step": 2441
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683883495145632e-05,
+      "loss": 0.0546,
+      "step": 2442
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683754045307443e-05,
+      "loss": 0.0592,
+      "step": 2443
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683624595469256e-05,
+      "loss": 0.0282,
+      "step": 2444
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683495145631069e-05,
+      "loss": 0.0494,
+      "step": 2445
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683365695792881e-05,
+      "loss": 0.0367,
+      "step": 2446
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683236245954694e-05,
+      "loss": 0.0381,
+      "step": 2447
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.683106796116505e-05,
+      "loss": 0.0732,
+      "step": 2448
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682977346278318e-05,
+      "loss": 0.0431,
+      "step": 2449
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.68284789644013e-05,
+      "loss": 0.0277,
+      "step": 2450
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682718446601943e-05,
+      "loss": 0.0504,
+      "step": 2451
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682588996763754e-05,
+      "loss": 0.0468,
+      "step": 2452
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682459546925566e-05,
+      "loss": 0.0394,
+      "step": 2453
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682330097087378e-05,
+      "loss": 0.049,
+      "step": 2454
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682200647249191e-05,
+      "loss": 0.0399,
+      "step": 2455
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.682071197411004e-05,
+      "loss": 0.0512,
+      "step": 2456
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681941747572816e-05,
+      "loss": 0.0607,
+      "step": 2457
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681812297734628e-05,
+      "loss": 0.0352,
+      "step": 2458
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.68168284789644e-05,
+      "loss": 0.0402,
+      "step": 2459
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681553398058253e-05,
+      "loss": 0.0422,
+      "step": 2460
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681423948220066e-05,
+      "loss": 0.0389,
+      "step": 2461
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681294498381877e-05,
+      "loss": 0.0589,
+      "step": 2462
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.68116504854369e-05,
+      "loss": 0.0345,
+      "step": 2463
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.681035598705502e-05,
+      "loss": 0.0189,
+      "step": 2464
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680906148867315e-05,
+      "loss": 0.0566,
+      "step": 2465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680776699029128e-05,
+      "loss": 0.0455,
+      "step": 2466
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680647249190939e-05,
+      "loss": 0.0531,
+      "step": 2467
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680517799352752e-05,
+      "loss": 0.0271,
+      "step": 2468
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680388349514563e-05,
+      "loss": 0.0732,
+      "step": 2469
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680258899676376e-05,
+      "loss": 0.0369,
+      "step": 2470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680129449838188e-05,
+      "loss": 0.0429,
+      "step": 2471
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 0.0443,
+      "step": 2472
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679870550161812e-05,
+      "loss": 0.027,
+      "step": 2473
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679741100323625e-05,
+      "loss": 0.0316,
+      "step": 2474
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679611650485437e-05,
+      "loss": 0.0403,
+      "step": 2475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.67948220064725e-05,
+      "loss": 0.0874,
+      "step": 2476
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679352750809061e-05,
+      "loss": 0.0499,
+      "step": 2477
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679223300970874e-05,
+      "loss": 0.1135,
+      "step": 2478
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.679093851132687e-05,
+      "loss": 0.0348,
+      "step": 2479
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.6789644012945e-05,
+      "loss": 0.0872,
+      "step": 2480
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678834951456312e-05,
+      "loss": 0.0491,
+      "step": 2481
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678705501618123e-05,
+      "loss": 0.0837,
+      "step": 2482
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678576051779936e-05,
+      "loss": 0.0758,
+      "step": 2483
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678446601941749e-05,
+      "loss": 0.0303,
+      "step": 2484
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.67831715210356e-05,
+      "loss": 0.0226,
+      "step": 2485
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678187702265373e-05,
+      "loss": 0.0578,
+      "step": 2486
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.678058252427184e-05,
+      "loss": 0.028,
+      "step": 2487
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677928802588997e-05,
+      "loss": 0.0541,
+      "step": 2488
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677799352750809e-05,
+      "loss": 0.0676,
+      "step": 2489
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677669902912622e-05,
+      "loss": 0.1032,
+      "step": 2490
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677540453074435e-05,
+      "loss": 0.0458,
+      "step": 2491
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677411003236246e-05,
+      "loss": 0.0426,
+      "step": 2492
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677281553398059e-05,
+      "loss": 0.0917,
+      "step": 2493
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677152103559871e-05,
+      "loss": 0.0378,
+      "step": 2494
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.677022653721684e-05,
+      "loss": 0.0797,
+      "step": 2495
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676893203883497e-05,
+      "loss": 0.0352,
+      "step": 2496
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676763754045308e-05,
+      "loss": 0.1286,
+      "step": 2497
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.67663430420712e-05,
+      "loss": 0.0483,
+      "step": 2498
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676504854368933e-05,
+      "loss": 0.0461,
+      "step": 2499
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676375404530746e-05,
+      "loss": 0.0567,
+      "step": 2500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676245954692557e-05,
+      "loss": 0.0547,
+      "step": 2501
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.676116504854368e-05,
+      "loss": 0.0449,
+      "step": 2502
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675987055016181e-05,
+      "loss": 0.0271,
+      "step": 2503
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675857605177994e-05,
+      "loss": 0.0442,
+      "step": 2504
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675728155339806e-05,
+      "loss": 0.0437,
+      "step": 2505
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675598705501619e-05,
+      "loss": 0.0296,
+      "step": 2506
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.67546925566343e-05,
+      "loss": 0.0384,
+      "step": 2507
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675339805825243e-05,
+      "loss": 0.0539,
+      "step": 2508
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675210355987056e-05,
+      "loss": 0.0378,
+      "step": 2509
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.675080906148868e-05,
+      "loss": 0.1487,
+      "step": 2510
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.67495145631068e-05,
+      "loss": 0.1017,
+      "step": 2511
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674822006472492e-05,
+      "loss": 0.0462,
+      "step": 2512
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674692556634305e-05,
+      "loss": 0.1133,
+      "step": 2513
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674563106796118e-05,
+      "loss": 0.0307,
+      "step": 2514
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.67443365695793e-05,
+      "loss": 0.053,
+      "step": 2515
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674304207119742e-05,
+      "loss": 0.049,
+      "step": 2516
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674174757281553e-05,
+      "loss": 0.0485,
+      "step": 2517
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.674045307443366e-05,
+      "loss": 0.0415,
+      "step": 2518
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673915857605178e-05,
+      "loss": 0.0553,
+      "step": 2519
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673786407766991e-05,
+      "loss": 0.0504,
+      "step": 2520
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673656957928802e-05,
+      "loss": 0.0322,
+      "step": 2521
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673527508090615e-05,
+      "loss": 0.049,
+      "step": 2522
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673398058252427e-05,
+      "loss": 0.0484,
+      "step": 2523
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.67326860841424e-05,
+      "loss": 0.0793,
+      "step": 2524
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673139158576053e-05,
+      "loss": 0.0448,
+      "step": 2525
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.673009708737864e-05,
+      "loss": 0.0911,
+      "step": 2526
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672880258899677e-05,
+      "loss": 0.0361,
+      "step": 2527
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.67275080906149e-05,
+      "loss": 0.0609,
+      "step": 2528
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672621359223302e-05,
+      "loss": 0.0264,
+      "step": 2529
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672491909385115e-05,
+      "loss": 0.1699,
+      "step": 2530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672362459546926e-05,
+      "loss": 0.0468,
+      "step": 2531
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672233009708739e-05,
+      "loss": 0.0561,
+      "step": 2532
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.672103559870551e-05,
+      "loss": 0.0479,
+      "step": 2533
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671974110032363e-05,
+      "loss": 0.0462,
+      "step": 2534
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671844660194175e-05,
+      "loss": 0.04,
+      "step": 2535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671715210355987e-05,
+      "loss": 0.0256,
+      "step": 2536
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671585760517799e-05,
+      "loss": 0.0327,
+      "step": 2537
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671456310679612e-05,
+      "loss": 0.0653,
+      "step": 2538
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671326860841425e-05,
+      "loss": 0.066,
+      "step": 2539
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671197411003237e-05,
+      "loss": 0.0377,
+      "step": 2540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.671067961165049e-05,
+      "loss": 0.0468,
+      "step": 2541
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670938511326861e-05,
+      "loss": 0.0281,
+      "step": 2542
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670809061488674e-05,
+      "loss": 0.0857,
+      "step": 2543
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670679611650487e-05,
+      "loss": 0.0419,
+      "step": 2544
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670550161812298e-05,
+      "loss": 0.03,
+      "step": 2545
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.67042071197411e-05,
+      "loss": 0.0402,
+      "step": 2546
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670291262135923e-05,
+      "loss": 0.0963,
+      "step": 2547
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670161812297736e-05,
+      "loss": 0.0542,
+      "step": 2548
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.670032362459548e-05,
+      "loss": 0.0828,
+      "step": 2549
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.66990291262136e-05,
+      "loss": 0.0712,
+      "step": 2550
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669773462783171e-05,
+      "loss": 0.0374,
+      "step": 2551
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669644012944984e-05,
+      "loss": 0.022,
+      "step": 2552
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669514563106796e-05,
+      "loss": 0.0244,
+      "step": 2553
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669385113268609e-05,
+      "loss": 0.0388,
+      "step": 2554
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.66925566343042e-05,
+      "loss": 0.0299,
+      "step": 2555
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.669126213592233e-05,
+      "loss": 0.0384,
+      "step": 2556
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668996763754046e-05,
+      "loss": 0.0316,
+      "step": 2557
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668867313915858e-05,
+      "loss": 0.0233,
+      "step": 2558
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668737864077671e-05,
+      "loss": 0.0459,
+      "step": 2559
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668608414239482e-05,
+      "loss": 0.0536,
+      "step": 2560
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668478964401295e-05,
+      "loss": 0.0321,
+      "step": 2561
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668349514563108e-05,
+      "loss": 0.0543,
+      "step": 2562
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.66822006472492e-05,
+      "loss": 0.062,
+      "step": 2563
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.668090614886733e-05,
+      "loss": 0.0467,
+      "step": 2564
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667961165048544e-05,
+      "loss": 0.0388,
+      "step": 2565
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667831715210356e-05,
+      "loss": 0.0831,
+      "step": 2566
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667702265372168e-05,
+      "loss": 0.0672,
+      "step": 2567
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667572815533981e-05,
+      "loss": 0.0263,
+      "step": 2568
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667443365695793e-05,
+      "loss": 0.0893,
+      "step": 2569
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667313915857605e-05,
+      "loss": 0.0795,
+      "step": 2570
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.667184466019417e-05,
+      "loss": 0.0464,
+      "step": 2571
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.66705501618123e-05,
+      "loss": 0.0623,
+      "step": 2572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666925566343043e-05,
+      "loss": 0.0398,
+      "step": 2573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666796116504855e-05,
+      "loss": 0.0519,
+      "step": 2574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666666666666667e-05,
+      "loss": 0.0601,
+      "step": 2575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.66653721682848e-05,
+      "loss": 0.07,
+      "step": 2576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666407766990292e-05,
+      "loss": 0.0858,
+      "step": 2577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666278317152105e-05,
+      "loss": 0.0564,
+      "step": 2578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666148867313916e-05,
+      "loss": 0.038,
+      "step": 2579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.666019417475729e-05,
+      "loss": 0.0489,
+      "step": 2580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665889967637541e-05,
+      "loss": 0.0478,
+      "step": 2581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665760517799354e-05,
+      "loss": 0.0472,
+      "step": 2582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665631067961165e-05,
+      "loss": 0.03,
+      "step": 2583
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665501618122978e-05,
+      "loss": 0.0424,
+      "step": 2584
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665372168284789e-05,
+      "loss": 0.0274,
+      "step": 2585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665242718446602e-05,
+      "loss": 0.0426,
+      "step": 2586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.665113268608415e-05,
+      "loss": 0.051,
+      "step": 2587
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664983818770227e-05,
+      "loss": 0.0615,
+      "step": 2588
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.66485436893204e-05,
+      "loss": 0.0318,
+      "step": 2589
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664724919093851e-05,
+      "loss": 0.067,
+      "step": 2590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664595469255664e-05,
+      "loss": 0.0349,
+      "step": 2591
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664466019417476e-05,
+      "loss": 0.0435,
+      "step": 2592
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664336569579289e-05,
+      "loss": 0.095,
+      "step": 2593
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.6642071197411e-05,
+      "loss": 0.051,
+      "step": 2594
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.664077669902913e-05,
+      "loss": 0.0333,
+      "step": 2595
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663948220064726e-05,
+      "loss": 0.0379,
+      "step": 2596
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663818770226538e-05,
+      "loss": 0.041,
+      "step": 2597
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663689320388351e-05,
+      "loss": 0.0276,
+      "step": 2598
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663559870550162e-05,
+      "loss": 0.0485,
+      "step": 2599
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663430420711974e-05,
+      "loss": 0.0755,
+      "step": 2600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663300970873786e-05,
+      "loss": 0.0621,
+      "step": 2601
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663171521035599e-05,
+      "loss": 0.0383,
+      "step": 2602
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.663042071197412e-05,
+      "loss": 0.06,
+      "step": 2603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662912621359223e-05,
+      "loss": 0.0481,
+      "step": 2604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662783171521036e-05,
+      "loss": 0.1129,
+      "step": 2605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662653721682848e-05,
+      "loss": 0.0808,
+      "step": 2606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662524271844661e-05,
+      "loss": 0.04,
+      "step": 2607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662394822006474e-05,
+      "loss": 0.039,
+      "step": 2608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662265372168285e-05,
+      "loss": 0.0508,
+      "step": 2609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.662135922330098e-05,
+      "loss": 0.0516,
+      "step": 2610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.66200647249191e-05,
+      "loss": 0.0544,
+      "step": 2611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661877022653723e-05,
+      "loss": 0.057,
+      "step": 2612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661747572815536e-05,
+      "loss": 0.0488,
+      "step": 2613
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661618122977347e-05,
+      "loss": 0.0974,
+      "step": 2614
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661488673139158e-05,
+      "loss": 0.0468,
+      "step": 2615
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661359223300971e-05,
+      "loss": 0.0649,
+      "step": 2616
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661229773462783e-05,
+      "loss": 0.0567,
+      "step": 2617
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.661100323624596e-05,
+      "loss": 0.0342,
+      "step": 2618
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660970873786407e-05,
+      "loss": 0.0918,
+      "step": 2619
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.66084142394822e-05,
+      "loss": 0.0562,
+      "step": 2620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660711974110033e-05,
+      "loss": 0.0702,
+      "step": 2621
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660582524271845e-05,
+      "loss": 0.0595,
+      "step": 2622
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660453074433658e-05,
+      "loss": 0.0303,
+      "step": 2623
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.66032362459547e-05,
+      "loss": 0.0634,
+      "step": 2624
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660194174757282e-05,
+      "loss": 0.0491,
+      "step": 2625
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.660064724919095e-05,
+      "loss": 0.025,
+      "step": 2626
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659935275080907e-05,
+      "loss": 0.045,
+      "step": 2627
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659805825242719e-05,
+      "loss": 0.0335,
+      "step": 2628
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659676375404531e-05,
+      "loss": 0.0494,
+      "step": 2629
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659546925566344e-05,
+      "loss": 0.0482,
+      "step": 2630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659417475728155e-05,
+      "loss": 0.0381,
+      "step": 2631
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659288025889968e-05,
+      "loss": 0.0397,
+      "step": 2632
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.65915857605178e-05,
+      "loss": 0.077,
+      "step": 2633
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.659029126213592e-05,
+      "loss": 0.0466,
+      "step": 2634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658899676375405e-05,
+      "loss": 0.0238,
+      "step": 2635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658770226537217e-05,
+      "loss": 0.0283,
+      "step": 2636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.65864077669903e-05,
+      "loss": 0.0472,
+      "step": 2637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658511326860841e-05,
+      "loss": 0.0663,
+      "step": 2638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658381877022654e-05,
+      "loss": 0.0309,
+      "step": 2639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658252427184466e-05,
+      "loss": 0.0623,
+      "step": 2640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.658122977346279e-05,
+      "loss": 0.0226,
+      "step": 2641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657993527508092e-05,
+      "loss": 0.0619,
+      "step": 2642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657864077669903e-05,
+      "loss": 0.0807,
+      "step": 2643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657734627831716e-05,
+      "loss": 0.0531,
+      "step": 2644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657605177993528e-05,
+      "loss": 0.0365,
+      "step": 2645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657475728155341e-05,
+      "loss": 0.0545,
+      "step": 2646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657346278317154e-05,
+      "loss": 0.0459,
+      "step": 2647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657216828478964e-05,
+      "loss": 0.0568,
+      "step": 2648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.657087378640776e-05,
+      "loss": 0.0345,
+      "step": 2649
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656957928802589e-05,
+      "loss": 0.0295,
+      "step": 2650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656828478964402e-05,
+      "loss": 0.0477,
+      "step": 2651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656699029126214e-05,
+      "loss": 0.0848,
+      "step": 2652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656569579288026e-05,
+      "loss": 0.041,
+      "step": 2653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656440129449838e-05,
+      "loss": 0.0543,
+      "step": 2654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656310679611651e-05,
+      "loss": 0.0272,
+      "step": 2655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656181229773464e-05,
+      "loss": 0.0273,
+      "step": 2656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.656051779935276e-05,
+      "loss": 0.0365,
+      "step": 2657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655922330097088e-05,
+      "loss": 0.0723,
+      "step": 2658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.6557928802589e-05,
+      "loss": 0.0613,
+      "step": 2659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655663430420713e-05,
+      "loss": 0.0719,
+      "step": 2660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655533980582526e-05,
+      "loss": 0.0493,
+      "step": 2661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655404530744337e-05,
+      "loss": 0.0336,
+      "step": 2662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.65527508090615e-05,
+      "loss": 0.0559,
+      "step": 2663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655145631067961e-05,
+      "loss": 0.0837,
+      "step": 2664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.655016181229773e-05,
+      "loss": 0.034,
+      "step": 2665
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654886731391586e-05,
+      "loss": 0.033,
+      "step": 2666
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654757281553399e-05,
+      "loss": 0.0589,
+      "step": 2667
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.65462783171521e-05,
+      "loss": 0.0313,
+      "step": 2668
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654498381877023e-05,
+      "loss": 0.0309,
+      "step": 2669
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654368932038835e-05,
+      "loss": 0.0521,
+      "step": 2670
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.654239482200648e-05,
+      "loss": 0.0926,
+      "step": 2671
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.65411003236246e-05,
+      "loss": 0.0406,
+      "step": 2672
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653980582524272e-05,
+      "loss": 0.0263,
+      "step": 2673
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653851132686085e-05,
+      "loss": 0.0345,
+      "step": 2674
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653721682847897e-05,
+      "loss": 0.0499,
+      "step": 2675
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.65359223300971e-05,
+      "loss": 0.0581,
+      "step": 2676
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653462783171521e-05,
+      "loss": 0.0612,
+      "step": 2677
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653333333333334e-05,
+      "loss": 0.0316,
+      "step": 2678
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653203883495147e-05,
+      "loss": 0.0457,
+      "step": 2679
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.653074433656958e-05,
+      "loss": 0.0708,
+      "step": 2680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.65294498381877e-05,
+      "loss": 0.024,
+      "step": 2681
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652815533980583e-05,
+      "loss": 0.066,
+      "step": 2682
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652686084142395e-05,
+      "loss": 0.0532,
+      "step": 2683
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652556634304207e-05,
+      "loss": 0.0487,
+      "step": 2684
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.65242718446602e-05,
+      "loss": 0.0741,
+      "step": 2685
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652297734627833e-05,
+      "loss": 0.065,
+      "step": 2686
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652168284789644e-05,
+      "loss": 0.054,
+      "step": 2687
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.652038834951456e-05,
+      "loss": 0.073,
+      "step": 2688
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651909385113269e-05,
+      "loss": 0.0336,
+      "step": 2689
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651779935275082e-05,
+      "loss": 0.0484,
+      "step": 2690
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651650485436894e-05,
+      "loss": 0.0671,
+      "step": 2691
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651521035598706e-05,
+      "loss": 0.0737,
+      "step": 2692
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651391585760518e-05,
+      "loss": 0.0343,
+      "step": 2693
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651262135922331e-05,
+      "loss": 0.0392,
+      "step": 2694
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651132686084144e-05,
+      "loss": 0.0227,
+      "step": 2695
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.651003236245955e-05,
+      "loss": 0.074,
+      "step": 2696
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650873786407766e-05,
+      "loss": 0.0456,
+      "step": 2697
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650744336569579e-05,
+      "loss": 0.0553,
+      "step": 2698
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650614886731392e-05,
+      "loss": 0.0572,
+      "step": 2699
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650485436893204e-05,
+      "loss": 0.0459,
+      "step": 2700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650355987055017e-05,
+      "loss": 0.0407,
+      "step": 2701
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650226537216828e-05,
+      "loss": 0.0674,
+      "step": 2702
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.650097087378641e-05,
+      "loss": 0.0376,
+      "step": 2703
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649967637540454e-05,
+      "loss": 0.0677,
+      "step": 2704
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649838187702266e-05,
+      "loss": 0.024,
+      "step": 2705
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649708737864079e-05,
+      "loss": 0.0398,
+      "step": 2706
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.64957928802589e-05,
+      "loss": 0.0641,
+      "step": 2707
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649449838187703e-05,
+      "loss": 0.1393,
+      "step": 2708
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649320388349516e-05,
+      "loss": 0.058,
+      "step": 2709
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.649190938511328e-05,
+      "loss": 0.056,
+      "step": 2710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.64906148867314e-05,
+      "loss": 0.032,
+      "step": 2711
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648932038834952e-05,
+      "loss": 0.0443,
+      "step": 2712
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648802588996763e-05,
+      "loss": 0.0611,
+      "step": 2713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648673139158576e-05,
+      "loss": 0.0396,
+      "step": 2714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648543689320389e-05,
+      "loss": 0.05,
+      "step": 2715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648414239482201e-05,
+      "loss": 0.0647,
+      "step": 2716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648284789644013e-05,
+      "loss": 0.046,
+      "step": 2717
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648155339805825e-05,
+      "loss": 0.0521,
+      "step": 2718
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.648025889967638e-05,
+      "loss": 0.0605,
+      "step": 2719
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647896440129451e-05,
+      "loss": 0.0456,
+      "step": 2720
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647766990291262e-05,
+      "loss": 0.0344,
+      "step": 2721
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647637540453075e-05,
+      "loss": 0.0414,
+      "step": 2722
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647508090614887e-05,
+      "loss": 0.0331,
+      "step": 2723
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.6473786407767e-05,
+      "loss": 0.025,
+      "step": 2724
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647249190938513e-05,
+      "loss": 0.0376,
+      "step": 2725
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.647119741100324e-05,
+      "loss": 0.0496,
+      "step": 2726
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646990291262137e-05,
+      "loss": 0.0689,
+      "step": 2727
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646860841423949e-05,
+      "loss": 0.0408,
+      "step": 2728
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.64673139158576e-05,
+      "loss": 0.0866,
+      "step": 2729
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646601941747573e-05,
+      "loss": 0.0442,
+      "step": 2730
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646472491909385e-05,
+      "loss": 0.078,
+      "step": 2731
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646343042071197e-05,
+      "loss": 0.0572,
+      "step": 2732
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.64621359223301e-05,
+      "loss": 0.0552,
+      "step": 2733
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.646084142394823e-05,
+      "loss": 0.032,
+      "step": 2734
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645954692556635e-05,
+      "loss": 0.0663,
+      "step": 2735
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645825242718446e-05,
+      "loss": 0.0723,
+      "step": 2736
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645695792880259e-05,
+      "loss": 0.0283,
+      "step": 2737
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645566343042072e-05,
+      "loss": 0.0971,
+      "step": 2738
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645436893203884e-05,
+      "loss": 0.0801,
+      "step": 2739
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645307443365697e-05,
+      "loss": 0.0366,
+      "step": 2740
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645177993527508e-05,
+      "loss": 0.0258,
+      "step": 2741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.645048543689321e-05,
+      "loss": 0.0632,
+      "step": 2742
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644919093851134e-05,
+      "loss": 0.0311,
+      "step": 2743
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644789644012946e-05,
+      "loss": 0.057,
+      "step": 2744
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644660194174758e-05,
+      "loss": 0.0595,
+      "step": 2745
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644530744336569e-05,
+      "loss": 0.0629,
+      "step": 2746
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644401294498382e-05,
+      "loss": 0.0411,
+      "step": 2747
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644271844660194e-05,
+      "loss": 0.0666,
+      "step": 2748
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.644142394822007e-05,
+      "loss": 0.0496,
+      "step": 2749
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.64401294498382e-05,
+      "loss": 0.0849,
+      "step": 2750
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643883495145631e-05,
+      "loss": 0.0697,
+      "step": 2751
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643754045307444e-05,
+      "loss": 0.0698,
+      "step": 2752
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643624595469256e-05,
+      "loss": 0.0461,
+      "step": 2753
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643495145631069e-05,
+      "loss": 0.0298,
+      "step": 2754
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.64336569579288e-05,
+      "loss": 0.0247,
+      "step": 2755
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643236245954693e-05,
+      "loss": 0.0949,
+      "step": 2756
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.643106796116506e-05,
+      "loss": 0.0535,
+      "step": 2757
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642977346278318e-05,
+      "loss": 0.0464,
+      "step": 2758
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642847896440131e-05,
+      "loss": 0.0301,
+      "step": 2759
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642718446601942e-05,
+      "loss": 0.0344,
+      "step": 2760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642588996763755e-05,
+      "loss": 0.0463,
+      "step": 2761
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642459546925566e-05,
+      "loss": 0.0343,
+      "step": 2762
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642330097087379e-05,
+      "loss": 0.0817,
+      "step": 2763
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642200647249191e-05,
+      "loss": 0.0534,
+      "step": 2764
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.642071197411003e-05,
+      "loss": 0.0676,
+      "step": 2765
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641941747572815e-05,
+      "loss": 0.0605,
+      "step": 2766
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641812297734628e-05,
+      "loss": 0.0573,
+      "step": 2767
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641682847896441e-05,
+      "loss": 0.0623,
+      "step": 2768
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641553398058253e-05,
+      "loss": 0.0644,
+      "step": 2769
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641423948220065e-05,
+      "loss": 0.0797,
+      "step": 2770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641294498381877e-05,
+      "loss": 0.0492,
+      "step": 2771
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.64116504854369e-05,
+      "loss": 0.0473,
+      "step": 2772
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.641035598705503e-05,
+      "loss": 0.0287,
+      "step": 2773
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640906148867315e-05,
+      "loss": 0.0524,
+      "step": 2774
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640776699029127e-05,
+      "loss": 0.0537,
+      "step": 2775
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640647249190939e-05,
+      "loss": 0.0851,
+      "step": 2776
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640517799352752e-05,
+      "loss": 0.0524,
+      "step": 2777
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640388349514563e-05,
+      "loss": 0.0534,
+      "step": 2778
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640258899676376e-05,
+      "loss": 0.0279,
+      "step": 2779
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.640129449838187e-05,
+      "loss": 0.0603,
+      "step": 2780
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.64e-05,
+      "loss": 0.0838,
+      "step": 2781
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639870550161813e-05,
+      "loss": 0.061,
+      "step": 2782
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639741100323625e-05,
+      "loss": 0.0311,
+      "step": 2783
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639611650485438e-05,
+      "loss": 0.0282,
+      "step": 2784
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639482200647249e-05,
+      "loss": 0.0594,
+      "step": 2785
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639352750809062e-05,
+      "loss": 0.0386,
+      "step": 2786
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639223300970874e-05,
+      "loss": 0.043,
+      "step": 2787
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.639093851132687e-05,
+      "loss": 0.0266,
+      "step": 2788
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638964401294498e-05,
+      "loss": 0.0465,
+      "step": 2789
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638834951456311e-05,
+      "loss": 0.0744,
+      "step": 2790
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638705501618124e-05,
+      "loss": 0.0878,
+      "step": 2791
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638576051779936e-05,
+      "loss": 0.0438,
+      "step": 2792
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638446601941749e-05,
+      "loss": 0.0596,
+      "step": 2793
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63831715210356e-05,
+      "loss": 0.039,
+      "step": 2794
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638187702265372e-05,
+      "loss": 0.0884,
+      "step": 2795
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.638058252427184e-05,
+      "loss": 0.0236,
+      "step": 2796
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637928802588997e-05,
+      "loss": 0.1215,
+      "step": 2797
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.63779935275081e-05,
+      "loss": 0.0409,
+      "step": 2798
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637669902912622e-05,
+      "loss": 0.0513,
+      "step": 2799
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637540453074434e-05,
+      "loss": 0.0699,
+      "step": 2800
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637411003236246e-05,
+      "loss": 0.0546,
+      "step": 2801
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637281553398059e-05,
+      "loss": 0.0318,
+      "step": 2802
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637152103559872e-05,
+      "loss": 0.0325,
+      "step": 2803
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.637022653721683e-05,
+      "loss": 0.0456,
+      "step": 2804
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636893203883496e-05,
+      "loss": 0.0556,
+      "step": 2805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636763754045308e-05,
+      "loss": 0.0491,
+      "step": 2806
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636634304207121e-05,
+      "loss": 0.0404,
+      "step": 2807
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636504854368934e-05,
+      "loss": 0.056,
+      "step": 2808
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636375404530745e-05,
+      "loss": 0.0345,
+      "step": 2809
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636245954692557e-05,
+      "loss": 0.046,
+      "step": 2810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.636116504854369e-05,
+      "loss": 0.0968,
+      "step": 2811
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635987055016181e-05,
+      "loss": 0.0424,
+      "step": 2812
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635857605177994e-05,
+      "loss": 0.0195,
+      "step": 2813
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635728155339805e-05,
+      "loss": 0.0596,
+      "step": 2814
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635598705501618e-05,
+      "loss": 0.0306,
+      "step": 2815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635469255663431e-05,
+      "loss": 0.0576,
+      "step": 2816
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635339805825243e-05,
+      "loss": 0.0556,
+      "step": 2817
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635210355987056e-05,
+      "loss": 0.0766,
+      "step": 2818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.635080906148867e-05,
+      "loss": 0.0533,
+      "step": 2819
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63495145631068e-05,
+      "loss": 0.041,
+      "step": 2820
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634822006472493e-05,
+      "loss": 0.0367,
+      "step": 2821
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634692556634305e-05,
+      "loss": 0.0606,
+      "step": 2822
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634563106796117e-05,
+      "loss": 0.0638,
+      "step": 2823
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634433656957929e-05,
+      "loss": 0.0552,
+      "step": 2824
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634304207119742e-05,
+      "loss": 0.042,
+      "step": 2825
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634174757281555e-05,
+      "loss": 0.1127,
+      "step": 2826
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.634045307443366e-05,
+      "loss": 0.0353,
+      "step": 2827
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633915857605179e-05,
+      "loss": 0.0523,
+      "step": 2828
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63378640776699e-05,
+      "loss": 0.0599,
+      "step": 2829
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633656957928803e-05,
+      "loss": 0.0357,
+      "step": 2830
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633527508090615e-05,
+      "loss": 0.034,
+      "step": 2831
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633398058252428e-05,
+      "loss": 0.0387,
+      "step": 2832
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.63326860841424e-05,
+      "loss": 0.0257,
+      "step": 2833
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633139158576052e-05,
+      "loss": 0.042,
+      "step": 2834
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.633009708737864e-05,
+      "loss": 0.0419,
+      "step": 2835
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632880258899677e-05,
+      "loss": 0.0669,
+      "step": 2836
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.63275080906149e-05,
+      "loss": 0.0515,
+      "step": 2837
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632621359223301e-05,
+      "loss": 0.0792,
+      "step": 2838
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632491909385114e-05,
+      "loss": 0.0421,
+      "step": 2839
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632362459546926e-05,
+      "loss": 0.0233,
+      "step": 2840
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632233009708739e-05,
+      "loss": 0.0837,
+      "step": 2841
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.632103559870552e-05,
+      "loss": 0.0941,
+      "step": 2842
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631974110032363e-05,
+      "loss": 0.0409,
+      "step": 2843
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631844660194174e-05,
+      "loss": 0.0291,
+      "step": 2844
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631715210355987e-05,
+      "loss": 0.0219,
+      "step": 2845
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.6315857605178e-05,
+      "loss": 0.061,
+      "step": 2846
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631456310679612e-05,
+      "loss": 0.0406,
+      "step": 2847
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631326860841424e-05,
+      "loss": 0.0402,
+      "step": 2848
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631197411003236e-05,
+      "loss": 0.0775,
+      "step": 2849
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.631067961165049e-05,
+      "loss": 0.0721,
+      "step": 2850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630938511326862e-05,
+      "loss": 0.0587,
+      "step": 2851
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630809061488674e-05,
+      "loss": 0.0366,
+      "step": 2852
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630679611650486e-05,
+      "loss": 0.0501,
+      "step": 2853
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630550161812298e-05,
+      "loss": 0.0476,
+      "step": 2854
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630420711974111e-05,
+      "loss": 0.0761,
+      "step": 2855
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630291262135924e-05,
+      "loss": 0.0317,
+      "step": 2856
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630161812297736e-05,
+      "loss": 0.038,
+      "step": 2857
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.630032362459547e-05,
+      "loss": 0.052,
+      "step": 2858
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629902912621359e-05,
+      "loss": 0.0319,
+      "step": 2859
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629773462783171e-05,
+      "loss": 0.0503,
+      "step": 2860
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629644012944984e-05,
+      "loss": 0.0371,
+      "step": 2861
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629514563106797e-05,
+      "loss": 0.0529,
+      "step": 2862
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629385113268608e-05,
+      "loss": 0.0569,
+      "step": 2863
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629255663430421e-05,
+      "loss": 0.0407,
+      "step": 2864
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.629126213592233e-05,
+      "loss": 0.0308,
+      "step": 2865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628996763754046e-05,
+      "loss": 0.0387,
+      "step": 2866
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628867313915859e-05,
+      "loss": 0.0599,
+      "step": 2867
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62873786407767e-05,
+      "loss": 0.0412,
+      "step": 2868
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628608414239483e-05,
+      "loss": 0.0515,
+      "step": 2869
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628478964401295e-05,
+      "loss": 0.0353,
+      "step": 2870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628349514563108e-05,
+      "loss": 0.0472,
+      "step": 2871
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628220064724919e-05,
+      "loss": 0.0741,
+      "step": 2872
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.628090614886732e-05,
+      "loss": 0.048,
+      "step": 2873
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627961165048545e-05,
+      "loss": 0.0388,
+      "step": 2874
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627831715210357e-05,
+      "loss": 0.046,
+      "step": 2875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627702265372169e-05,
+      "loss": 0.0307,
+      "step": 2876
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627572815533981e-05,
+      "loss": 0.0643,
+      "step": 2877
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627443365695793e-05,
+      "loss": 0.0377,
+      "step": 2878
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627313915857605e-05,
+      "loss": 0.0607,
+      "step": 2879
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.627184466019418e-05,
+      "loss": 0.0279,
+      "step": 2880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.62705501618123e-05,
+      "loss": 0.0612,
+      "step": 2881
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626925566343042e-05,
+      "loss": 0.0336,
+      "step": 2882
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626796116504854e-05,
+      "loss": 0.047,
+      "step": 2883
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626666666666667e-05,
+      "loss": 0.0557,
+      "step": 2884
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62653721682848e-05,
+      "loss": 0.0306,
+      "step": 2885
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626407766990292e-05,
+      "loss": 0.0353,
+      "step": 2886
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626278317152104e-05,
+      "loss": 0.0885,
+      "step": 2887
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626148867313916e-05,
+      "loss": 0.0496,
+      "step": 2888
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.626019417475729e-05,
+      "loss": 0.0424,
+      "step": 2889
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625889967637542e-05,
+      "loss": 0.0866,
+      "step": 2890
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625760517799354e-05,
+      "loss": 0.0423,
+      "step": 2891
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625631067961164e-05,
+      "loss": 0.048,
+      "step": 2892
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625501618122977e-05,
+      "loss": 0.0486,
+      "step": 2893
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.62537216828479e-05,
+      "loss": 0.034,
+      "step": 2894
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625242718446602e-05,
+      "loss": 0.0814,
+      "step": 2895
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.625113268608415e-05,
+      "loss": 0.0509,
+      "step": 2896
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624983818770226e-05,
+      "loss": 0.0384,
+      "step": 2897
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624854368932039e-05,
+      "loss": 0.0408,
+      "step": 2898
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624724919093852e-05,
+      "loss": 0.0404,
+      "step": 2899
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624595469255664e-05,
+      "loss": 0.0339,
+      "step": 2900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624466019417477e-05,
+      "loss": 0.0286,
+      "step": 2901
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624336569579288e-05,
+      "loss": 0.0442,
+      "step": 2902
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624207119741101e-05,
+      "loss": 0.0995,
+      "step": 2903
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.624077669902914e-05,
+      "loss": 0.0435,
+      "step": 2904
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623948220064726e-05,
+      "loss": 0.0609,
+      "step": 2905
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623818770226537e-05,
+      "loss": 0.04,
+      "step": 2906
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.62368932038835e-05,
+      "loss": 0.0261,
+      "step": 2907
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623559870550161e-05,
+      "loss": 0.089,
+      "step": 2908
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623430420711974e-05,
+      "loss": 0.054,
+      "step": 2909
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623300970873787e-05,
+      "loss": 0.0941,
+      "step": 2910
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.6231715210356e-05,
+      "loss": 0.1114,
+      "step": 2911
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.623042071197411e-05,
+      "loss": 0.0871,
+      "step": 2912
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622912621359223e-05,
+      "loss": 0.0461,
+      "step": 2913
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622783171521036e-05,
+      "loss": 0.0442,
+      "step": 2914
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622653721682849e-05,
+      "loss": 0.044,
+      "step": 2915
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62252427184466e-05,
+      "loss": 0.0676,
+      "step": 2916
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622394822006473e-05,
+      "loss": 0.1836,
+      "step": 2917
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622265372168285e-05,
+      "loss": 0.0454,
+      "step": 2918
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.622135922330098e-05,
+      "loss": 0.0504,
+      "step": 2919
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62200647249191e-05,
+      "loss": 0.0456,
+      "step": 2920
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621877022653722e-05,
+      "loss": 0.0347,
+      "step": 2921
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621747572815535e-05,
+      "loss": 0.0688,
+      "step": 2922
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621618122977347e-05,
+      "loss": 0.0686,
+      "step": 2923
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.62148867313916e-05,
+      "loss": 0.0612,
+      "step": 2924
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621359223300971e-05,
+      "loss": 0.0534,
+      "step": 2925
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621229773462784e-05,
+      "loss": 0.0844,
+      "step": 2926
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.621100323624595e-05,
+      "loss": 0.0722,
+      "step": 2927
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620970873786408e-05,
+      "loss": 0.0426,
+      "step": 2928
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.62084142394822e-05,
+      "loss": 0.0454,
+      "step": 2929
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620711974110033e-05,
+      "loss": 0.05,
+      "step": 2930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620582524271844e-05,
+      "loss": 0.0672,
+      "step": 2931
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620453074433657e-05,
+      "loss": 0.0669,
+      "step": 2932
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.62032362459547e-05,
+      "loss": 0.058,
+      "step": 2933
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620194174757282e-05,
+      "loss": 0.0601,
+      "step": 2934
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.620064724919095e-05,
+      "loss": 0.0417,
+      "step": 2935
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619935275080906e-05,
+      "loss": 0.0305,
+      "step": 2936
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619805825242719e-05,
+      "loss": 0.0315,
+      "step": 2937
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619676375404532e-05,
+      "loss": 0.0629,
+      "step": 2938
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619546925566344e-05,
+      "loss": 0.0377,
+      "step": 2939
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619417475728156e-05,
+      "loss": 0.0701,
+      "step": 2940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619288025889967e-05,
+      "loss": 0.0694,
+      "step": 2941
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.61915857605178e-05,
+      "loss": 0.0415,
+      "step": 2942
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.619029126213592e-05,
+      "loss": 0.0413,
+      "step": 2943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618899676375405e-05,
+      "loss": 0.1203,
+      "step": 2944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618770226537218e-05,
+      "loss": 0.1242,
+      "step": 2945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618640776699029e-05,
+      "loss": 0.0505,
+      "step": 2946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618511326860842e-05,
+      "loss": 0.0401,
+      "step": 2947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618381877022654e-05,
+      "loss": 0.0367,
+      "step": 2948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.618252427184467e-05,
+      "loss": 0.0492,
+      "step": 2949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.61812297734628e-05,
+      "loss": 0.0566,
+      "step": 2950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617993527508091e-05,
+      "loss": 0.0345,
+      "step": 2951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617864077669903e-05,
+      "loss": 0.0525,
+      "step": 2952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617734627831716e-05,
+      "loss": 0.0393,
+      "step": 2953
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617605177993529e-05,
+      "loss": 0.0279,
+      "step": 2954
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.61747572815534e-05,
+      "loss": 0.0538,
+      "step": 2955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617346278317153e-05,
+      "loss": 0.0595,
+      "step": 2956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617216828478964e-05,
+      "loss": 0.046,
+      "step": 2957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.617087378640777e-05,
+      "loss": 0.0374,
+      "step": 2958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61695792880259e-05,
+      "loss": 0.0503,
+      "step": 2959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616828478964402e-05,
+      "loss": 0.0452,
+      "step": 2960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616699029126213e-05,
+      "loss": 0.0457,
+      "step": 2961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616569579288026e-05,
+      "loss": 0.0612,
+      "step": 2962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616440129449839e-05,
+      "loss": 0.0758,
+      "step": 2963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616310679611651e-05,
+      "loss": 0.0328,
+      "step": 2964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616181229773463e-05,
+      "loss": 0.0506,
+      "step": 2965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.616051779935275e-05,
+      "loss": 0.0559,
+      "step": 2966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615922330097088e-05,
+      "loss": 0.051,
+      "step": 2967
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.6157928802589e-05,
+      "loss": 0.053,
+      "step": 2968
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615663430420713e-05,
+      "loss": 0.0294,
+      "step": 2969
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615533980582525e-05,
+      "loss": 0.0213,
+      "step": 2970
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615404530744337e-05,
+      "loss": 0.069,
+      "step": 2971
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.61527508090615e-05,
+      "loss": 0.0555,
+      "step": 2972
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615145631067963e-05,
+      "loss": 0.0234,
+      "step": 2973
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.615016181229774e-05,
+      "loss": 0.04,
+      "step": 2974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614886731391585e-05,
+      "loss": 0.0606,
+      "step": 2975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614757281553398e-05,
+      "loss": 0.0377,
+      "step": 2976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.61462783171521e-05,
+      "loss": 0.0455,
+      "step": 2977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614498381877023e-05,
+      "loss": 0.0306,
+      "step": 2978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614368932038836e-05,
+      "loss": 0.033,
+      "step": 2979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.614239482200647e-05,
+      "loss": 0.0463,
+      "step": 2980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.61411003236246e-05,
+      "loss": 0.0521,
+      "step": 2981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613980582524272e-05,
+      "loss": 0.0782,
+      "step": 2982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613851132686085e-05,
+      "loss": 0.0918,
+      "step": 2983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613721682847898e-05,
+      "loss": 0.0566,
+      "step": 2984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613592233009709e-05,
+      "loss": 0.0413,
+      "step": 2985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613462783171522e-05,
+      "loss": 0.0492,
+      "step": 2986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613333333333334e-05,
+      "loss": 0.1565,
+      "step": 2987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613203883495147e-05,
+      "loss": 0.0328,
+      "step": 2988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.613074433656958e-05,
+      "loss": 0.0831,
+      "step": 2989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.61294498381877e-05,
+      "loss": 0.0332,
+      "step": 2990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612815533980582e-05,
+      "loss": 0.0445,
+      "step": 2991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612686084142395e-05,
+      "loss": 0.0333,
+      "step": 2992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612556634304208e-05,
+      "loss": 0.118,
+      "step": 2993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.61242718446602e-05,
+      "loss": 0.0298,
+      "step": 2994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612297734627832e-05,
+      "loss": 0.0509,
+      "step": 2995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612168284789644e-05,
+      "loss": 0.0568,
+      "step": 2996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.612038834951457e-05,
+      "loss": 0.0337,
+      "step": 2997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.61190938511327e-05,
+      "loss": 0.0933,
+      "step": 2998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611779935275081e-05,
+      "loss": 0.0581,
+      "step": 2999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611650485436893e-05,
+      "loss": 0.043,
+      "step": 3000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 0.0690242126584053,
+      "eval_mean_accuracy": 0.6971518497190766,
+      "eval_mean_iou": 0.6232414436607869,
+      "eval_overall_accuracy": 0.9760269145965577,
+      "eval_per_category_accuracy": [
+        0.992147723855591,
+        0.5985086487906774,
+        0.5983058015648758,
+        0.5996452246651622
+      ],
+      "eval_per_category_iou": [
+        0.977313306772953,
+        0.5171795551108308,
+        0.519708449764995,
+        0.478764462994369
+      ],
+      "eval_runtime": 125.9236,
+      "eval_samples_per_second": 15.883,
+      "eval_steps_per_second": 1.985,
+      "step": 3000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611521035598706e-05,
+      "loss": 0.0575,
+      "step": 3001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611391585760519e-05,
+      "loss": 0.0495,
+      "step": 3002
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611262135922331e-05,
+      "loss": 0.0477,
+      "step": 3003
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611132686084143e-05,
+      "loss": 0.0809,
+      "step": 3004
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.611003236245955e-05,
+      "loss": 0.1108,
+      "step": 3005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610873786407767e-05,
+      "loss": 0.0539,
+      "step": 3006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.61074433656958e-05,
+      "loss": 0.0363,
+      "step": 3007
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610614886731392e-05,
+      "loss": 0.0444,
+      "step": 3008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610485436893203e-05,
+      "loss": 0.0585,
+      "step": 3009
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610355987055016e-05,
+      "loss": 0.0419,
+      "step": 3010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610226537216829e-05,
+      "loss": 0.0836,
+      "step": 3011
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.610097087378641e-05,
+      "loss": 0.0439,
+      "step": 3012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609967637540454e-05,
+      "loss": 0.0207,
+      "step": 3013
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609838187702265e-05,
+      "loss": 0.0615,
+      "step": 3014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609708737864078e-05,
+      "loss": 0.0966,
+      "step": 3015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.60957928802589e-05,
+      "loss": 0.0319,
+      "step": 3016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609449838187703e-05,
+      "loss": 0.0578,
+      "step": 3017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609320388349516e-05,
+      "loss": 0.0724,
+      "step": 3018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.609190938511327e-05,
+      "loss": 0.0677,
+      "step": 3019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.60906148867314e-05,
+      "loss": 0.0496,
+      "step": 3020
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608932038834953e-05,
+      "loss": 0.0938,
+      "step": 3021
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608802588996764e-05,
+      "loss": 0.0286,
+      "step": 3022
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608673139158577e-05,
+      "loss": 0.0552,
+      "step": 3023
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608543689320388e-05,
+      "loss": 0.0336,
+      "step": 3024
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.6084142394822e-05,
+      "loss": 0.0404,
+      "step": 3025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608284789644013e-05,
+      "loss": 0.0541,
+      "step": 3026
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608155339805826e-05,
+      "loss": 0.0699,
+      "step": 3027
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.608025889967638e-05,
+      "loss": 0.0526,
+      "step": 3028
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.60789644012945e-05,
+      "loss": 0.0766,
+      "step": 3029
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607766990291262e-05,
+      "loss": 0.0626,
+      "step": 3030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607637540453075e-05,
+      "loss": 0.0381,
+      "step": 3031
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607508090614888e-05,
+      "loss": 0.0482,
+      "step": 3032
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607378640776699e-05,
+      "loss": 0.0427,
+      "step": 3033
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607249190938512e-05,
+      "loss": 0.0451,
+      "step": 3034
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.607119741100324e-05,
+      "loss": 0.0337,
+      "step": 3035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606990291262137e-05,
+      "loss": 0.0331,
+      "step": 3036
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.60686084142395e-05,
+      "loss": 0.0586,
+      "step": 3037
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606731391585761e-05,
+      "loss": 0.0558,
+      "step": 3038
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606601941747572e-05,
+      "loss": 0.0256,
+      "step": 3039
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606472491909385e-05,
+      "loss": 0.0572,
+      "step": 3040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606343042071198e-05,
+      "loss": 0.058,
+      "step": 3041
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.60621359223301e-05,
+      "loss": 0.0883,
+      "step": 3042
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.606084142394823e-05,
+      "loss": 0.0461,
+      "step": 3043
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605954692556634e-05,
+      "loss": 0.0509,
+      "step": 3044
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605825242718447e-05,
+      "loss": 0.0481,
+      "step": 3045
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.60569579288026e-05,
+      "loss": 0.0447,
+      "step": 3046
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605566343042072e-05,
+      "loss": 0.0301,
+      "step": 3047
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605436893203883e-05,
+      "loss": 0.0409,
+      "step": 3048
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605307443365696e-05,
+      "loss": 0.0267,
+      "step": 3049
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605177993527509e-05,
+      "loss": 0.035,
+      "step": 3050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.605048543689321e-05,
+      "loss": 0.0642,
+      "step": 3051
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604919093851134e-05,
+      "loss": 0.0374,
+      "step": 3052
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604789644012945e-05,
+      "loss": 0.0462,
+      "step": 3053
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604660194174758e-05,
+      "loss": 0.054,
+      "step": 3054
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.60453074433657e-05,
+      "loss": 0.0314,
+      "step": 3055
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604401294498382e-05,
+      "loss": 0.0411,
+      "step": 3056
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604271844660195e-05,
+      "loss": 0.036,
+      "step": 3057
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604142394822006e-05,
+      "loss": 0.0443,
+      "step": 3058
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.604012944983819e-05,
+      "loss": 0.0429,
+      "step": 3059
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603883495145631e-05,
+      "loss": 0.0382,
+      "step": 3060
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603754045307444e-05,
+      "loss": 0.0453,
+      "step": 3061
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603624595469257e-05,
+      "loss": 0.033,
+      "step": 3062
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603495145631068e-05,
+      "loss": 0.0604,
+      "step": 3063
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.60336569579288e-05,
+      "loss": 0.0373,
+      "step": 3064
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603236245954693e-05,
+      "loss": 0.0442,
+      "step": 3065
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.603106796116506e-05,
+      "loss": 0.0284,
+      "step": 3066
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602977346278319e-05,
+      "loss": 0.0284,
+      "step": 3067
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.60284789644013e-05,
+      "loss": 0.0596,
+      "step": 3068
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602718446601943e-05,
+      "loss": 0.0793,
+      "step": 3069
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602588996763755e-05,
+      "loss": 0.0308,
+      "step": 3070
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602459546925567e-05,
+      "loss": 0.0348,
+      "step": 3071
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602330097087379e-05,
+      "loss": 0.0567,
+      "step": 3072
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.60220064724919e-05,
+      "loss": 0.038,
+      "step": 3073
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.602071197411003e-05,
+      "loss": 0.0472,
+      "step": 3074
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601941747572816e-05,
+      "loss": 0.0226,
+      "step": 3075
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601812297734628e-05,
+      "loss": 0.0445,
+      "step": 3076
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601682847896441e-05,
+      "loss": 0.0425,
+      "step": 3077
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601553398058252e-05,
+      "loss": 0.0354,
+      "step": 3078
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601423948220065e-05,
+      "loss": 0.0323,
+      "step": 3079
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601294498381878e-05,
+      "loss": 0.0337,
+      "step": 3080
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.60116504854369e-05,
+      "loss": 0.0534,
+      "step": 3081
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.601035598705502e-05,
+      "loss": 0.0592,
+      "step": 3082
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600906148867314e-05,
+      "loss": 0.0557,
+      "step": 3083
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600776699029127e-05,
+      "loss": 0.0455,
+      "step": 3084
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.60064724919094e-05,
+      "loss": 0.0854,
+      "step": 3085
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600517799352752e-05,
+      "loss": 0.0337,
+      "step": 3086
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600388349514564e-05,
+      "loss": 0.0384,
+      "step": 3087
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600258899676375e-05,
+      "loss": 0.1169,
+      "step": 3088
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.600129449838188e-05,
+      "loss": 0.0363,
+      "step": 3089
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.6e-05,
+      "loss": 0.0346,
+      "step": 3090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599870550161813e-05,
+      "loss": 0.0479,
+      "step": 3091
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599741100323624e-05,
+      "loss": 0.0304,
+      "step": 3092
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599611650485437e-05,
+      "loss": 0.0671,
+      "step": 3093
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.59948220064725e-05,
+      "loss": 0.0224,
+      "step": 3094
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599352750809062e-05,
+      "loss": 0.0343,
+      "step": 3095
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599223300970875e-05,
+      "loss": 0.0519,
+      "step": 3096
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.599093851132686e-05,
+      "loss": 0.0491,
+      "step": 3097
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598964401294499e-05,
+      "loss": 0.0432,
+      "step": 3098
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598834951456311e-05,
+      "loss": 0.0226,
+      "step": 3099
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598705501618124e-05,
+      "loss": 0.0605,
+      "step": 3100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598576051779937e-05,
+      "loss": 0.0611,
+      "step": 3101
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598446601941748e-05,
+      "loss": 0.0429,
+      "step": 3102
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598317152103561e-05,
+      "loss": 0.0377,
+      "step": 3103
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598187702265372e-05,
+      "loss": 0.0567,
+      "step": 3104
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.598058252427185e-05,
+      "loss": 0.0418,
+      "step": 3105
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597928802588997e-05,
+      "loss": 0.0286,
+      "step": 3106
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597799352750809e-05,
+      "loss": 0.0245,
+      "step": 3107
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597669902912621e-05,
+      "loss": 0.0285,
+      "step": 3108
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597540453074434e-05,
+      "loss": 0.0471,
+      "step": 3109
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597411003236247e-05,
+      "loss": 0.0263,
+      "step": 3110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597281553398059e-05,
+      "loss": 0.0297,
+      "step": 3111
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.59715210355987e-05,
+      "loss": 0.04,
+      "step": 3112
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.597022653721683e-05,
+      "loss": 0.0448,
+      "step": 3113
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596893203883496e-05,
+      "loss": 0.0249,
+      "step": 3114
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596763754045309e-05,
+      "loss": 0.0341,
+      "step": 3115
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.59663430420712e-05,
+      "loss": 0.0532,
+      "step": 3116
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596504854368933e-05,
+      "loss": 0.0489,
+      "step": 3117
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596375404530745e-05,
+      "loss": 0.0322,
+      "step": 3118
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596245954692558e-05,
+      "loss": 0.0627,
+      "step": 3119
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.596116504854369e-05,
+      "loss": 0.0286,
+      "step": 3120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595987055016182e-05,
+      "loss": 0.024,
+      "step": 3121
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595857605177993e-05,
+      "loss": 0.041,
+      "step": 3122
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595728155339806e-05,
+      "loss": 0.0189,
+      "step": 3123
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595598705501618e-05,
+      "loss": 0.0204,
+      "step": 3124
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595469255663431e-05,
+      "loss": 0.0226,
+      "step": 3125
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595339805825242e-05,
+      "loss": 0.0361,
+      "step": 3126
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595210355987055e-05,
+      "loss": 0.0431,
+      "step": 3127
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.595080906148868e-05,
+      "loss": 0.062,
+      "step": 3128
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.59495145631068e-05,
+      "loss": 0.0539,
+      "step": 3129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594822006472493e-05,
+      "loss": 0.026,
+      "step": 3130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594692556634304e-05,
+      "loss": 0.0476,
+      "step": 3131
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594563106796117e-05,
+      "loss": 0.0449,
+      "step": 3132
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.59443365695793e-05,
+      "loss": 0.0635,
+      "step": 3133
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594304207119742e-05,
+      "loss": 0.06,
+      "step": 3134
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594174757281555e-05,
+      "loss": 0.0946,
+      "step": 3135
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.594045307443366e-05,
+      "loss": 0.0603,
+      "step": 3136
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593915857605178e-05,
+      "loss": 0.038,
+      "step": 3137
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.59378640776699e-05,
+      "loss": 0.0616,
+      "step": 3138
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593656957928803e-05,
+      "loss": 0.0213,
+      "step": 3139
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593527508090616e-05,
+      "loss": 0.0412,
+      "step": 3140
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593398058252427e-05,
+      "loss": 0.0314,
+      "step": 3141
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.59326860841424e-05,
+      "loss": 0.0184,
+      "step": 3142
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593139158576052e-05,
+      "loss": 0.038,
+      "step": 3143
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.593009708737865e-05,
+      "loss": 0.0211,
+      "step": 3144
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592880258899677e-05,
+      "loss": 0.0269,
+      "step": 3145
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592750809061489e-05,
+      "loss": 0.0156,
+      "step": 3146
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592621359223301e-05,
+      "loss": 0.1307,
+      "step": 3147
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592491909385114e-05,
+      "loss": 0.0413,
+      "step": 3148
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592362459546927e-05,
+      "loss": 0.0715,
+      "step": 3149
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592233009708738e-05,
+      "loss": 0.0389,
+      "step": 3150
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.592103559870551e-05,
+      "loss": 0.0619,
+      "step": 3151
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591974110032363e-05,
+      "loss": 0.0409,
+      "step": 3152
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591844660194175e-05,
+      "loss": 0.0384,
+      "step": 3153
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591715210355987e-05,
+      "loss": 0.0419,
+      "step": 3154
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.5915857605178e-05,
+      "loss": 0.0185,
+      "step": 3155
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591456310679611e-05,
+      "loss": 0.0259,
+      "step": 3156
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591326860841424e-05,
+      "loss": 0.0422,
+      "step": 3157
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591197411003237e-05,
+      "loss": 0.0645,
+      "step": 3158
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.591067961165049e-05,
+      "loss": 0.0232,
+      "step": 3159
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590938511326862e-05,
+      "loss": 0.055,
+      "step": 3160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590809061488673e-05,
+      "loss": 0.0191,
+      "step": 3161
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590679611650486e-05,
+      "loss": 0.0338,
+      "step": 3162
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590550161812299e-05,
+      "loss": 0.0325,
+      "step": 3163
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590420711974111e-05,
+      "loss": 0.0374,
+      "step": 3164
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590291262135923e-05,
+      "loss": 0.0493,
+      "step": 3165
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590161812297735e-05,
+      "loss": 0.0547,
+      "step": 3166
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.590032362459548e-05,
+      "loss": 0.0373,
+      "step": 3167
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.58990291262136e-05,
+      "loss": 0.0265,
+      "step": 3168
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589773462783172e-05,
+      "loss": 0.0329,
+      "step": 3169
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589644012944984e-05,
+      "loss": 0.0712,
+      "step": 3170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589514563106796e-05,
+      "loss": 0.0231,
+      "step": 3171
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589385113268608e-05,
+      "loss": 0.0403,
+      "step": 3172
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589255663430421e-05,
+      "loss": 0.0263,
+      "step": 3173
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.589126213592234e-05,
+      "loss": 0.0291,
+      "step": 3174
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588996763754045e-05,
+      "loss": 0.0234,
+      "step": 3175
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588867313915858e-05,
+      "loss": 0.0251,
+      "step": 3176
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.58873786407767e-05,
+      "loss": 0.0269,
+      "step": 3177
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588608414239483e-05,
+      "loss": 0.0302,
+      "step": 3178
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588478964401296e-05,
+      "loss": 0.0378,
+      "step": 3179
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588349514563107e-05,
+      "loss": 0.0362,
+      "step": 3180
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.58822006472492e-05,
+      "loss": 0.0627,
+      "step": 3181
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.588090614886732e-05,
+      "loss": 0.0329,
+      "step": 3182
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587961165048545e-05,
+      "loss": 0.0318,
+      "step": 3183
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587831715210356e-05,
+      "loss": 0.0192,
+      "step": 3184
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587702265372168e-05,
+      "loss": 0.0403,
+      "step": 3185
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.58757281553398e-05,
+      "loss": 0.0256,
+      "step": 3186
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587443365695793e-05,
+      "loss": 0.0323,
+      "step": 3187
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587313915857606e-05,
+      "loss": 0.0239,
+      "step": 3188
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.587184466019418e-05,
+      "loss": 0.0342,
+      "step": 3189
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.58705501618123e-05,
+      "loss": 0.0336,
+      "step": 3190
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586925566343042e-05,
+      "loss": 0.033,
+      "step": 3191
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586796116504855e-05,
+      "loss": 0.0274,
+      "step": 3192
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586666666666667e-05,
+      "loss": 0.0858,
+      "step": 3193
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.58653721682848e-05,
+      "loss": 0.0243,
+      "step": 3194
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586407766990291e-05,
+      "loss": 0.0923,
+      "step": 3195
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586278317152104e-05,
+      "loss": 0.0538,
+      "step": 3196
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.586148867313917e-05,
+      "loss": 0.0253,
+      "step": 3197
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.58601941747573e-05,
+      "loss": 0.0323,
+      "step": 3198
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585889967637541e-05,
+      "loss": 0.074,
+      "step": 3199
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585760517799353e-05,
+      "loss": 0.0292,
+      "step": 3200
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585631067961166e-05,
+      "loss": 0.047,
+      "step": 3201
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585501618122977e-05,
+      "loss": 0.0298,
+      "step": 3202
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.58537216828479e-05,
+      "loss": 0.0275,
+      "step": 3203
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585242718446603e-05,
+      "loss": 0.0448,
+      "step": 3204
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.585113268608414e-05,
+      "loss": 0.0272,
+      "step": 3205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584983818770227e-05,
+      "loss": 0.0283,
+      "step": 3206
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584854368932039e-05,
+      "loss": 0.0313,
+      "step": 3207
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584724919093852e-05,
+      "loss": 0.0272,
+      "step": 3208
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584595469255663e-05,
+      "loss": 0.0686,
+      "step": 3209
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584466019417476e-05,
+      "loss": 0.0308,
+      "step": 3210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584336569579289e-05,
+      "loss": 0.0236,
+      "step": 3211
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584207119741101e-05,
+      "loss": 0.068,
+      "step": 3212
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.584077669902914e-05,
+      "loss": 0.0368,
+      "step": 3213
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583948220064725e-05,
+      "loss": 0.0293,
+      "step": 3214
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583818770226538e-05,
+      "loss": 0.0289,
+      "step": 3215
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.58368932038835e-05,
+      "loss": 0.0328,
+      "step": 3216
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583559870550163e-05,
+      "loss": 0.036,
+      "step": 3217
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583430420711974e-05,
+      "loss": 0.0355,
+      "step": 3218
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583300970873786e-05,
+      "loss": 0.0262,
+      "step": 3219
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583171521035598e-05,
+      "loss": 0.043,
+      "step": 3220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.583042071197411e-05,
+      "loss": 0.0694,
+      "step": 3221
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582912621359224e-05,
+      "loss": 0.0324,
+      "step": 3222
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582783171521036e-05,
+      "loss": 0.0341,
+      "step": 3223
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582653721682848e-05,
+      "loss": 0.0359,
+      "step": 3224
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.58252427184466e-05,
+      "loss": 0.0536,
+      "step": 3225
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582394822006473e-05,
+      "loss": 0.0263,
+      "step": 3226
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582265372168286e-05,
+      "loss": 0.0478,
+      "step": 3227
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.582135922330098e-05,
+      "loss": 0.0308,
+      "step": 3228
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.58200647249191e-05,
+      "loss": 0.0374,
+      "step": 3229
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581877022653722e-05,
+      "loss": 0.0398,
+      "step": 3230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581747572815535e-05,
+      "loss": 0.0255,
+      "step": 3231
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581618122977348e-05,
+      "loss": 0.0348,
+      "step": 3232
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581488673139159e-05,
+      "loss": 0.0326,
+      "step": 3233
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.58135922330097e-05,
+      "loss": 0.0226,
+      "step": 3234
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581229773462783e-05,
+      "loss": 0.046,
+      "step": 3235
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.581100323624596e-05,
+      "loss": 0.064,
+      "step": 3236
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580970873786408e-05,
+      "loss": 0.0291,
+      "step": 3237
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580841423948221e-05,
+      "loss": 0.0307,
+      "step": 3238
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580711974110032e-05,
+      "loss": 0.0241,
+      "step": 3239
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580582524271845e-05,
+      "loss": 0.0373,
+      "step": 3240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580453074433657e-05,
+      "loss": 0.0336,
+      "step": 3241
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.58032362459547e-05,
+      "loss": 0.0563,
+      "step": 3242
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580194174757281e-05,
+      "loss": 0.051,
+      "step": 3243
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.580064724919094e-05,
+      "loss": 0.0148,
+      "step": 3244
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579935275080907e-05,
+      "loss": 0.0178,
+      "step": 3245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.57980582524272e-05,
+      "loss": 0.0335,
+      "step": 3246
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579676375404532e-05,
+      "loss": 0.0576,
+      "step": 3247
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579546925566343e-05,
+      "loss": 0.02,
+      "step": 3248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579417475728156e-05,
+      "loss": 0.0407,
+      "step": 3249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579288025889967e-05,
+      "loss": 0.0493,
+      "step": 3250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.57915857605178e-05,
+      "loss": 0.0368,
+      "step": 3251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.579029126213593e-05,
+      "loss": 0.046,
+      "step": 3252
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578899676375404e-05,
+      "loss": 0.0392,
+      "step": 3253
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578770226537217e-05,
+      "loss": 0.0251,
+      "step": 3254
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578640776699029e-05,
+      "loss": 0.0472,
+      "step": 3255
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578511326860842e-05,
+      "loss": 0.0337,
+      "step": 3256
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578381877022655e-05,
+      "loss": 0.0481,
+      "step": 3257
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578252427184466e-05,
+      "loss": 0.0294,
+      "step": 3258
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578122977346279e-05,
+      "loss": 0.0314,
+      "step": 3259
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577993527508091e-05,
+      "loss": 0.0562,
+      "step": 3260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577864077669904e-05,
+      "loss": 0.0408,
+      "step": 3261
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577734627831717e-05,
+      "loss": 0.0425,
+      "step": 3262
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577605177993528e-05,
+      "loss": 0.031,
+      "step": 3263
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.57747572815534e-05,
+      "loss": 0.025,
+      "step": 3264
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577346278317153e-05,
+      "loss": 0.0529,
+      "step": 3265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577216828478966e-05,
+      "loss": 0.0398,
+      "step": 3266
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.577087378640777e-05,
+      "loss": 0.0503,
+      "step": 3267
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576957928802588e-05,
+      "loss": 0.0535,
+      "step": 3268
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576828478964401e-05,
+      "loss": 0.0286,
+      "step": 3269
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576699029126214e-05,
+      "loss": 0.0401,
+      "step": 3270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576569579288026e-05,
+      "loss": 0.0358,
+      "step": 3271
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576440129449839e-05,
+      "loss": 0.0284,
+      "step": 3272
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.57631067961165e-05,
+      "loss": 0.0259,
+      "step": 3273
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576181229773463e-05,
+      "loss": 0.0307,
+      "step": 3274
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.576051779935276e-05,
+      "loss": 0.0876,
+      "step": 3275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575922330097088e-05,
+      "loss": 0.049,
+      "step": 3276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.5757928802589e-05,
+      "loss": 0.043,
+      "step": 3277
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575663430420712e-05,
+      "loss": 0.0224,
+      "step": 3278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575533980582525e-05,
+      "loss": 0.0323,
+      "step": 3279
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575404530744338e-05,
+      "loss": 0.0288,
+      "step": 3280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.57527508090615e-05,
+      "loss": 0.0252,
+      "step": 3281
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575145631067962e-05,
+      "loss": 0.0196,
+      "step": 3282
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.575016181229773e-05,
+      "loss": 0.0374,
+      "step": 3283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574886731391586e-05,
+      "loss": 0.0267,
+      "step": 3284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574757281553398e-05,
+      "loss": 0.0374,
+      "step": 3285
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574627831715211e-05,
+      "loss": 0.0193,
+      "step": 3286
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574498381877024e-05,
+      "loss": 0.0464,
+      "step": 3287
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574368932038835e-05,
+      "loss": 0.0463,
+      "step": 3288
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.574239482200647e-05,
+      "loss": 0.03,
+      "step": 3289
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.57411003236246e-05,
+      "loss": 0.0271,
+      "step": 3290
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573980582524273e-05,
+      "loss": 0.0317,
+      "step": 3291
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573851132686084e-05,
+      "loss": 0.0403,
+      "step": 3292
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573721682847897e-05,
+      "loss": 0.042,
+      "step": 3293
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.57359223300971e-05,
+      "loss": 0.0279,
+      "step": 3294
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573462783171522e-05,
+      "loss": 0.0232,
+      "step": 3295
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573333333333335e-05,
+      "loss": 0.0241,
+      "step": 3296
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573203883495146e-05,
+      "loss": 0.032,
+      "step": 3297
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.573074433656959e-05,
+      "loss": 0.0306,
+      "step": 3298
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.57294498381877e-05,
+      "loss": 0.0438,
+      "step": 3299
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572815533980583e-05,
+      "loss": 0.0712,
+      "step": 3300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572686084142395e-05,
+      "loss": 0.0393,
+      "step": 3301
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572556634304207e-05,
+      "loss": 0.0359,
+      "step": 3302
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572427184466019e-05,
+      "loss": 0.0407,
+      "step": 3303
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572297734627832e-05,
+      "loss": 0.0285,
+      "step": 3304
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572168284789645e-05,
+      "loss": 0.0483,
+      "step": 3305
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.572038834951457e-05,
+      "loss": 0.0238,
+      "step": 3306
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571909385113269e-05,
+      "loss": 0.0708,
+      "step": 3307
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571779935275081e-05,
+      "loss": 0.0566,
+      "step": 3308
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571650485436894e-05,
+      "loss": 0.0397,
+      "step": 3309
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571521035598707e-05,
+      "loss": 0.04,
+      "step": 3310
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571391585760519e-05,
+      "loss": 0.0378,
+      "step": 3311
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.57126213592233e-05,
+      "loss": 0.0181,
+      "step": 3312
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571132686084143e-05,
+      "loss": 0.056,
+      "step": 3313
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.571003236245956e-05,
+      "loss": 0.0519,
+      "step": 3314
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570873786407768e-05,
+      "loss": 0.038,
+      "step": 3315
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.57074433656958e-05,
+      "loss": 0.0292,
+      "step": 3316
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570614886731391e-05,
+      "loss": 0.0549,
+      "step": 3317
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570485436893204e-05,
+      "loss": 0.0296,
+      "step": 3318
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570355987055016e-05,
+      "loss": 0.0192,
+      "step": 3319
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570226537216829e-05,
+      "loss": 0.0634,
+      "step": 3320
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.570097087378642e-05,
+      "loss": 0.1005,
+      "step": 3321
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569967637540453e-05,
+      "loss": 0.0531,
+      "step": 3322
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569838187702266e-05,
+      "loss": 0.0366,
+      "step": 3323
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569708737864078e-05,
+      "loss": 0.0613,
+      "step": 3324
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569579288025891e-05,
+      "loss": 0.037,
+      "step": 3325
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569449838187702e-05,
+      "loss": 0.0421,
+      "step": 3326
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569320388349515e-05,
+      "loss": 0.0355,
+      "step": 3327
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.569190938511328e-05,
+      "loss": 0.0376,
+      "step": 3328
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.56906148867314e-05,
+      "loss": 0.0586,
+      "step": 3329
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568932038834953e-05,
+      "loss": 0.0182,
+      "step": 3330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568802588996764e-05,
+      "loss": 0.0377,
+      "step": 3331
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568673139158576e-05,
+      "loss": 0.0233,
+      "step": 3332
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568543689320388e-05,
+      "loss": 0.0293,
+      "step": 3333
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568414239482201e-05,
+      "loss": 0.038,
+      "step": 3334
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568284789644014e-05,
+      "loss": 0.0402,
+      "step": 3335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568155339805825e-05,
+      "loss": 0.0322,
+      "step": 3336
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.568025889967637e-05,
+      "loss": 0.0181,
+      "step": 3337
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.56789644012945e-05,
+      "loss": 0.0371,
+      "step": 3338
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567766990291263e-05,
+      "loss": 0.0315,
+      "step": 3339
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567637540453075e-05,
+      "loss": 0.035,
+      "step": 3340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567508090614887e-05,
+      "loss": 0.0392,
+      "step": 3341
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.5673786407767e-05,
+      "loss": 0.0219,
+      "step": 3342
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567249190938512e-05,
+      "loss": 0.0219,
+      "step": 3343
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.567119741100325e-05,
+      "loss": 0.0288,
+      "step": 3344
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566990291262137e-05,
+      "loss": 0.042,
+      "step": 3345
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566860841423949e-05,
+      "loss": 0.0292,
+      "step": 3346
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566731391585761e-05,
+      "loss": 0.0255,
+      "step": 3347
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566601941747573e-05,
+      "loss": 0.0589,
+      "step": 3348
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566472491909385e-05,
+      "loss": 0.0348,
+      "step": 3349
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566343042071198e-05,
+      "loss": 0.0477,
+      "step": 3350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566213592233009e-05,
+      "loss": 0.0479,
+      "step": 3351
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.566084142394822e-05,
+      "loss": 0.0601,
+      "step": 3352
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565954692556635e-05,
+      "loss": 0.0557,
+      "step": 3353
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565825242718447e-05,
+      "loss": 0.0256,
+      "step": 3354
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.56569579288026e-05,
+      "loss": 0.0231,
+      "step": 3355
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565566343042071e-05,
+      "loss": 0.0232,
+      "step": 3356
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565436893203884e-05,
+      "loss": 0.0406,
+      "step": 3357
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565307443365697e-05,
+      "loss": 0.0292,
+      "step": 3358
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.565177993527509e-05,
+      "loss": 0.0315,
+      "step": 3359
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.56504854368932e-05,
+      "loss": 0.0384,
+      "step": 3360
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564919093851133e-05,
+      "loss": 0.0562,
+      "step": 3361
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564789644012946e-05,
+      "loss": 0.027,
+      "step": 3362
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564660194174758e-05,
+      "loss": 0.0345,
+      "step": 3363
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564530744336571e-05,
+      "loss": 0.0331,
+      "step": 3364
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564401294498382e-05,
+      "loss": 0.0376,
+      "step": 3365
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564271844660194e-05,
+      "loss": 0.0346,
+      "step": 3366
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564142394822006e-05,
+      "loss": 0.0915,
+      "step": 3367
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.564012944983819e-05,
+      "loss": 0.0326,
+      "step": 3368
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563883495145632e-05,
+      "loss": 0.0219,
+      "step": 3369
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563754045307443e-05,
+      "loss": 0.0455,
+      "step": 3370
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563624595469256e-05,
+      "loss": 0.0444,
+      "step": 3371
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563495145631068e-05,
+      "loss": 0.0318,
+      "step": 3372
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563365695792881e-05,
+      "loss": 0.0291,
+      "step": 3373
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563236245954694e-05,
+      "loss": 0.0281,
+      "step": 3374
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.563106796116505e-05,
+      "loss": 0.0239,
+      "step": 3375
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562977346278318e-05,
+      "loss": 0.0336,
+      "step": 3376
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.56284789644013e-05,
+      "loss": 0.027,
+      "step": 3377
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562718446601943e-05,
+      "loss": 0.0331,
+      "step": 3378
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562588996763756e-05,
+      "loss": 0.0296,
+      "step": 3379
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562459546925567e-05,
+      "loss": 0.0443,
+      "step": 3380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562330097087378e-05,
+      "loss": 0.0265,
+      "step": 3381
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562200647249191e-05,
+      "loss": 0.037,
+      "step": 3382
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.562071197411004e-05,
+      "loss": 0.0265,
+      "step": 3383
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561941747572816e-05,
+      "loss": 0.0586,
+      "step": 3384
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561812297734627e-05,
+      "loss": 0.0766,
+      "step": 3385
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.56168284789644e-05,
+      "loss": 0.0251,
+      "step": 3386
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561553398058253e-05,
+      "loss": 0.036,
+      "step": 3387
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561423948220065e-05,
+      "loss": 0.0322,
+      "step": 3388
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561294498381878e-05,
+      "loss": 0.0246,
+      "step": 3389
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.56116504854369e-05,
+      "loss": 0.053,
+      "step": 3390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.561035598705502e-05,
+      "loss": 0.0423,
+      "step": 3391
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560906148867315e-05,
+      "loss": 0.0328,
+      "step": 3392
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560776699029127e-05,
+      "loss": 0.0602,
+      "step": 3393
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560647249190939e-05,
+      "loss": 0.034,
+      "step": 3394
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560517799352751e-05,
+      "loss": 0.0358,
+      "step": 3395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560388349514564e-05,
+      "loss": 0.0191,
+      "step": 3396
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560258899676375e-05,
+      "loss": 0.0394,
+      "step": 3397
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.560129449838188e-05,
+      "loss": 0.0377,
+      "step": 3398
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.56e-05,
+      "loss": 0.0335,
+      "step": 3399
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559870550161812e-05,
+      "loss": 0.0329,
+      "step": 3400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559741100323625e-05,
+      "loss": 0.0621,
+      "step": 3401
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559611650485437e-05,
+      "loss": 0.0372,
+      "step": 3402
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.55948220064725e-05,
+      "loss": 0.0303,
+      "step": 3403
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559352750809063e-05,
+      "loss": 0.0165,
+      "step": 3404
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559223300970874e-05,
+      "loss": 0.0271,
+      "step": 3405
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.559093851132687e-05,
+      "loss": 0.0366,
+      "step": 3406
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558964401294499e-05,
+      "loss": 0.0241,
+      "step": 3407
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558834951456312e-05,
+      "loss": 0.0286,
+      "step": 3408
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558705501618123e-05,
+      "loss": 0.0384,
+      "step": 3409
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558576051779936e-05,
+      "loss": 0.0492,
+      "step": 3410
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558446601941748e-05,
+      "loss": 0.0435,
+      "step": 3411
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558317152103561e-05,
+      "loss": 0.0429,
+      "step": 3412
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558187702265372e-05,
+      "loss": 0.0519,
+      "step": 3413
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.558058252427185e-05,
+      "loss": 0.0379,
+      "step": 3414
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557928802588996e-05,
+      "loss": 0.0381,
+      "step": 3415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557799352750809e-05,
+      "loss": 0.0513,
+      "step": 3416
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557669902912622e-05,
+      "loss": 0.034,
+      "step": 3417
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557540453074434e-05,
+      "loss": 0.0146,
+      "step": 3418
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557411003236246e-05,
+      "loss": 0.032,
+      "step": 3419
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557281553398058e-05,
+      "loss": 0.0319,
+      "step": 3420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557152103559871e-05,
+      "loss": 0.0213,
+      "step": 3421
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.557022653721684e-05,
+      "loss": 0.0505,
+      "step": 3422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556893203883496e-05,
+      "loss": 0.0365,
+      "step": 3423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556763754045308e-05,
+      "loss": 0.0237,
+      "step": 3424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.55663430420712e-05,
+      "loss": 0.0314,
+      "step": 3425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556504854368933e-05,
+      "loss": 0.0301,
+      "step": 3426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556375404530746e-05,
+      "loss": 0.0259,
+      "step": 3427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.556245954692558e-05,
+      "loss": 0.0292,
+      "step": 3428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.55611650485437e-05,
+      "loss": 0.0382,
+      "step": 3429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555987055016181e-05,
+      "loss": 0.0216,
+      "step": 3430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555857605177994e-05,
+      "loss": 0.0917,
+      "step": 3431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555728155339806e-05,
+      "loss": 0.0218,
+      "step": 3432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555598705501619e-05,
+      "loss": 0.0201,
+      "step": 3433
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.55546925566343e-05,
+      "loss": 0.0664,
+      "step": 3434
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555339805825243e-05,
+      "loss": 0.0192,
+      "step": 3435
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555210355987055e-05,
+      "loss": 0.0658,
+      "step": 3436
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.555080906148868e-05,
+      "loss": 0.0282,
+      "step": 3437
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554951456310681e-05,
+      "loss": 0.0393,
+      "step": 3438
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554822006472492e-05,
+      "loss": 0.0385,
+      "step": 3439
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554692556634305e-05,
+      "loss": 0.0407,
+      "step": 3440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554563106796117e-05,
+      "loss": 0.0279,
+      "step": 3441
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55443365695793e-05,
+      "loss": 0.0359,
+      "step": 3442
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554304207119741e-05,
+      "loss": 0.0414,
+      "step": 3443
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554174757281554e-05,
+      "loss": 0.0283,
+      "step": 3444
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.554045307443367e-05,
+      "loss": 0.0426,
+      "step": 3445
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553915857605178e-05,
+      "loss": 0.0479,
+      "step": 3446
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55378640776699e-05,
+      "loss": 0.03,
+      "step": 3447
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553656957928803e-05,
+      "loss": 0.0247,
+      "step": 3448
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553527508090615e-05,
+      "loss": 0.0209,
+      "step": 3449
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553398058252427e-05,
+      "loss": 0.0307,
+      "step": 3450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.55326860841424e-05,
+      "loss": 0.0321,
+      "step": 3451
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553139158576053e-05,
+      "loss": 0.0391,
+      "step": 3452
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.553009708737864e-05,
+      "loss": 0.0176,
+      "step": 3453
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552880258899677e-05,
+      "loss": 0.0207,
+      "step": 3454
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552750809061489e-05,
+      "loss": 0.0459,
+      "step": 3455
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552621359223302e-05,
+      "loss": 0.0272,
+      "step": 3456
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552491909385114e-05,
+      "loss": 0.0499,
+      "step": 3457
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552362459546926e-05,
+      "loss": 0.0329,
+      "step": 3458
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552233009708738e-05,
+      "loss": 0.0691,
+      "step": 3459
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.552103559870551e-05,
+      "loss": 0.0342,
+      "step": 3460
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551974110032364e-05,
+      "loss": 0.0197,
+      "step": 3461
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551844660194175e-05,
+      "loss": 0.0453,
+      "step": 3462
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551715210355986e-05,
+      "loss": 0.0329,
+      "step": 3463
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551585760517799e-05,
+      "loss": 0.0312,
+      "step": 3464
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551456310679612e-05,
+      "loss": 0.0293,
+      "step": 3465
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551326860841424e-05,
+      "loss": 0.0189,
+      "step": 3466
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551197411003237e-05,
+      "loss": 0.0182,
+      "step": 3467
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.551067961165048e-05,
+      "loss": 0.0226,
+      "step": 3468
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550938511326861e-05,
+      "loss": 0.0223,
+      "step": 3469
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550809061488674e-05,
+      "loss": 0.0445,
+      "step": 3470
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550679611650486e-05,
+      "loss": 0.0302,
+      "step": 3471
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550550161812299e-05,
+      "loss": 0.0291,
+      "step": 3472
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.55042071197411e-05,
+      "loss": 0.0593,
+      "step": 3473
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550291262135923e-05,
+      "loss": 0.0302,
+      "step": 3474
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550161812297736e-05,
+      "loss": 0.0902,
+      "step": 3475
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.550032362459548e-05,
+      "loss": 0.0457,
+      "step": 3476
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.54990291262136e-05,
+      "loss": 0.0438,
+      "step": 3477
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549773462783172e-05,
+      "loss": 0.058,
+      "step": 3478
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549644012944984e-05,
+      "loss": 0.0314,
+      "step": 3479
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549514563106796e-05,
+      "loss": 0.0391,
+      "step": 3480
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549385113268609e-05,
+      "loss": 0.0341,
+      "step": 3481
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549255663430421e-05,
+      "loss": 0.0447,
+      "step": 3482
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.549126213592233e-05,
+      "loss": 0.0676,
+      "step": 3483
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548996763754045e-05,
+      "loss": 0.0463,
+      "step": 3484
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548867313915858e-05,
+      "loss": 0.0398,
+      "step": 3485
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548737864077671e-05,
+      "loss": 0.0294,
+      "step": 3486
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548608414239482e-05,
+      "loss": 0.0319,
+      "step": 3487
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548478964401295e-05,
+      "loss": 0.0389,
+      "step": 3488
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548349514563107e-05,
+      "loss": 0.0441,
+      "step": 3489
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.54822006472492e-05,
+      "loss": 0.0308,
+      "step": 3490
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.548090614886733e-05,
+      "loss": 0.042,
+      "step": 3491
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547961165048544e-05,
+      "loss": 0.0213,
+      "step": 3492
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547831715210357e-05,
+      "loss": 0.0398,
+      "step": 3493
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547702265372169e-05,
+      "loss": 0.0308,
+      "step": 3494
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.54757281553398e-05,
+      "loss": 0.0299,
+      "step": 3495
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547443365695793e-05,
+      "loss": 0.0512,
+      "step": 3496
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547313915857606e-05,
+      "loss": 0.0424,
+      "step": 3497
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.547184466019417e-05,
+      "loss": 0.0327,
+      "step": 3498
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.54705501618123e-05,
+      "loss": 0.0458,
+      "step": 3499
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546925566343043e-05,
+      "loss": 0.0208,
+      "step": 3500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546796116504855e-05,
+      "loss": 0.041,
+      "step": 3501
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546666666666667e-05,
+      "loss": 0.0209,
+      "step": 3502
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546537216828479e-05,
+      "loss": 0.0472,
+      "step": 3503
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546407766990292e-05,
+      "loss": 0.0308,
+      "step": 3504
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546278317152104e-05,
+      "loss": 0.0377,
+      "step": 3505
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546148867313917e-05,
+      "loss": 0.0516,
+      "step": 3506
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.546019417475728e-05,
+      "loss": 0.0342,
+      "step": 3507
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545889967637541e-05,
+      "loss": 0.0443,
+      "step": 3508
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545760517799354e-05,
+      "loss": 0.0265,
+      "step": 3509
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545631067961166e-05,
+      "loss": 0.0758,
+      "step": 3510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545501618122978e-05,
+      "loss": 0.0395,
+      "step": 3511
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545372168284789e-05,
+      "loss": 0.0196,
+      "step": 3512
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545242718446602e-05,
+      "loss": 0.0516,
+      "step": 3513
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.545113268608414e-05,
+      "loss": 0.0229,
+      "step": 3514
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544983818770227e-05,
+      "loss": 0.0317,
+      "step": 3515
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.54485436893204e-05,
+      "loss": 0.0174,
+      "step": 3516
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544724919093851e-05,
+      "loss": 0.0368,
+      "step": 3517
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544595469255664e-05,
+      "loss": 0.0366,
+      "step": 3518
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544466019417476e-05,
+      "loss": 0.0304,
+      "step": 3519
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544336569579289e-05,
+      "loss": 0.041,
+      "step": 3520
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544207119741102e-05,
+      "loss": 0.0316,
+      "step": 3521
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.544077669902913e-05,
+      "loss": 0.0353,
+      "step": 3522
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543948220064726e-05,
+      "loss": 0.0753,
+      "step": 3523
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543818770226538e-05,
+      "loss": 0.0403,
+      "step": 3524
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543689320388351e-05,
+      "loss": 0.0272,
+      "step": 3525
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543559870550162e-05,
+      "loss": 0.0281,
+      "step": 3526
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543430420711973e-05,
+      "loss": 0.0205,
+      "step": 3527
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543300970873786e-05,
+      "loss": 0.0242,
+      "step": 3528
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543171521035599e-05,
+      "loss": 0.1482,
+      "step": 3529
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.543042071197411e-05,
+      "loss": 0.0331,
+      "step": 3530
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542912621359224e-05,
+      "loss": 0.0346,
+      "step": 3531
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542783171521035e-05,
+      "loss": 0.0149,
+      "step": 3532
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542653721682848e-05,
+      "loss": 0.049,
+      "step": 3533
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542524271844661e-05,
+      "loss": 0.0351,
+      "step": 3534
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542394822006473e-05,
+      "loss": 0.0426,
+      "step": 3535
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542265372168285e-05,
+      "loss": 0.0429,
+      "step": 3536
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.542135922330097e-05,
+      "loss": 0.0405,
+      "step": 3537
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.54200647249191e-05,
+      "loss": 0.0293,
+      "step": 3538
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541877022653723e-05,
+      "loss": 0.0381,
+      "step": 3539
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541747572815535e-05,
+      "loss": 0.0231,
+      "step": 3540
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541618122977347e-05,
+      "loss": 0.043,
+      "step": 3541
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541488673139159e-05,
+      "loss": 0.0417,
+      "step": 3542
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541359223300972e-05,
+      "loss": 0.034,
+      "step": 3543
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541229773462783e-05,
+      "loss": 0.0352,
+      "step": 3544
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.541100323624596e-05,
+      "loss": 0.0407,
+      "step": 3545
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540970873786407e-05,
+      "loss": 0.0336,
+      "step": 3546
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.54084142394822e-05,
+      "loss": 0.0406,
+      "step": 3547
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540711974110033e-05,
+      "loss": 0.0344,
+      "step": 3548
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540582524271845e-05,
+      "loss": 0.0183,
+      "step": 3549
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540453074433658e-05,
+      "loss": 0.0394,
+      "step": 3550
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540323624595469e-05,
+      "loss": 0.0471,
+      "step": 3551
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540194174757282e-05,
+      "loss": 0.0418,
+      "step": 3552
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.540064724919094e-05,
+      "loss": 0.0503,
+      "step": 3553
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539935275080907e-05,
+      "loss": 0.0275,
+      "step": 3554
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.53980582524272e-05,
+      "loss": 0.0371,
+      "step": 3555
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539676375404531e-05,
+      "loss": 0.0281,
+      "step": 3556
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539546925566344e-05,
+      "loss": 0.0678,
+      "step": 3557
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539417475728156e-05,
+      "loss": 0.029,
+      "step": 3558
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539288025889969e-05,
+      "loss": 0.0513,
+      "step": 3559
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.53915857605178e-05,
+      "loss": 0.0349,
+      "step": 3560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.539029126213592e-05,
+      "loss": 0.04,
+      "step": 3561
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538899676375404e-05,
+      "loss": 0.0363,
+      "step": 3562
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538770226537217e-05,
+      "loss": 0.0253,
+      "step": 3563
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.53864077669903e-05,
+      "loss": 0.0376,
+      "step": 3564
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538511326860842e-05,
+      "loss": 0.0377,
+      "step": 3565
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538381877022654e-05,
+      "loss": 0.0513,
+      "step": 3566
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538252427184466e-05,
+      "loss": 0.0513,
+      "step": 3567
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.538122977346279e-05,
+      "loss": 0.0543,
+      "step": 3568
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537993527508092e-05,
+      "loss": 0.0596,
+      "step": 3569
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537864077669903e-05,
+      "loss": 0.0426,
+      "step": 3570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537734627831716e-05,
+      "loss": 0.0655,
+      "step": 3571
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537605177993528e-05,
+      "loss": 0.0327,
+      "step": 3572
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537475728155341e-05,
+      "loss": 0.0406,
+      "step": 3573
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537346278317154e-05,
+      "loss": 0.043,
+      "step": 3574
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537216828478965e-05,
+      "loss": 0.0387,
+      "step": 3575
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.537087378640776e-05,
+      "loss": 0.0417,
+      "step": 3576
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536957928802589e-05,
+      "loss": 0.0506,
+      "step": 3577
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536828478964401e-05,
+      "loss": 0.0214,
+      "step": 3578
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536699029126214e-05,
+      "loss": 0.0376,
+      "step": 3579
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536569579288025e-05,
+      "loss": 0.0416,
+      "step": 3580
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536440129449838e-05,
+      "loss": 0.0358,
+      "step": 3581
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536310679611651e-05,
+      "loss": 0.021,
+      "step": 3582
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536181229773463e-05,
+      "loss": 0.0491,
+      "step": 3583
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.536051779935276e-05,
+      "loss": 0.0564,
+      "step": 3584
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535922330097087e-05,
+      "loss": 0.0323,
+      "step": 3585
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.5357928802589e-05,
+      "loss": 0.0536,
+      "step": 3586
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535663430420713e-05,
+      "loss": 0.064,
+      "step": 3587
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535533980582525e-05,
+      "loss": 0.0242,
+      "step": 3588
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535404530744338e-05,
+      "loss": 0.0295,
+      "step": 3589
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535275080906149e-05,
+      "loss": 0.0522,
+      "step": 3590
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535145631067962e-05,
+      "loss": 0.0582,
+      "step": 3591
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.535016181229775e-05,
+      "loss": 0.0291,
+      "step": 3592
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534886731391586e-05,
+      "loss": 0.0264,
+      "step": 3593
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534757281553399e-05,
+      "loss": 0.0246,
+      "step": 3594
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53462783171521e-05,
+      "loss": 0.0404,
+      "step": 3595
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534498381877023e-05,
+      "loss": 0.0349,
+      "step": 3596
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534368932038835e-05,
+      "loss": 0.0273,
+      "step": 3597
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.534239482200648e-05,
+      "loss": 0.052,
+      "step": 3598
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53411003236246e-05,
+      "loss": 0.0213,
+      "step": 3599
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533980582524272e-05,
+      "loss": 0.0415,
+      "step": 3600
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533851132686084e-05,
+      "loss": 0.0154,
+      "step": 3601
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533721682847897e-05,
+      "loss": 0.0356,
+      "step": 3602
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.53359223300971e-05,
+      "loss": 0.0388,
+      "step": 3603
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533462783171521e-05,
+      "loss": 0.0238,
+      "step": 3604
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533333333333334e-05,
+      "loss": 0.0331,
+      "step": 3605
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533203883495146e-05,
+      "loss": 0.0278,
+      "step": 3606
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.533074433656959e-05,
+      "loss": 0.0238,
+      "step": 3607
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532944983818772e-05,
+      "loss": 0.0249,
+      "step": 3608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532815533980583e-05,
+      "loss": 0.0326,
+      "step": 3609
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532686084142394e-05,
+      "loss": 0.0513,
+      "step": 3610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532556634304207e-05,
+      "loss": 0.0617,
+      "step": 3611
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.53242718446602e-05,
+      "loss": 0.0395,
+      "step": 3612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532297734627832e-05,
+      "loss": 0.0328,
+      "step": 3613
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532168284789644e-05,
+      "loss": 0.043,
+      "step": 3614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.532038834951456e-05,
+      "loss": 0.0499,
+      "step": 3615
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531909385113269e-05,
+      "loss": 0.0332,
+      "step": 3616
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531779935275082e-05,
+      "loss": 0.0346,
+      "step": 3617
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531650485436894e-05,
+      "loss": 0.0304,
+      "step": 3618
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531521035598706e-05,
+      "loss": 0.0369,
+      "step": 3619
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531391585760518e-05,
+      "loss": 0.0647,
+      "step": 3620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531262135922331e-05,
+      "loss": 0.0456,
+      "step": 3621
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531132686084144e-05,
+      "loss": 0.0146,
+      "step": 3622
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.531003236245956e-05,
+      "loss": 0.0469,
+      "step": 3623
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530873786407768e-05,
+      "loss": 0.0246,
+      "step": 3624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530744336569579e-05,
+      "loss": 0.0302,
+      "step": 3625
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530614886731391e-05,
+      "loss": 0.029,
+      "step": 3626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530485436893204e-05,
+      "loss": 0.0717,
+      "step": 3627
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530355987055017e-05,
+      "loss": 0.0371,
+      "step": 3628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530226537216828e-05,
+      "loss": 0.0599,
+      "step": 3629
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.530097087378641e-05,
+      "loss": 0.0541,
+      "step": 3630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529967637540453e-05,
+      "loss": 0.0168,
+      "step": 3631
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529838187702266e-05,
+      "loss": 0.0757,
+      "step": 3632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529708737864079e-05,
+      "loss": 0.0236,
+      "step": 3633
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.52957928802589e-05,
+      "loss": 0.038,
+      "step": 3634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529449838187703e-05,
+      "loss": 0.0244,
+      "step": 3635
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529320388349515e-05,
+      "loss": 0.0312,
+      "step": 3636
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529190938511328e-05,
+      "loss": 0.028,
+      "step": 3637
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.529061488673139e-05,
+      "loss": 0.0569,
+      "step": 3638
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528932038834952e-05,
+      "loss": 0.0202,
+      "step": 3639
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528802588996765e-05,
+      "loss": 0.0244,
+      "step": 3640
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528673139158576e-05,
+      "loss": 0.0233,
+      "step": 3641
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528543689320389e-05,
+      "loss": 0.0402,
+      "step": 3642
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528414239482201e-05,
+      "loss": 0.016,
+      "step": 3643
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528284789644013e-05,
+      "loss": 0.0336,
+      "step": 3644
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528155339805825e-05,
+      "loss": 0.0176,
+      "step": 3645
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.528025889967638e-05,
+      "loss": 0.032,
+      "step": 3646
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.52789644012945e-05,
+      "loss": 0.0575,
+      "step": 3647
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527766990291263e-05,
+      "loss": 0.0164,
+      "step": 3648
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527637540453074e-05,
+      "loss": 0.0214,
+      "step": 3649
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527508090614887e-05,
+      "loss": 0.0313,
+      "step": 3650
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.5273786407767e-05,
+      "loss": 0.0328,
+      "step": 3651
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527249190938512e-05,
+      "loss": 0.0341,
+      "step": 3652
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.527119741100324e-05,
+      "loss": 0.0326,
+      "step": 3653
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526990291262136e-05,
+      "loss": 0.0305,
+      "step": 3654
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526860841423949e-05,
+      "loss": 0.0481,
+      "step": 3655
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526731391585762e-05,
+      "loss": 0.1018,
+      "step": 3656
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526601941747574e-05,
+      "loss": 0.021,
+      "step": 3657
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526472491909386e-05,
+      "loss": 0.0454,
+      "step": 3658
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526343042071197e-05,
+      "loss": 0.0331,
+      "step": 3659
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.52621359223301e-05,
+      "loss": 0.0224,
+      "step": 3660
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.526084142394822e-05,
+      "loss": 0.033,
+      "step": 3661
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525954692556635e-05,
+      "loss": 0.0859,
+      "step": 3662
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525825242718446e-05,
+      "loss": 0.0281,
+      "step": 3663
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525695792880259e-05,
+      "loss": 0.0454,
+      "step": 3664
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525566343042072e-05,
+      "loss": 0.0205,
+      "step": 3665
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525436893203884e-05,
+      "loss": 0.0522,
+      "step": 3666
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525307443365697e-05,
+      "loss": 0.0211,
+      "step": 3667
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525177993527508e-05,
+      "loss": 0.0347,
+      "step": 3668
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.525048543689321e-05,
+      "loss": 0.0232,
+      "step": 3669
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524919093851134e-05,
+      "loss": 0.0307,
+      "step": 3670
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524789644012946e-05,
+      "loss": 0.0441,
+      "step": 3671
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524660194174759e-05,
+      "loss": 0.0488,
+      "step": 3672
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52453074433657e-05,
+      "loss": 0.0282,
+      "step": 3673
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524401294498381e-05,
+      "loss": 0.049,
+      "step": 3674
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524271844660194e-05,
+      "loss": 0.0433,
+      "step": 3675
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.524142394822007e-05,
+      "loss": 0.0323,
+      "step": 3676
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.52401294498382e-05,
+      "loss": 0.0687,
+      "step": 3677
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523883495145631e-05,
+      "loss": 0.0196,
+      "step": 3678
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523754045307443e-05,
+      "loss": 0.0314,
+      "step": 3679
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523624595469256e-05,
+      "loss": 0.0433,
+      "step": 3680
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523495145631069e-05,
+      "loss": 0.0214,
+      "step": 3681
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523365695792881e-05,
+      "loss": 0.0285,
+      "step": 3682
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523236245954693e-05,
+      "loss": 0.0265,
+      "step": 3683
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.523106796116505e-05,
+      "loss": 0.027,
+      "step": 3684
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522977346278318e-05,
+      "loss": 0.0341,
+      "step": 3685
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.52284789644013e-05,
+      "loss": 0.0711,
+      "step": 3686
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522718446601942e-05,
+      "loss": 0.0207,
+      "step": 3687
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522588996763755e-05,
+      "loss": 0.042,
+      "step": 3688
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522459546925567e-05,
+      "loss": 0.0295,
+      "step": 3689
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522330097087379e-05,
+      "loss": 0.0328,
+      "step": 3690
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522200647249191e-05,
+      "loss": 0.0387,
+      "step": 3691
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.522071197411004e-05,
+      "loss": 0.0246,
+      "step": 3692
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521941747572815e-05,
+      "loss": 0.0238,
+      "step": 3693
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521812297734628e-05,
+      "loss": 0.0558,
+      "step": 3694
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.52168284789644e-05,
+      "loss": 0.0197,
+      "step": 3695
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521553398058253e-05,
+      "loss": 0.0247,
+      "step": 3696
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521423948220064e-05,
+      "loss": 0.0229,
+      "step": 3697
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521294498381877e-05,
+      "loss": 0.034,
+      "step": 3698
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.52116504854369e-05,
+      "loss": 0.0245,
+      "step": 3699
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.521035598705502e-05,
+      "loss": 0.035,
+      "step": 3700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520906148867315e-05,
+      "loss": 0.0772,
+      "step": 3701
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520776699029126e-05,
+      "loss": 0.0523,
+      "step": 3702
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520647249190939e-05,
+      "loss": 0.028,
+      "step": 3703
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520517799352752e-05,
+      "loss": 0.0179,
+      "step": 3704
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520388349514564e-05,
+      "loss": 0.0264,
+      "step": 3705
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520258899676377e-05,
+      "loss": 0.0266,
+      "step": 3706
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.520129449838187e-05,
+      "loss": 0.0416,
+      "step": 3707
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.52e-05,
+      "loss": 0.0287,
+      "step": 3708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519870550161812e-05,
+      "loss": 0.0322,
+      "step": 3709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519741100323625e-05,
+      "loss": 0.0304,
+      "step": 3710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519611650485438e-05,
+      "loss": 0.031,
+      "step": 3711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519482200647249e-05,
+      "loss": 0.0276,
+      "step": 3712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519352750809062e-05,
+      "loss": 0.0234,
+      "step": 3713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519223300970874e-05,
+      "loss": 0.0364,
+      "step": 3714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.519093851132687e-05,
+      "loss": 0.0334,
+      "step": 3715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.5189644012945e-05,
+      "loss": 0.0366,
+      "step": 3716
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518834951456311e-05,
+      "loss": 0.019,
+      "step": 3717
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518705501618124e-05,
+      "loss": 0.0346,
+      "step": 3718
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518576051779936e-05,
+      "loss": 0.0273,
+      "step": 3719
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518446601941749e-05,
+      "loss": 0.0222,
+      "step": 3720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.51831715210356e-05,
+      "loss": 0.0325,
+      "step": 3721
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518187702265373e-05,
+      "loss": 0.0265,
+      "step": 3722
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.518058252427184e-05,
+      "loss": 0.0319,
+      "step": 3723
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517928802588997e-05,
+      "loss": 0.0528,
+      "step": 3724
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.51779935275081e-05,
+      "loss": 0.0306,
+      "step": 3725
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517669902912622e-05,
+      "loss": 0.0436,
+      "step": 3726
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517540453074433e-05,
+      "loss": 0.0355,
+      "step": 3727
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517411003236246e-05,
+      "loss": 0.0302,
+      "step": 3728
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517281553398059e-05,
+      "loss": 0.0327,
+      "step": 3729
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517152103559871e-05,
+      "loss": 0.0399,
+      "step": 3730
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.517022653721683e-05,
+      "loss": 0.0475,
+      "step": 3731
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516893203883495e-05,
+      "loss": 0.0259,
+      "step": 3732
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516763754045308e-05,
+      "loss": 0.0323,
+      "step": 3733
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.51663430420712e-05,
+      "loss": 0.0793,
+      "step": 3734
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516504854368933e-05,
+      "loss": 0.021,
+      "step": 3735
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516375404530745e-05,
+      "loss": 0.0343,
+      "step": 3736
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.516245954692557e-05,
+      "loss": 0.0448,
+      "step": 3737
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.51611650485437e-05,
+      "loss": 0.0448,
+      "step": 3738
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515987055016181e-05,
+      "loss": 0.0336,
+      "step": 3739
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515857605177994e-05,
+      "loss": 0.0301,
+      "step": 3740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515728155339807e-05,
+      "loss": 0.0328,
+      "step": 3741
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515598705501618e-05,
+      "loss": 0.0308,
+      "step": 3742
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.51546925566343e-05,
+      "loss": 0.03,
+      "step": 3743
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515339805825243e-05,
+      "loss": 0.0392,
+      "step": 3744
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515210355987056e-05,
+      "loss": 0.013,
+      "step": 3745
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.515080906148867e-05,
+      "loss": 0.0378,
+      "step": 3746
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51495145631068e-05,
+      "loss": 0.0489,
+      "step": 3747
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514822006472492e-05,
+      "loss": 0.0537,
+      "step": 3748
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514692556634305e-05,
+      "loss": 0.0373,
+      "step": 3749
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514563106796118e-05,
+      "loss": 0.0458,
+      "step": 3750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514433656957929e-05,
+      "loss": 0.0461,
+      "step": 3751
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514304207119742e-05,
+      "loss": 0.056,
+      "step": 3752
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514174757281554e-05,
+      "loss": 0.0379,
+      "step": 3753
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.514045307443367e-05,
+      "loss": 0.0242,
+      "step": 3754
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513915857605178e-05,
+      "loss": 0.0302,
+      "step": 3755
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51378640776699e-05,
+      "loss": 0.0319,
+      "step": 3756
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513656957928802e-05,
+      "loss": 0.0309,
+      "step": 3757
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513527508090615e-05,
+      "loss": 0.0287,
+      "step": 3758
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513398058252428e-05,
+      "loss": 0.0325,
+      "step": 3759
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.51326860841424e-05,
+      "loss": 0.0358,
+      "step": 3760
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513139158576052e-05,
+      "loss": 0.032,
+      "step": 3761
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.513009708737864e-05,
+      "loss": 0.0385,
+      "step": 3762
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512880258899677e-05,
+      "loss": 0.0422,
+      "step": 3763
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.51275080906149e-05,
+      "loss": 0.0269,
+      "step": 3764
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512621359223302e-05,
+      "loss": 0.0699,
+      "step": 3765
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512491909385114e-05,
+      "loss": 0.0373,
+      "step": 3766
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512362459546926e-05,
+      "loss": 0.0388,
+      "step": 3767
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512233009708739e-05,
+      "loss": 0.0444,
+      "step": 3768
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.512103559870552e-05,
+      "loss": 0.0336,
+      "step": 3769
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511974110032363e-05,
+      "loss": 0.038,
+      "step": 3770
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511844660194175e-05,
+      "loss": 0.061,
+      "step": 3771
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511715210355987e-05,
+      "loss": 0.0145,
+      "step": 3772
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.5115857605178e-05,
+      "loss": 0.0304,
+      "step": 3773
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511456310679612e-05,
+      "loss": 0.0226,
+      "step": 3774
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511326860841425e-05,
+      "loss": 0.0154,
+      "step": 3775
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511197411003236e-05,
+      "loss": 0.0138,
+      "step": 3776
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.511067961165049e-05,
+      "loss": 0.0281,
+      "step": 3777
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510938511326861e-05,
+      "loss": 0.0217,
+      "step": 3778
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510809061488674e-05,
+      "loss": 0.0442,
+      "step": 3779
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510679611650485e-05,
+      "loss": 0.0303,
+      "step": 3780
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510550161812298e-05,
+      "loss": 0.0176,
+      "step": 3781
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.51042071197411e-05,
+      "loss": 0.0818,
+      "step": 3782
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510291262135923e-05,
+      "loss": 0.0387,
+      "step": 3783
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510161812297736e-05,
+      "loss": 0.0226,
+      "step": 3784
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.510032362459547e-05,
+      "loss": 0.031,
+      "step": 3785
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.50990291262136e-05,
+      "loss": 0.032,
+      "step": 3786
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509773462783173e-05,
+      "loss": 0.0274,
+      "step": 3787
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509644012944984e-05,
+      "loss": 0.0455,
+      "step": 3788
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509514563106797e-05,
+      "loss": 0.021,
+      "step": 3789
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509385113268608e-05,
+      "loss": 0.0539,
+      "step": 3790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.50925566343042e-05,
+      "loss": 0.0202,
+      "step": 3791
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.509126213592233e-05,
+      "loss": 0.0432,
+      "step": 3792
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508996763754046e-05,
+      "loss": 0.0336,
+      "step": 3793
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508867313915858e-05,
+      "loss": 0.034,
+      "step": 3794
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.50873786407767e-05,
+      "loss": 0.0391,
+      "step": 3795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508608414239482e-05,
+      "loss": 0.0176,
+      "step": 3796
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508478964401295e-05,
+      "loss": 0.0246,
+      "step": 3797
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508349514563108e-05,
+      "loss": 0.0148,
+      "step": 3798
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.50822006472492e-05,
+      "loss": 0.0283,
+      "step": 3799
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.508090614886732e-05,
+      "loss": 0.0211,
+      "step": 3800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507961165048544e-05,
+      "loss": 0.0226,
+      "step": 3801
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507831715210357e-05,
+      "loss": 0.0556,
+      "step": 3802
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.50770226537217e-05,
+      "loss": 0.0425,
+      "step": 3803
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507572815533981e-05,
+      "loss": 0.0506,
+      "step": 3804
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507443365695792e-05,
+      "loss": 0.0755,
+      "step": 3805
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507313915857605e-05,
+      "loss": 0.0209,
+      "step": 3806
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.507184466019418e-05,
+      "loss": 0.0207,
+      "step": 3807
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.50705501618123e-05,
+      "loss": 0.0403,
+      "step": 3808
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506925566343043e-05,
+      "loss": 0.0333,
+      "step": 3809
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506796116504854e-05,
+      "loss": 0.052,
+      "step": 3810
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506666666666667e-05,
+      "loss": 0.0387,
+      "step": 3811
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.50653721682848e-05,
+      "loss": 0.0361,
+      "step": 3812
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506407766990292e-05,
+      "loss": 0.0217,
+      "step": 3813
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506278317152104e-05,
+      "loss": 0.0467,
+      "step": 3814
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506148867313916e-05,
+      "loss": 0.0287,
+      "step": 3815
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.506019417475729e-05,
+      "loss": 0.0189,
+      "step": 3816
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505889967637541e-05,
+      "loss": 0.0292,
+      "step": 3817
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505760517799354e-05,
+      "loss": 0.0399,
+      "step": 3818
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505631067961165e-05,
+      "loss": 0.028,
+      "step": 3819
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505501618122978e-05,
+      "loss": 0.0375,
+      "step": 3820
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.50537216828479e-05,
+      "loss": 0.0245,
+      "step": 3821
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505242718446602e-05,
+      "loss": 0.049,
+      "step": 3822
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.505113268608415e-05,
+      "loss": 0.0516,
+      "step": 3823
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504983818770226e-05,
+      "loss": 0.0551,
+      "step": 3824
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504854368932039e-05,
+      "loss": 0.0281,
+      "step": 3825
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504724919093851e-05,
+      "loss": 0.0265,
+      "step": 3826
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504595469255664e-05,
+      "loss": 0.0513,
+      "step": 3827
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504466019417477e-05,
+      "loss": 0.0339,
+      "step": 3828
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504336569579288e-05,
+      "loss": 0.0353,
+      "step": 3829
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.5042071197411e-05,
+      "loss": 0.0598,
+      "step": 3830
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.504077669902913e-05,
+      "loss": 0.1021,
+      "step": 3831
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503948220064726e-05,
+      "loss": 0.0442,
+      "step": 3832
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503818770226539e-05,
+      "loss": 0.0401,
+      "step": 3833
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.50368932038835e-05,
+      "loss": 0.0301,
+      "step": 3834
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503559870550163e-05,
+      "loss": 0.0388,
+      "step": 3835
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503430420711975e-05,
+      "loss": 0.0881,
+      "step": 3836
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503300970873787e-05,
+      "loss": 0.0276,
+      "step": 3837
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.503171521035599e-05,
+      "loss": 0.0255,
+      "step": 3838
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.50304207119741e-05,
+      "loss": 0.0181,
+      "step": 3839
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502912621359223e-05,
+      "loss": 0.0494,
+      "step": 3840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502783171521036e-05,
+      "loss": 0.0757,
+      "step": 3841
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502653721682848e-05,
+      "loss": 0.053,
+      "step": 3842
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502524271844661e-05,
+      "loss": 0.0352,
+      "step": 3843
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502394822006472e-05,
+      "loss": 0.0347,
+      "step": 3844
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502265372168285e-05,
+      "loss": 0.0618,
+      "step": 3845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.502135922330098e-05,
+      "loss": 0.0378,
+      "step": 3846
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.50200647249191e-05,
+      "loss": 0.0385,
+      "step": 3847
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501877022653722e-05,
+      "loss": 0.0375,
+      "step": 3848
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501747572815534e-05,
+      "loss": 0.0553,
+      "step": 3849
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501618122977347e-05,
+      "loss": 0.0597,
+      "step": 3850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.50148867313916e-05,
+      "loss": 0.0361,
+      "step": 3851
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501359223300972e-05,
+      "loss": 0.074,
+      "step": 3852
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501229773462784e-05,
+      "loss": 0.0793,
+      "step": 3853
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.501100323624595e-05,
+      "loss": 0.0894,
+      "step": 3854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500970873786408e-05,
+      "loss": 0.0348,
+      "step": 3855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.50084142394822e-05,
+      "loss": 0.0322,
+      "step": 3856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500711974110033e-05,
+      "loss": 0.0509,
+      "step": 3857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500582524271846e-05,
+      "loss": 0.0432,
+      "step": 3858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500453074433657e-05,
+      "loss": 0.0438,
+      "step": 3859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.50032362459547e-05,
+      "loss": 0.031,
+      "step": 3860
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500194174757282e-05,
+      "loss": 0.0417,
+      "step": 3861
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.500064724919095e-05,
+      "loss": 0.0242,
+      "step": 3862
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499935275080906e-05,
+      "loss": 0.0373,
+      "step": 3863
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499805825242719e-05,
+      "loss": 0.0246,
+      "step": 3864
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499676375404531e-05,
+      "loss": 0.0593,
+      "step": 3865
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499546925566344e-05,
+      "loss": 0.0305,
+      "step": 3866
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499417475728157e-05,
+      "loss": 0.0295,
+      "step": 3867
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499288025889968e-05,
+      "loss": 0.0248,
+      "step": 3868
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499158576051781e-05,
+      "loss": 0.0364,
+      "step": 3869
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.499029126213592e-05,
+      "loss": 0.0471,
+      "step": 3870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498899676375405e-05,
+      "loss": 0.037,
+      "step": 3871
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498770226537217e-05,
+      "loss": 0.0256,
+      "step": 3872
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498640776699029e-05,
+      "loss": 0.0254,
+      "step": 3873
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498511326860841e-05,
+      "loss": 0.0421,
+      "step": 3874
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498381877022654e-05,
+      "loss": 0.0204,
+      "step": 3875
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.498252427184467e-05,
+      "loss": 0.0504,
+      "step": 3876
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.49812297734628e-05,
+      "loss": 0.0271,
+      "step": 3877
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.49799352750809e-05,
+      "loss": 0.0385,
+      "step": 3878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497864077669903e-05,
+      "loss": 0.0249,
+      "step": 3879
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497734627831716e-05,
+      "loss": 0.0553,
+      "step": 3880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497605177993529e-05,
+      "loss": 0.0377,
+      "step": 3881
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497475728155341e-05,
+      "loss": 0.0547,
+      "step": 3882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497346278317153e-05,
+      "loss": 0.0294,
+      "step": 3883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497216828478965e-05,
+      "loss": 0.0342,
+      "step": 3884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.497087378640778e-05,
+      "loss": 0.0518,
+      "step": 3885
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496957928802589e-05,
+      "loss": 0.039,
+      "step": 3886
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496828478964402e-05,
+      "loss": 0.0268,
+      "step": 3887
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496699029126213e-05,
+      "loss": 0.0367,
+      "step": 3888
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496569579288026e-05,
+      "loss": 0.0238,
+      "step": 3889
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496440129449838e-05,
+      "loss": 0.028,
+      "step": 3890
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496310679611651e-05,
+      "loss": 0.0586,
+      "step": 3891
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496181229773464e-05,
+      "loss": 0.0134,
+      "step": 3892
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.496051779935275e-05,
+      "loss": 0.0332,
+      "step": 3893
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495922330097088e-05,
+      "loss": 0.0519,
+      "step": 3894
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.4957928802589e-05,
+      "loss": 0.02,
+      "step": 3895
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495663430420713e-05,
+      "loss": 0.0265,
+      "step": 3896
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495533980582524e-05,
+      "loss": 0.0433,
+      "step": 3897
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495404530744337e-05,
+      "loss": 0.0267,
+      "step": 3898
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.49527508090615e-05,
+      "loss": 0.0256,
+      "step": 3899
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495145631067962e-05,
+      "loss": 0.0196,
+      "step": 3900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.495016181229775e-05,
+      "loss": 0.0512,
+      "step": 3901
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494886731391586e-05,
+      "loss": 0.0342,
+      "step": 3902
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494757281553398e-05,
+      "loss": 0.0167,
+      "step": 3903
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.49462783171521e-05,
+      "loss": 0.0387,
+      "step": 3904
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494498381877023e-05,
+      "loss": 0.0611,
+      "step": 3905
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494368932038836e-05,
+      "loss": 0.0222,
+      "step": 3906
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.494239482200647e-05,
+      "loss": 0.0255,
+      "step": 3907
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.49411003236246e-05,
+      "loss": 0.024,
+      "step": 3908
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493980582524272e-05,
+      "loss": 0.0713,
+      "step": 3909
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493851132686085e-05,
+      "loss": 0.0368,
+      "step": 3910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493721682847898e-05,
+      "loss": 0.0366,
+      "step": 3911
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493592233009709e-05,
+      "loss": 0.034,
+      "step": 3912
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493462783171521e-05,
+      "loss": 0.0331,
+      "step": 3913
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493333333333334e-05,
+      "loss": 0.0449,
+      "step": 3914
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.493203883495147e-05,
+      "loss": 0.0263,
+      "step": 3915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.49307443365696e-05,
+      "loss": 0.03,
+      "step": 3916
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492944983818771e-05,
+      "loss": 0.014,
+      "step": 3917
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492815533980582e-05,
+      "loss": 0.0194,
+      "step": 3918
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492686084142395e-05,
+      "loss": 0.0288,
+      "step": 3919
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492556634304207e-05,
+      "loss": 0.0445,
+      "step": 3920
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.49242718446602e-05,
+      "loss": 0.0577,
+      "step": 3921
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492297734627831e-05,
+      "loss": 0.0317,
+      "step": 3922
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492168284789644e-05,
+      "loss": 0.0159,
+      "step": 3923
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.492038834951457e-05,
+      "loss": 0.034,
+      "step": 3924
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.49190938511327e-05,
+      "loss": 0.026,
+      "step": 3925
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491779935275082e-05,
+      "loss": 0.0178,
+      "step": 3926
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491650485436893e-05,
+      "loss": 0.0232,
+      "step": 3927
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491521035598706e-05,
+      "loss": 0.0242,
+      "step": 3928
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491391585760519e-05,
+      "loss": 0.0294,
+      "step": 3929
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491262135922331e-05,
+      "loss": 0.0353,
+      "step": 3930
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491132686084143e-05,
+      "loss": 0.0274,
+      "step": 3931
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.491003236245955e-05,
+      "loss": 0.0217,
+      "step": 3932
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490873786407768e-05,
+      "loss": 0.0316,
+      "step": 3933
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.49074433656958e-05,
+      "loss": 0.0291,
+      "step": 3934
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490614886731392e-05,
+      "loss": 0.0241,
+      "step": 3935
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490485436893205e-05,
+      "loss": 0.0551,
+      "step": 3936
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490355987055016e-05,
+      "loss": 0.0228,
+      "step": 3937
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490226537216828e-05,
+      "loss": 0.0516,
+      "step": 3938
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.490097087378641e-05,
+      "loss": 0.0417,
+      "step": 3939
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489967637540454e-05,
+      "loss": 0.0252,
+      "step": 3940
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489838187702265e-05,
+      "loss": 0.0346,
+      "step": 3941
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489708737864078e-05,
+      "loss": 0.0342,
+      "step": 3942
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.48957928802589e-05,
+      "loss": 0.0616,
+      "step": 3943
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489449838187703e-05,
+      "loss": 0.0286,
+      "step": 3944
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489320388349516e-05,
+      "loss": 0.0652,
+      "step": 3945
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.489190938511327e-05,
+      "loss": 0.0233,
+      "step": 3946
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.48906148867314e-05,
+      "loss": 0.0191,
+      "step": 3947
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488932038834952e-05,
+      "loss": 0.0374,
+      "step": 3948
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488802588996765e-05,
+      "loss": 0.0538,
+      "step": 3949
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488673139158578e-05,
+      "loss": 0.0165,
+      "step": 3950
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488543689320389e-05,
+      "loss": 0.0316,
+      "step": 3951
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.4884142394822e-05,
+      "loss": 0.0516,
+      "step": 3952
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488284789644013e-05,
+      "loss": 0.0506,
+      "step": 3953
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488155339805826e-05,
+      "loss": 0.0488,
+      "step": 3954
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.488025889967638e-05,
+      "loss": 0.0429,
+      "step": 3955
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.48789644012945e-05,
+      "loss": 0.026,
+      "step": 3956
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487766990291262e-05,
+      "loss": 0.049,
+      "step": 3957
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487637540453075e-05,
+      "loss": 0.0326,
+      "step": 3958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487508090614888e-05,
+      "loss": 0.0353,
+      "step": 3959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.4873786407767e-05,
+      "loss": 0.1407,
+      "step": 3960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487249190938511e-05,
+      "loss": 0.0413,
+      "step": 3961
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.487119741100324e-05,
+      "loss": 0.0278,
+      "step": 3962
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486990291262137e-05,
+      "loss": 0.0324,
+      "step": 3963
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.48686084142395e-05,
+      "loss": 0.0521,
+      "step": 3964
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486731391585761e-05,
+      "loss": 0.0434,
+      "step": 3965
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486601941747573e-05,
+      "loss": 0.0539,
+      "step": 3966
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486472491909385e-05,
+      "loss": 0.0407,
+      "step": 3967
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486343042071197e-05,
+      "loss": 0.0442,
+      "step": 3968
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.48621359223301e-05,
+      "loss": 0.0334,
+      "step": 3969
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.486084142394823e-05,
+      "loss": 0.0287,
+      "step": 3970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485954692556634e-05,
+      "loss": 0.0487,
+      "step": 3971
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485825242718447e-05,
+      "loss": 0.0239,
+      "step": 3972
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485695792880259e-05,
+      "loss": 0.039,
+      "step": 3973
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485566343042072e-05,
+      "loss": 0.0279,
+      "step": 3974
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485436893203883e-05,
+      "loss": 0.0246,
+      "step": 3975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485307443365696e-05,
+      "loss": 0.0758,
+      "step": 3976
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485177993527509e-05,
+      "loss": 0.0365,
+      "step": 3977
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.485048543689321e-05,
+      "loss": 0.0372,
+      "step": 3978
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484919093851134e-05,
+      "loss": 0.048,
+      "step": 3979
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484789644012945e-05,
+      "loss": 0.0219,
+      "step": 3980
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484660194174758e-05,
+      "loss": 0.0465,
+      "step": 3981
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.48453074433657e-05,
+      "loss": 0.0227,
+      "step": 3982
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484401294498383e-05,
+      "loss": 0.0546,
+      "step": 3983
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484271844660194e-05,
+      "loss": 0.0469,
+      "step": 3984
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484142394822007e-05,
+      "loss": 0.0555,
+      "step": 3985
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.484012944983818e-05,
+      "loss": 0.0255,
+      "step": 3986
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483883495145631e-05,
+      "loss": 0.0356,
+      "step": 3987
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483754045307444e-05,
+      "loss": 0.0249,
+      "step": 3988
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483624595469256e-05,
+      "loss": 0.021,
+      "step": 3989
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483495145631068e-05,
+      "loss": 0.0257,
+      "step": 3990
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.48336569579288e-05,
+      "loss": 0.0289,
+      "step": 3991
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483236245954693e-05,
+      "loss": 0.0308,
+      "step": 3992
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.483106796116506e-05,
+      "loss": 0.0307,
+      "step": 3993
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482977346278318e-05,
+      "loss": 0.0239,
+      "step": 3994
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.48284789644013e-05,
+      "loss": 0.0799,
+      "step": 3995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482718446601942e-05,
+      "loss": 0.0239,
+      "step": 3996
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482588996763755e-05,
+      "loss": 0.0547,
+      "step": 3997
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482459546925568e-05,
+      "loss": 0.0196,
+      "step": 3998
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482330097087379e-05,
+      "loss": 0.0283,
+      "step": 3999
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.48220064724919e-05,
+      "loss": 0.0384,
+      "step": 4000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.482071197411003e-05,
+      "loss": 0.024,
+      "step": 4001
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481941747572816e-05,
+      "loss": 0.0282,
+      "step": 4002
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481812297734628e-05,
+      "loss": 0.0462,
+      "step": 4003
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481682847896441e-05,
+      "loss": 0.028,
+      "step": 4004
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481553398058252e-05,
+      "loss": 0.0367,
+      "step": 4005
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481423948220065e-05,
+      "loss": 0.0445,
+      "step": 4006
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481294498381878e-05,
+      "loss": 0.0153,
+      "step": 4007
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.48116504854369e-05,
+      "loss": 0.0381,
+      "step": 4008
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.481035598705503e-05,
+      "loss": 0.0478,
+      "step": 4009
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480906148867314e-05,
+      "loss": 0.0543,
+      "step": 4010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480776699029127e-05,
+      "loss": 0.0387,
+      "step": 4011
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.48064724919094e-05,
+      "loss": 0.056,
+      "step": 4012
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480517799352752e-05,
+      "loss": 0.0164,
+      "step": 4013
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480388349514563e-05,
+      "loss": 0.0304,
+      "step": 4014
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480258899676376e-05,
+      "loss": 0.047,
+      "step": 4015
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.480129449838187e-05,
+      "loss": 0.0285,
+      "step": 4016
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.48e-05,
+      "loss": 0.0381,
+      "step": 4017
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479870550161813e-05,
+      "loss": 0.0268,
+      "step": 4018
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479741100323625e-05,
+      "loss": 0.0324,
+      "step": 4019
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479611650485437e-05,
+      "loss": 0.0415,
+      "step": 4020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479482200647249e-05,
+      "loss": 0.0334,
+      "step": 4021
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479352750809062e-05,
+      "loss": 0.0398,
+      "step": 4022
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479223300970875e-05,
+      "loss": 0.0348,
+      "step": 4023
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.479093851132686e-05,
+      "loss": 0.0375,
+      "step": 4024
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478964401294499e-05,
+      "loss": 0.0267,
+      "step": 4025
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478834951456311e-05,
+      "loss": 0.0375,
+      "step": 4026
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478705501618124e-05,
+      "loss": 0.0338,
+      "step": 4027
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478576051779937e-05,
+      "loss": 0.0295,
+      "step": 4028
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478446601941748e-05,
+      "loss": 0.0222,
+      "step": 4029
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.47831715210356e-05,
+      "loss": 0.0236,
+      "step": 4030
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478187702265373e-05,
+      "loss": 0.0227,
+      "step": 4031
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.478058252427184e-05,
+      "loss": 0.0333,
+      "step": 4032
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477928802588997e-05,
+      "loss": 0.0265,
+      "step": 4033
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477799352750808e-05,
+      "loss": 0.049,
+      "step": 4034
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477669902912621e-05,
+      "loss": 0.0725,
+      "step": 4035
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477540453074434e-05,
+      "loss": 0.0322,
+      "step": 4036
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477411003236246e-05,
+      "loss": 0.024,
+      "step": 4037
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477281553398059e-05,
+      "loss": 0.0349,
+      "step": 4038
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.47715210355987e-05,
+      "loss": 0.036,
+      "step": 4039
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.477022653721683e-05,
+      "loss": 0.0469,
+      "step": 4040
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476893203883496e-05,
+      "loss": 0.0204,
+      "step": 4041
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476763754045308e-05,
+      "loss": 0.0277,
+      "step": 4042
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476634304207121e-05,
+      "loss": 0.0337,
+      "step": 4043
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476504854368932e-05,
+      "loss": 0.0312,
+      "step": 4044
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476375404530745e-05,
+      "loss": 0.0317,
+      "step": 4045
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.476245954692558e-05,
+      "loss": 0.0269,
+      "step": 4046
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.47611650485437e-05,
+      "loss": 0.0388,
+      "step": 4047
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475987055016182e-05,
+      "loss": 0.0448,
+      "step": 4048
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475857605177993e-05,
+      "loss": 0.0497,
+      "step": 4049
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475728155339806e-05,
+      "loss": 0.0397,
+      "step": 4050
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475598705501618e-05,
+      "loss": 0.0326,
+      "step": 4051
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475469255663431e-05,
+      "loss": 0.043,
+      "step": 4052
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475339805825244e-05,
+      "loss": 0.0292,
+      "step": 4053
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475210355987055e-05,
+      "loss": 0.1191,
+      "step": 4054
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.475080906148868e-05,
+      "loss": 0.0584,
+      "step": 4055
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.47495145631068e-05,
+      "loss": 0.0367,
+      "step": 4056
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474822006472493e-05,
+      "loss": 0.0452,
+      "step": 4057
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474692556634304e-05,
+      "loss": 0.0271,
+      "step": 4058
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474563106796117e-05,
+      "loss": 0.0334,
+      "step": 4059
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.47443365695793e-05,
+      "loss": 0.0303,
+      "step": 4060
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474304207119742e-05,
+      "loss": 0.0362,
+      "step": 4061
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474174757281555e-05,
+      "loss": 0.023,
+      "step": 4062
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.474045307443366e-05,
+      "loss": 0.0564,
+      "step": 4063
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473915857605179e-05,
+      "loss": 0.0402,
+      "step": 4064
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.47378640776699e-05,
+      "loss": 0.029,
+      "step": 4065
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473656957928803e-05,
+      "loss": 0.0294,
+      "step": 4066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473527508090615e-05,
+      "loss": 0.054,
+      "step": 4067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473398058252427e-05,
+      "loss": 0.0538,
+      "step": 4068
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473268608414239e-05,
+      "loss": 0.0175,
+      "step": 4069
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473139158576052e-05,
+      "loss": 0.0541,
+      "step": 4070
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473009708737865e-05,
+      "loss": 0.0354,
+      "step": 4071
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472880258899677e-05,
+      "loss": 0.0726,
+      "step": 4072
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472750809061489e-05,
+      "loss": 0.0351,
+      "step": 4073
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472621359223301e-05,
+      "loss": 0.0506,
+      "step": 4074
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472491909385114e-05,
+      "loss": 0.0434,
+      "step": 4075
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472362459546927e-05,
+      "loss": 0.0384,
+      "step": 4076
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.472233009708739e-05,
+      "loss": 0.0301,
+      "step": 4077
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.47210355987055e-05,
+      "loss": 0.0334,
+      "step": 4078
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471974110032363e-05,
+      "loss": 0.0402,
+      "step": 4079
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471844660194176e-05,
+      "loss": 0.0155,
+      "step": 4080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471715210355987e-05,
+      "loss": 0.0445,
+      "step": 4081
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.4715857605178e-05,
+      "loss": 0.0333,
+      "step": 4082
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471456310679611e-05,
+      "loss": 0.0452,
+      "step": 4083
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471326860841424e-05,
+      "loss": 0.0496,
+      "step": 4084
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471197411003236e-05,
+      "loss": 0.0409,
+      "step": 4085
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.471067961165049e-05,
+      "loss": 0.0233,
+      "step": 4086
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470938511326862e-05,
+      "loss": 0.0366,
+      "step": 4087
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470809061488673e-05,
+      "loss": 0.0348,
+      "step": 4088
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470679611650486e-05,
+      "loss": 0.0415,
+      "step": 4089
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470550161812298e-05,
+      "loss": 0.0387,
+      "step": 4090
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470420711974111e-05,
+      "loss": 0.02,
+      "step": 4091
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470291262135922e-05,
+      "loss": 0.0223,
+      "step": 4092
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470161812297735e-05,
+      "loss": 0.037,
+      "step": 4093
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.470032362459548e-05,
+      "loss": 0.0423,
+      "step": 4094
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.46990291262136e-05,
+      "loss": 0.0341,
+      "step": 4095
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469773462783173e-05,
+      "loss": 0.0283,
+      "step": 4096
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469644012944984e-05,
+      "loss": 0.043,
+      "step": 4097
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469514563106796e-05,
+      "loss": 0.0542,
+      "step": 4098
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469385113268608e-05,
+      "loss": 0.0403,
+      "step": 4099
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469255663430421e-05,
+      "loss": 0.0484,
+      "step": 4100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.469126213592234e-05,
+      "loss": 0.0408,
+      "step": 4101
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468996763754046e-05,
+      "loss": 0.0752,
+      "step": 4102
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468867313915858e-05,
+      "loss": 0.0154,
+      "step": 4103
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46873786407767e-05,
+      "loss": 0.0369,
+      "step": 4104
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468608414239483e-05,
+      "loss": 0.0268,
+      "step": 4105
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468478964401295e-05,
+      "loss": 0.0403,
+      "step": 4106
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468349514563107e-05,
+      "loss": 0.0613,
+      "step": 4107
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.46822006472492e-05,
+      "loss": 0.0365,
+      "step": 4108
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468090614886732e-05,
+      "loss": 0.0309,
+      "step": 4109
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467961165048545e-05,
+      "loss": 0.0251,
+      "step": 4110
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467831715210357e-05,
+      "loss": 0.056,
+      "step": 4111
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467702265372169e-05,
+      "loss": 0.0383,
+      "step": 4112
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467572815533981e-05,
+      "loss": 0.087,
+      "step": 4113
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467443365695793e-05,
+      "loss": 0.0386,
+      "step": 4114
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467313915857605e-05,
+      "loss": 0.0292,
+      "step": 4115
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467184466019418e-05,
+      "loss": 0.0278,
+      "step": 4116
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.467055016181229e-05,
+      "loss": 0.0379,
+      "step": 4117
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466925566343042e-05,
+      "loss": 0.0226,
+      "step": 4118
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466796116504855e-05,
+      "loss": 0.0473,
+      "step": 4119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466666666666667e-05,
+      "loss": 0.039,
+      "step": 4120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.46653721682848e-05,
+      "loss": 0.0361,
+      "step": 4121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466407766990291e-05,
+      "loss": 0.0402,
+      "step": 4122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466278317152104e-05,
+      "loss": 0.0258,
+      "step": 4123
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466148867313917e-05,
+      "loss": 0.0285,
+      "step": 4124
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.466019417475729e-05,
+      "loss": 0.0404,
+      "step": 4125
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465889967637542e-05,
+      "loss": 0.0473,
+      "step": 4126
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465760517799353e-05,
+      "loss": 0.0462,
+      "step": 4127
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465631067961166e-05,
+      "loss": 0.0368,
+      "step": 4128
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465501618122979e-05,
+      "loss": 0.0332,
+      "step": 4129
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.46537216828479e-05,
+      "loss": 0.0345,
+      "step": 4130
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465242718446602e-05,
+      "loss": 0.0382,
+      "step": 4131
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.465113268608414e-05,
+      "loss": 0.0617,
+      "step": 4132
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464983818770226e-05,
+      "loss": 0.0469,
+      "step": 4133
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464854368932039e-05,
+      "loss": 0.0304,
+      "step": 4134
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464724919093852e-05,
+      "loss": 0.0553,
+      "step": 4135
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464595469255664e-05,
+      "loss": 0.0339,
+      "step": 4136
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464466019417476e-05,
+      "loss": 0.0264,
+      "step": 4137
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464336569579288e-05,
+      "loss": 0.0612,
+      "step": 4138
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464207119741101e-05,
+      "loss": 0.0322,
+      "step": 4139
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.464077669902914e-05,
+      "loss": 0.0659,
+      "step": 4140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463948220064725e-05,
+      "loss": 0.0386,
+      "step": 4141
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463818770226538e-05,
+      "loss": 0.0438,
+      "step": 4142
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.46368932038835e-05,
+      "loss": 0.0323,
+      "step": 4143
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463559870550163e-05,
+      "loss": 0.0253,
+      "step": 4144
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463430420711976e-05,
+      "loss": 0.0371,
+      "step": 4145
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463300970873787e-05,
+      "loss": 0.0259,
+      "step": 4146
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463171521035598e-05,
+      "loss": 0.0216,
+      "step": 4147
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.463042071197411e-05,
+      "loss": 0.0291,
+      "step": 4148
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462912621359224e-05,
+      "loss": 0.0476,
+      "step": 4149
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462783171521036e-05,
+      "loss": 0.0574,
+      "step": 4150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462653721682848e-05,
+      "loss": 0.0273,
+      "step": 4151
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.46252427184466e-05,
+      "loss": 0.0383,
+      "step": 4152
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462394822006473e-05,
+      "loss": 0.0511,
+      "step": 4153
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462265372168285e-05,
+      "loss": 0.0553,
+      "step": 4154
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.462135922330098e-05,
+      "loss": 0.0264,
+      "step": 4155
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.46200647249191e-05,
+      "loss": 0.0347,
+      "step": 4156
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461877022653722e-05,
+      "loss": 0.0297,
+      "step": 4157
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461747572815535e-05,
+      "loss": 0.0624,
+      "step": 4158
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461618122977347e-05,
+      "loss": 0.0215,
+      "step": 4159
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.46148867313916e-05,
+      "loss": 0.0296,
+      "step": 4160
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461359223300971e-05,
+      "loss": 0.0598,
+      "step": 4161
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461229773462784e-05,
+      "loss": 0.0342,
+      "step": 4162
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.461100323624595e-05,
+      "loss": 0.0266,
+      "step": 4163
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460970873786408e-05,
+      "loss": 0.0494,
+      "step": 4164
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.46084142394822e-05,
+      "loss": 0.0246,
+      "step": 4165
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460711974110032e-05,
+      "loss": 0.0283,
+      "step": 4166
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460582524271845e-05,
+      "loss": 0.0312,
+      "step": 4167
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460453074433657e-05,
+      "loss": 0.0225,
+      "step": 4168
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.46032362459547e-05,
+      "loss": 0.0192,
+      "step": 4169
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460194174757283e-05,
+      "loss": 0.0421,
+      "step": 4170
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.460064724919094e-05,
+      "loss": 0.0241,
+      "step": 4171
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459935275080907e-05,
+      "loss": 0.0309,
+      "step": 4172
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459805825242719e-05,
+      "loss": 0.0221,
+      "step": 4173
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459676375404532e-05,
+      "loss": 0.0241,
+      "step": 4174
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459546925566343e-05,
+      "loss": 0.025,
+      "step": 4175
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459417475728156e-05,
+      "loss": 0.0253,
+      "step": 4176
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459288025889968e-05,
+      "loss": 0.019,
+      "step": 4177
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459158576051781e-05,
+      "loss": 0.03,
+      "step": 4178
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.459029126213592e-05,
+      "loss": 0.0201,
+      "step": 4179
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458899676375405e-05,
+      "loss": 0.0296,
+      "step": 4180
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458770226537216e-05,
+      "loss": 0.0333,
+      "step": 4181
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458640776699029e-05,
+      "loss": 0.0285,
+      "step": 4182
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458511326860842e-05,
+      "loss": 0.0321,
+      "step": 4183
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458381877022654e-05,
+      "loss": 0.0253,
+      "step": 4184
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458252427184466e-05,
+      "loss": 0.035,
+      "step": 4185
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.458122977346278e-05,
+      "loss": 0.0241,
+      "step": 4186
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457993527508091e-05,
+      "loss": 0.0365,
+      "step": 4187
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457864077669904e-05,
+      "loss": 0.0173,
+      "step": 4188
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457734627831716e-05,
+      "loss": 0.0344,
+      "step": 4189
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457605177993528e-05,
+      "loss": 0.0238,
+      "step": 4190
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.45747572815534e-05,
+      "loss": 0.0228,
+      "step": 4191
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457346278317153e-05,
+      "loss": 0.0603,
+      "step": 4192
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457216828478966e-05,
+      "loss": 0.0256,
+      "step": 4193
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.457087378640778e-05,
+      "loss": 0.1009,
+      "step": 4194
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.45695792880259e-05,
+      "loss": 0.0318,
+      "step": 4195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456828478964401e-05,
+      "loss": 0.0294,
+      "step": 4196
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456699029126214e-05,
+      "loss": 0.0186,
+      "step": 4197
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456569579288026e-05,
+      "loss": 0.0253,
+      "step": 4198
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456440129449839e-05,
+      "loss": 0.021,
+      "step": 4199
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.45631067961165e-05,
+      "loss": 0.0308,
+      "step": 4200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456181229773463e-05,
+      "loss": 0.0331,
+      "step": 4201
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.456051779935275e-05,
+      "loss": 0.055,
+      "step": 4202
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455922330097088e-05,
+      "loss": 0.0341,
+      "step": 4203
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455792880258901e-05,
+      "loss": 0.0274,
+      "step": 4204
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455663430420712e-05,
+      "loss": 0.0413,
+      "step": 4205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455533980582525e-05,
+      "loss": 0.0261,
+      "step": 4206
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455404530744337e-05,
+      "loss": 0.0569,
+      "step": 4207
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.45527508090615e-05,
+      "loss": 0.0184,
+      "step": 4208
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455145631067961e-05,
+      "loss": 0.026,
+      "step": 4209
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.455016181229774e-05,
+      "loss": 0.0142,
+      "step": 4210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454886731391587e-05,
+      "loss": 0.0428,
+      "step": 4211
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454757281553398e-05,
+      "loss": 0.0337,
+      "step": 4212
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45462783171521e-05,
+      "loss": 0.0198,
+      "step": 4213
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454498381877023e-05,
+      "loss": 0.0188,
+      "step": 4214
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454368932038835e-05,
+      "loss": 0.0561,
+      "step": 4215
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.454239482200647e-05,
+      "loss": 0.0223,
+      "step": 4216
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.45411003236246e-05,
+      "loss": 0.039,
+      "step": 4217
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453980582524273e-05,
+      "loss": 0.0464,
+      "step": 4218
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453851132686085e-05,
+      "loss": 0.032,
+      "step": 4219
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453721682847897e-05,
+      "loss": 0.0172,
+      "step": 4220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453592233009709e-05,
+      "loss": 0.038,
+      "step": 4221
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453462783171522e-05,
+      "loss": 0.0193,
+      "step": 4222
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453333333333335e-05,
+      "loss": 0.0425,
+      "step": 4223
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453203883495146e-05,
+      "loss": 0.034,
+      "step": 4224
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.453074433656958e-05,
+      "loss": 0.0406,
+      "step": 4225
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452944983818771e-05,
+      "loss": 0.0225,
+      "step": 4226
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452815533980584e-05,
+      "loss": 0.0274,
+      "step": 4227
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452686084142395e-05,
+      "loss": 0.0282,
+      "step": 4228
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452556634304208e-05,
+      "loss": 0.032,
+      "step": 4229
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452427184466019e-05,
+      "loss": 0.032,
+      "step": 4230
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452297734627832e-05,
+      "loss": 0.0526,
+      "step": 4231
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452168284789644e-05,
+      "loss": 0.0182,
+      "step": 4232
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.452038834951457e-05,
+      "loss": 0.028,
+      "step": 4233
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451909385113268e-05,
+      "loss": 0.0445,
+      "step": 4234
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451779935275081e-05,
+      "loss": 0.0288,
+      "step": 4235
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451650485436894e-05,
+      "loss": 0.017,
+      "step": 4236
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451521035598706e-05,
+      "loss": 0.0546,
+      "step": 4237
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451391585760519e-05,
+      "loss": 0.0329,
+      "step": 4238
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.45126213592233e-05,
+      "loss": 0.0292,
+      "step": 4239
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451132686084143e-05,
+      "loss": 0.0241,
+      "step": 4240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.451003236245956e-05,
+      "loss": 0.0598,
+      "step": 4241
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450873786407768e-05,
+      "loss": 0.057,
+      "step": 4242
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450744336569581e-05,
+      "loss": 0.0319,
+      "step": 4243
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450614886731391e-05,
+      "loss": 0.025,
+      "step": 4244
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450485436893204e-05,
+      "loss": 0.0175,
+      "step": 4245
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450355987055016e-05,
+      "loss": 0.0266,
+      "step": 4246
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450226537216829e-05,
+      "loss": 0.0377,
+      "step": 4247
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.450097087378642e-05,
+      "loss": 0.0563,
+      "step": 4248
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449967637540453e-05,
+      "loss": 0.0439,
+      "step": 4249
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449838187702265e-05,
+      "loss": 0.0334,
+      "step": 4250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449708737864078e-05,
+      "loss": 0.0331,
+      "step": 4251
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449579288025891e-05,
+      "loss": 0.0567,
+      "step": 4252
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449449838187703e-05,
+      "loss": 0.048,
+      "step": 4253
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449320388349515e-05,
+      "loss": 0.0398,
+      "step": 4254
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.449190938511327e-05,
+      "loss": 0.0199,
+      "step": 4255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.44906148867314e-05,
+      "loss": 0.0397,
+      "step": 4256
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448932038834953e-05,
+      "loss": 0.0294,
+      "step": 4257
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448802588996764e-05,
+      "loss": 0.0159,
+      "step": 4258
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448673139158577e-05,
+      "loss": 0.0265,
+      "step": 4259
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448543689320388e-05,
+      "loss": 0.0283,
+      "step": 4260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.4484142394822e-05,
+      "loss": 0.0426,
+      "step": 4261
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448284789644013e-05,
+      "loss": 0.036,
+      "step": 4262
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448155339805826e-05,
+      "loss": 0.0432,
+      "step": 4263
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.448025889967637e-05,
+      "loss": 0.0286,
+      "step": 4264
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.44789644012945e-05,
+      "loss": 0.028,
+      "step": 4265
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447766990291263e-05,
+      "loss": 0.0253,
+      "step": 4266
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447637540453075e-05,
+      "loss": 0.0505,
+      "step": 4267
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447508090614887e-05,
+      "loss": 0.0467,
+      "step": 4268
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447378640776699e-05,
+      "loss": 0.0238,
+      "step": 4269
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447249190938512e-05,
+      "loss": 0.0399,
+      "step": 4270
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.447119741100325e-05,
+      "loss": 0.059,
+      "step": 4271
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446990291262137e-05,
+      "loss": 0.0216,
+      "step": 4272
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446860841423948e-05,
+      "loss": 0.0876,
+      "step": 4273
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446731391585761e-05,
+      "loss": 0.0596,
+      "step": 4274
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446601941747574e-05,
+      "loss": 0.064,
+      "step": 4275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446472491909386e-05,
+      "loss": 0.041,
+      "step": 4276
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446343042071198e-05,
+      "loss": 0.0225,
+      "step": 4277
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446213592233009e-05,
+      "loss": 0.0436,
+      "step": 4278
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.446084142394822e-05,
+      "loss": 0.0586,
+      "step": 4279
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445954692556634e-05,
+      "loss": 0.0454,
+      "step": 4280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445825242718447e-05,
+      "loss": 0.0493,
+      "step": 4281
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.44569579288026e-05,
+      "loss": 0.0705,
+      "step": 4282
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445566343042071e-05,
+      "loss": 0.0277,
+      "step": 4283
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445436893203884e-05,
+      "loss": 0.0474,
+      "step": 4284
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445307443365696e-05,
+      "loss": 0.0321,
+      "step": 4285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445177993527509e-05,
+      "loss": 0.0302,
+      "step": 4286
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.445048543689322e-05,
+      "loss": 0.0283,
+      "step": 4287
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444919093851133e-05,
+      "loss": 0.0401,
+      "step": 4288
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444789644012946e-05,
+      "loss": 0.0163,
+      "step": 4289
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444660194174758e-05,
+      "loss": 0.0291,
+      "step": 4290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444530744336571e-05,
+      "loss": 0.0555,
+      "step": 4291
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444401294498382e-05,
+      "loss": 0.0571,
+      "step": 4292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444271844660194e-05,
+      "loss": 0.0348,
+      "step": 4293
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444142394822006e-05,
+      "loss": 0.0442,
+      "step": 4294
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.444012944983819e-05,
+      "loss": 0.0644,
+      "step": 4295
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443883495145632e-05,
+      "loss": 0.0424,
+      "step": 4296
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443754045307444e-05,
+      "loss": 0.0313,
+      "step": 4297
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443624595469255e-05,
+      "loss": 0.0376,
+      "step": 4298
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443495145631068e-05,
+      "loss": 0.0644,
+      "step": 4299
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443365695792881e-05,
+      "loss": 0.0336,
+      "step": 4300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443236245954693e-05,
+      "loss": 0.031,
+      "step": 4301
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.443106796116505e-05,
+      "loss": 0.0381,
+      "step": 4302
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.442977346278317e-05,
+      "loss": 0.0197,
+      "step": 4303
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.44284789644013e-05,
+      "loss": 0.0437,
+      "step": 4304
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.442718446601943e-05,
+      "loss": 0.0269,
+      "step": 4305
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.442588996763755e-05,
+      "loss": 0.0464,
+      "step": 4306
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.442459546925567e-05,
+      "loss": 0.0326,
+      "step": 4307
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.44233009708738e-05,
+      "loss": 0.0519,
+      "step": 4308
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.44220064724919e-05,
+      "loss": 0.0165,
+      "step": 4309
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.442071197411003e-05,
+      "loss": 0.0333,
+      "step": 4310
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441941747572816e-05,
+      "loss": 0.0322,
+      "step": 4311
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441812297734629e-05,
+      "loss": 0.0345,
+      "step": 4312
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.44168284789644e-05,
+      "loss": 0.0197,
+      "step": 4313
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441553398058253e-05,
+      "loss": 0.0289,
+      "step": 4314
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441423948220065e-05,
+      "loss": 0.0302,
+      "step": 4315
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441294498381878e-05,
+      "loss": 0.0298,
+      "step": 4316
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441165048543689e-05,
+      "loss": 0.0501,
+      "step": 4317
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.441035598705502e-05,
+      "loss": 0.0348,
+      "step": 4318
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440906148867315e-05,
+      "loss": 0.0488,
+      "step": 4319
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440776699029127e-05,
+      "loss": 0.0401,
+      "step": 4320
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.44064724919094e-05,
+      "loss": 0.0409,
+      "step": 4321
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440517799352751e-05,
+      "loss": 0.0347,
+      "step": 4322
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440388349514564e-05,
+      "loss": 0.0317,
+      "step": 4323
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440258899676376e-05,
+      "loss": 0.0472,
+      "step": 4324
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.440129449838189e-05,
+      "loss": 0.0471,
+      "step": 4325
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.44e-05,
+      "loss": 0.0463,
+      "step": 4326
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439870550161812e-05,
+      "loss": 0.0201,
+      "step": 4327
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439741100323624e-05,
+      "loss": 0.0373,
+      "step": 4328
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439611650485437e-05,
+      "loss": 0.0734,
+      "step": 4329
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.43948220064725e-05,
+      "loss": 0.0312,
+      "step": 4330
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439352750809062e-05,
+      "loss": 0.0313,
+      "step": 4331
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439223300970874e-05,
+      "loss": 0.0319,
+      "step": 4332
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.439093851132686e-05,
+      "loss": 0.0165,
+      "step": 4333
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438964401294499e-05,
+      "loss": 0.0437,
+      "step": 4334
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438834951456312e-05,
+      "loss": 0.0473,
+      "step": 4335
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438705501618123e-05,
+      "loss": 0.052,
+      "step": 4336
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438576051779936e-05,
+      "loss": 0.0277,
+      "step": 4337
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438446601941748e-05,
+      "loss": 0.0327,
+      "step": 4338
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438317152103561e-05,
+      "loss": 0.0353,
+      "step": 4339
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438187702265374e-05,
+      "loss": 0.0143,
+      "step": 4340
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.438058252427185e-05,
+      "loss": 0.0448,
+      "step": 4341
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437928802588996e-05,
+      "loss": 0.0446,
+      "step": 4342
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437799352750809e-05,
+      "loss": 0.0264,
+      "step": 4343
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437669902912621e-05,
+      "loss": 0.0304,
+      "step": 4344
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437540453074434e-05,
+      "loss": 0.0168,
+      "step": 4345
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437411003236247e-05,
+      "loss": 0.0693,
+      "step": 4346
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437281553398058e-05,
+      "loss": 0.0472,
+      "step": 4347
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437152103559871e-05,
+      "loss": 0.0377,
+      "step": 4348
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.437022653721683e-05,
+      "loss": 0.0227,
+      "step": 4349
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.436893203883496e-05,
+      "loss": 0.0249,
+      "step": 4350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.436763754045307e-05,
+      "loss": 0.0825,
+      "step": 4351
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.43663430420712e-05,
+      "loss": 0.0452,
+      "step": 4352
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.436504854368933e-05,
+      "loss": 0.0247,
+      "step": 4353
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.436375404530745e-05,
+      "loss": 0.0445,
+      "step": 4354
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.436245954692558e-05,
+      "loss": 0.0461,
+      "step": 4355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.43611650485437e-05,
+      "loss": 0.0858,
+      "step": 4356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435987055016182e-05,
+      "loss": 0.0265,
+      "step": 4357
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435857605177993e-05,
+      "loss": 0.0341,
+      "step": 4358
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435728155339806e-05,
+      "loss": 0.0506,
+      "step": 4359
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435598705501619e-05,
+      "loss": 0.0653,
+      "step": 4360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.43546925566343e-05,
+      "loss": 0.0665,
+      "step": 4361
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435339805825243e-05,
+      "loss": 0.0465,
+      "step": 4362
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435210355987055e-05,
+      "loss": 0.0513,
+      "step": 4363
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.435080906148868e-05,
+      "loss": 0.0517,
+      "step": 4364
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.43495145631068e-05,
+      "loss": 0.0173,
+      "step": 4365
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434822006472492e-05,
+      "loss": 0.0288,
+      "step": 4366
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434692556634305e-05,
+      "loss": 0.0444,
+      "step": 4367
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434563106796117e-05,
+      "loss": 0.028,
+      "step": 4368
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.43443365695793e-05,
+      "loss": 0.0376,
+      "step": 4369
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434304207119742e-05,
+      "loss": 0.0335,
+      "step": 4370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434174757281554e-05,
+      "loss": 0.0309,
+      "step": 4371
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.434045307443366e-05,
+      "loss": 0.0308,
+      "step": 4372
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433915857605179e-05,
+      "loss": 0.0406,
+      "step": 4373
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433786407766992e-05,
+      "loss": 0.0702,
+      "step": 4374
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433656957928803e-05,
+      "loss": 0.0292,
+      "step": 4375
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433527508090614e-05,
+      "loss": 0.0388,
+      "step": 4376
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433398058252427e-05,
+      "loss": 0.0439,
+      "step": 4377
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.43326860841424e-05,
+      "loss": 0.0359,
+      "step": 4378
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433139158576052e-05,
+      "loss": 0.0573,
+      "step": 4379
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.433009708737865e-05,
+      "loss": 0.0399,
+      "step": 4380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432880258899676e-05,
+      "loss": 0.0187,
+      "step": 4381
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432750809061489e-05,
+      "loss": 0.037,
+      "step": 4382
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432621359223302e-05,
+      "loss": 0.0481,
+      "step": 4383
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432491909385114e-05,
+      "loss": 0.0401,
+      "step": 4384
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432362459546926e-05,
+      "loss": 0.0277,
+      "step": 4385
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432233009708738e-05,
+      "loss": 0.039,
+      "step": 4386
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.432103559870551e-05,
+      "loss": 0.0342,
+      "step": 4387
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431974110032364e-05,
+      "loss": 0.0414,
+      "step": 4388
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431844660194176e-05,
+      "loss": 0.0292,
+      "step": 4389
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431715210355988e-05,
+      "loss": 0.0335,
+      "step": 4390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431585760517799e-05,
+      "loss": 0.0283,
+      "step": 4391
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431456310679611e-05,
+      "loss": 0.0245,
+      "step": 4392
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431326860841424e-05,
+      "loss": 0.0408,
+      "step": 4393
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431197411003237e-05,
+      "loss": 0.0316,
+      "step": 4394
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.431067961165048e-05,
+      "loss": 0.0217,
+      "step": 4395
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430938511326861e-05,
+      "loss": 0.0306,
+      "step": 4396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430809061488673e-05,
+      "loss": 0.0331,
+      "step": 4397
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430679611650486e-05,
+      "loss": 0.036,
+      "step": 4398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430550161812299e-05,
+      "loss": 0.0252,
+      "step": 4399
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.43042071197411e-05,
+      "loss": 0.0264,
+      "step": 4400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430291262135923e-05,
+      "loss": 0.0602,
+      "step": 4401
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430161812297735e-05,
+      "loss": 0.0484,
+      "step": 4402
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.430032362459548e-05,
+      "loss": 0.0503,
+      "step": 4403
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429902912621361e-05,
+      "loss": 0.0487,
+      "step": 4404
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429773462783172e-05,
+      "loss": 0.0213,
+      "step": 4405
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429644012944985e-05,
+      "loss": 0.0431,
+      "step": 4406
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429514563106796e-05,
+      "loss": 0.0545,
+      "step": 4407
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429385113268609e-05,
+      "loss": 0.0612,
+      "step": 4408
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429255663430421e-05,
+      "loss": 0.0414,
+      "step": 4409
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.429126213592233e-05,
+      "loss": 0.0284,
+      "step": 4410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428996763754045e-05,
+      "loss": 0.0353,
+      "step": 4411
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428867313915858e-05,
+      "loss": 0.0163,
+      "step": 4412
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.42873786407767e-05,
+      "loss": 0.0345,
+      "step": 4413
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428608414239483e-05,
+      "loss": 0.026,
+      "step": 4414
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428478964401295e-05,
+      "loss": 0.0203,
+      "step": 4415
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428349514563107e-05,
+      "loss": 0.067,
+      "step": 4416
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.42822006472492e-05,
+      "loss": 0.0258,
+      "step": 4417
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.428090614886732e-05,
+      "loss": 0.0375,
+      "step": 4418
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427961165048544e-05,
+      "loss": 0.0207,
+      "step": 4419
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427831715210356e-05,
+      "loss": 0.088,
+      "step": 4420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427702265372169e-05,
+      "loss": 0.0291,
+      "step": 4421
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427572815533982e-05,
+      "loss": 0.0571,
+      "step": 4422
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427443365695793e-05,
+      "loss": 0.0357,
+      "step": 4423
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427313915857606e-05,
+      "loss": 0.0172,
+      "step": 4424
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.427184466019417e-05,
+      "loss": 0.0426,
+      "step": 4425
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.42705501618123e-05,
+      "loss": 0.054,
+      "step": 4426
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426925566343042e-05,
+      "loss": 0.0469,
+      "step": 4427
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426796116504855e-05,
+      "loss": 0.0313,
+      "step": 4428
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426666666666666e-05,
+      "loss": 0.0381,
+      "step": 4429
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426537216828479e-05,
+      "loss": 0.0403,
+      "step": 4430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426407766990292e-05,
+      "loss": 0.0393,
+      "step": 4431
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426278317152104e-05,
+      "loss": 0.032,
+      "step": 4432
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426148867313917e-05,
+      "loss": 0.0388,
+      "step": 4433
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.426019417475728e-05,
+      "loss": 0.0437,
+      "step": 4434
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425889967637541e-05,
+      "loss": 0.0611,
+      "step": 4435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425760517799354e-05,
+      "loss": 0.0379,
+      "step": 4436
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425631067961166e-05,
+      "loss": 0.0347,
+      "step": 4437
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425501618122979e-05,
+      "loss": 0.0423,
+      "step": 4438
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.42537216828479e-05,
+      "loss": 0.0471,
+      "step": 4439
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425242718446601e-05,
+      "loss": 0.0782,
+      "step": 4440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.425113268608414e-05,
+      "loss": 0.0312,
+      "step": 4441
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424983818770227e-05,
+      "loss": 0.0419,
+      "step": 4442
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.42485436893204e-05,
+      "loss": 0.0242,
+      "step": 4443
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424724919093851e-05,
+      "loss": 0.0603,
+      "step": 4444
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424595469255663e-05,
+      "loss": 0.0259,
+      "step": 4445
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424466019417476e-05,
+      "loss": 0.0454,
+      "step": 4446
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424336569579289e-05,
+      "loss": 0.0321,
+      "step": 4447
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424207119741101e-05,
+      "loss": 0.0335,
+      "step": 4448
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.424077669902913e-05,
+      "loss": 0.0334,
+      "step": 4449
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423948220064725e-05,
+      "loss": 0.0773,
+      "step": 4450
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423818770226538e-05,
+      "loss": 0.0311,
+      "step": 4451
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423689320388351e-05,
+      "loss": 0.0487,
+      "step": 4452
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423559870550162e-05,
+      "loss": 0.0255,
+      "step": 4453
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423430420711975e-05,
+      "loss": 0.0297,
+      "step": 4454
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423300970873787e-05,
+      "loss": 0.0312,
+      "step": 4455
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423171521035599e-05,
+      "loss": 0.0414,
+      "step": 4456
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.423042071197411e-05,
+      "loss": 0.0459,
+      "step": 4457
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422912621359224e-05,
+      "loss": 0.038,
+      "step": 4458
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422783171521035e-05,
+      "loss": 0.0361,
+      "step": 4459
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422653721682848e-05,
+      "loss": 0.1096,
+      "step": 4460
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.42252427184466e-05,
+      "loss": 0.0325,
+      "step": 4461
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422394822006473e-05,
+      "loss": 0.025,
+      "step": 4462
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422265372168286e-05,
+      "loss": 0.0668,
+      "step": 4463
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.422135922330097e-05,
+      "loss": 0.0214,
+      "step": 4464
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.42200647249191e-05,
+      "loss": 0.0337,
+      "step": 4465
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421877022653722e-05,
+      "loss": 0.0359,
+      "step": 4466
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421747572815535e-05,
+      "loss": 0.0268,
+      "step": 4467
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421618122977346e-05,
+      "loss": 0.048,
+      "step": 4468
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421488673139159e-05,
+      "loss": 0.1075,
+      "step": 4469
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421359223300972e-05,
+      "loss": 0.0367,
+      "step": 4470
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421229773462784e-05,
+      "loss": 0.0431,
+      "step": 4471
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.421100323624596e-05,
+      "loss": 0.0376,
+      "step": 4472
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420970873786408e-05,
+      "loss": 0.0195,
+      "step": 4473
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.42084142394822e-05,
+      "loss": 0.0451,
+      "step": 4474
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420711974110032e-05,
+      "loss": 0.039,
+      "step": 4475
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420582524271845e-05,
+      "loss": 0.0432,
+      "step": 4476
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420453074433658e-05,
+      "loss": 0.057,
+      "step": 4477
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420323624595469e-05,
+      "loss": 0.0296,
+      "step": 4478
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420194174757282e-05,
+      "loss": 0.0519,
+      "step": 4479
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.420064724919094e-05,
+      "loss": 0.031,
+      "step": 4480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419935275080907e-05,
+      "loss": 0.0682,
+      "step": 4481
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.41980582524272e-05,
+      "loss": 0.0415,
+      "step": 4482
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419676375404531e-05,
+      "loss": 0.0346,
+      "step": 4483
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419546925566344e-05,
+      "loss": 0.044,
+      "step": 4484
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419417475728156e-05,
+      "loss": 0.0416,
+      "step": 4485
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419288025889969e-05,
+      "loss": 0.0237,
+      "step": 4486
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419158576051782e-05,
+      "loss": 0.0182,
+      "step": 4487
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.419029126213593e-05,
+      "loss": 0.0277,
+      "step": 4488
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418899676375404e-05,
+      "loss": 0.0494,
+      "step": 4489
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418770226537217e-05,
+      "loss": 0.0747,
+      "step": 4490
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.41864077669903e-05,
+      "loss": 0.0558,
+      "step": 4491
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418511326860842e-05,
+      "loss": 0.0426,
+      "step": 4492
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418381877022653e-05,
+      "loss": 0.0293,
+      "step": 4493
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418252427184466e-05,
+      "loss": 0.0253,
+      "step": 4494
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.418122977346279e-05,
+      "loss": 0.0206,
+      "step": 4495
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.417993527508091e-05,
+      "loss": 0.0287,
+      "step": 4496
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.417864077669904e-05,
+      "loss": 0.047,
+      "step": 4497
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.417734627831715e-05,
+      "loss": 0.0327,
+      "step": 4498
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.417605177993528e-05,
+      "loss": 0.0394,
+      "step": 4499
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.417475728155341e-05,
+      "loss": 0.0401,
+      "step": 4500
+    },
+    {
+      "epoch": 2.91,
+      "eval_loss": 0.07027395814657211,
+      "eval_mean_accuracy": 0.7252884404155394,
+      "eval_mean_iou": 0.6522348662509885,
+      "eval_overall_accuracy": 0.9774481277465821,
+      "eval_per_category_accuracy": [
+        0.9929064781398524,
+        0.6886110933311822,
+        0.6187015599247686,
+        0.6009346302663543
+      ],
+      "eval_per_category_iou": [
+        0.9779869976895426,
+        0.5833176118613678,
+        0.5477439797211661,
+        0.4998908757318773
+      ],
+      "eval_runtime": 132.4286,
+      "eval_samples_per_second": 15.102,
+      "eval_steps_per_second": 1.888,
+      "step": 4500
+    }
+  ],
+  "max_steps": 77250,
+  "num_train_epochs": 50,
+  "total_flos": 8.020387284186563e+18,
+  "trial_name": null,
+  "trial_params": null
+}